Sea anemone Bartholomea annulata venom inhibits voltage-gated Na+ channels and activates GABAA receptors from mammals

Toxin production in nematocysts by Cnidaria phylum represents an important source of bioactive compounds. Using electrophysiology and, heterologous expression of mammalian ion channels in the Xenopus oocyte membrane, we identified two main effects produced by the sea anemone Bartholomea annulata venom. Nematocysts isolation and controlled discharge of their content, revealed that venom had potent effects on both voltage-dependent Na+ (Nav) channels and GABA type A channel receptors (GABAAR), two essential proteins in central nervous system signaling. Unlike many others sea anemone toxins, which slow the inactivation rate of Nav channels, B. annulata venom potently inhibited the neuronal action potential and the Na+ currents generated by distinct Nav channels opening, including human TTX-sensitive (hNav1.6) and TTX-insensitive Nav channels (hNav1.5). A second effect of B. annulata venom was an agonistic action on GABAAR that activated distinct receptors conformed by either α1β2γ2, α3β2γ1 or, ρ1 homomeric receptors. Since GABA was detected in venom samples by ELISA assay at low nanomolar range, it was excluded that GABA from nematocysts directly activated the GABAARs. This revealed that substances in B. annulata nematocysts generated at least two potent and novel effects on mammalian ion channels that are crucial for nervous system signaling.


Results
Bioactivity. To test bioactivity of B. annulata ( Fig. 1) venom, either crude sea anemone extract (AE) or crude extract isolated from nematocysts (NE; see below) was administered to Ocypode quadrata crabs (n = 4 for each case) injecting 15-20 µg of the extract per g/crab in 50 µl PBS. In all cases, strong alterations, including tremor of legs, chelas, and eyes, uncontrolled movements, variable periods of freezing that increased until causing total paralysis, and finally death. Crabs (n = 4) injected with vehicle did not present any response. All these alterations were compatible with effects on the neuromuscular system and confirmed previous observations 29, 33 .
To conduct a broad screening of molecules involved in nerve transmission, we systematically analyzed the venom effect on membrane proteins from the rat brain that were heterologously expressed in Xenopus oocytes through mRNA injection 15,33,34 . It is well known that the oocyte has great experimental advantages for molecular target screening, including its capability to heterologously express functional proteins in its membrane that can be analyzed by electrophysiology. An example of the variety of electrical signals susceptible to analysis is shown it is also commonly known as "corkscrew anemone", due to spiral bands on its tentacles. Regularly is found inside crevices of rocks and coral rubble and has mutualist symbiosis with the cleaner shrimp Ancylomenes pedersoni (d) and the pistol shrimp Alpheus armatus.  Supplementary Fig. S1. Using mRNA-injected oocytes, AE samples from B. annulata were tested. Figure 2 illustrates these experiments, first AE (0.1-1 mg ml −1 ) applied to control oocytes (H 2 O-injected; Fig. 2a) did not elicit any electrical response (n = 6, 3 frogs). However, AE application in oocytes expressing brain proteins generated several electrical responses (Fig. 2b-e), particularly three effects on three different responses were observed in all the oocytes tested. Thus, in oocytes held at − 60 mV, AE generated two types of currents that can be distinguished by their time-course (Fig. 2b) and their dependency on an intracellular Ca 2+ concentration ([Ca 2+ ] i ) increase (Fig. 2c). These current responses are indicated as I s (smooth) and I osc (oscillatory) in Fig. 2b. The first response, I s , was an inward current associated with an increase in membrane conductance. The second response, I osc , began to increase a few seconds later and corresponded to an inward oscillatory current associated with an increase in conductance. Both responses, were co-activated while AE perfusion was maintained and were washed completely by NR superfusion. I osc resembled the typical responses to various neurotransmitters when their receptors are expressed through brain mRNA injection in the oocyte (Fig. S1 35 ). These are currents dependent on [Ca 2+ ] i increase promoted by phospholipase C activation and are due to Ca 2+ -dependent Cl − channels opening 36,37 . To explore if I osc elicited by the extract was activated through the same mechanism, oocytes that generated I osc by AE, were washed, loaded with EGTA 38 , and then tested again with AE (Fig. 2c, orange traces). In all cases, AE elicited only I s responses in EGTA-loaded oocytes, while I osc were eliminated, indicating that I s was not Ca 2+ -dependent, and confirming that I osc was more probably activated through [Ca 2+ ] i increase. I osc was the most robust response, whereas I s showed lower amplitude, reaching values of 93 ± 83 nA (n = 22) (orange bar graph in Fig. 2c). No further analysis was performed on the mechanisms of I osc activation. To explore the basic characteristics of I s , its reversal potential (E rev ) was estimated in EGTA-loaded oocytes that were held at distinct membrane potentials while AE was superfused (Fig. 2d). I s response had an E rev close to − 20 mV in all cases  Supplementary Fig. S1). Increase of the trace timeresolution signal in the insert (blue box) discloses activation of two currents with distinct temporal course, an early and smooth inward current, denoted I s , and a late and oscillatory inward current, denoted I osc . In (c) traces show that I osc current observed in control response (black traces; EGTA non loaded oocytes) was eliminated in EGTA-loaded oocytes, while inward current EGTA-resistant corresponded with I s time-course (orange traces). Total current responses (I m ) obtained in 40 control and EGTA-loaded oocytes (right graph) indicated that I s was of 93 ± 17.7 nA (mean ± S.D.). (d) I s was elicited by 0.5 mg ml −1 AE superfusion in mRNA-injected oocytes that were EGTA-loaded, and holding the membrane potential as indicated in each trace, it is shown that I s reverted close to − 20 mV, similar result was obtained in 2 more oocytes from different frog. (e) Traces illustrate a voltagestep protocol applied in EGTA-loaded oocytes (n = 4) held at − 100 mV in the absence of AE (control, gray traces) and during the application of the venom (0.5 mg ml −1 ; orange traces); the current-voltage relationship was built for the current elicited by AE subtracting the control current as it is shown in the graph. The black line to the left for each set of traces signals zero current. The green circle in control traces signals a fast inward current elicited at − 20 mV; this current, more probably due to opening of Na v channels, was eliminated by AE application.  36,39 . In the responses elicited by the pulse protocol, a fast voltage-dependent inward current that desensitizes within a few milliseconds was regularly activated (green circle in Fig. 2e; see also Supplementary Fig. S1). This current (denoted as I Nav ) corresponds with the opening of voltage-dependent Na + channels commonly expressed in brain mRNA-injected oocytes 40,41 . As depicted in Fig. 2e, I Nav appears to be inhibited during the application of AE. Based on this initial screening, we decided to characterize in detail the AE effects on I s activation and I Nav inhibition.
Inhibition of I Nav . AE was applied to mRNA-injected oocytes expressing I Nav to quantify its effect within a potential range of − 100 mV to + 40 mV and explore its reproducibility in different preparations. The peak I Nav amplitude at − 20 mV was inhibited 96.5 ± 4.97% by 1 mg ml −1 AE (7 oocytes, 3 frogs, Fig. 3a-c). However, the extract at 0.1 mg ml −1 still caused a potent inhibition of 70.8 ± 15.89% (18 oocytes, 7 frogs). AE samples boiled for 45 min fully maintained their inhibitory activity (7 oocytes, 3 frogs), indicating that the inhibitory substance was thermostable (Fig. 3b). I Nav kinetics monitored in the oocyte, within the entire range of potentials tested, results uncertain due to the oocyte size. Due to this, and to explore the effect of AE on endogenous molecules (i.e., I Nav in neurons), the following experiments were performed using cultured rat cortical neurons. Figure 3c shows the I-V relationships for I Nav generated either in mRNA-injected oocytes or in neurons, and for both cases in the absence (black circles) and the presence (green circles) of AE; it was found that inhibition by AE was similar in both preparations. Thus, I Nav inhibition occurred in the entire range of membrane potentials and, importantly, AE had the same effect on neurons; for example, 0.1 mg ml −1 AE inhibited neuronal I Nav by 95.3 ± 3% (5 cells). Figure 3d illustrates a typical example of neuronal I Nav elicited by 20 ms voltage steps within the − 10 to − 40 mV range from a holding potential of − 100 mV, regularly, the maximum current amplitude was obtained at − 20 mV. To study the time-course of the Voltage-dependent inward current at − 20 mV was potently inhibited by AE in concentrations of 0.1 mg ml −1 (18 oocytes, 7 frogs) and 1 mg ml −1 (7 oocytes, 3 frogs), the inhibitory effect was not eliminated in samples "boiled" by 45 min (7 oocytes, 3 frogs). (c) I-V relationship for the inward current response in oocytes (left graph; n = 8) suggested that it corresponds to opening of Na v channels; this was confirmed recording Na v currents in cultured cortical neurons (right I/V graph, black circles, n = 5) that were also potently inhibited by 1 mg ml −1 superfusion (green circles); in both cases currents were inhibited in the whole voltage-range analyzed. (d) Traces illustrate the I Nav activated in neurons at − 30, − 20 and − 10 mV from a holding potential of − 100 mV. Maximal currents were regularly obtained at − 20 mV, to test the effect of extracts in this and subsequent figures, I Nav was elicited periodically, stepping the membrane potential from − 100 to − 20 mV for 20 ms every 0.75-1 s as illustrated in the traces on the right. To show the complete time-course of a particular experiment, traces were superimposed as this is shown in the insert that illustrates 12 consecutive stimuli. (e) In neurons, I Nav current was activated periodically as in (d). After monitoring the control amplitude (black traces), neurons were exposed to 0.1 mg ml −1 AE (green traces) for 10 s. During AE application I Nav was rapidly inhibited, and after this, AE removal allowed current recovery. The graph in (f) shows the effect monitored in 5 neurons (mean ± S.D.) using the protocol in (e), top black bar signals AE application; inward currents (green circles) were eliminated while small outward currents depicted in yellow circles showed no change.  Fig. 3d) in such a way that the different treatments were applied for 5-10 s after a control period, followed by washing with external solution. The insert in 3d illustrates a series of control responses obtained using this protocol that have been superimposed to show I Nav reproducibility, the time scale corresponds to that of a single depolarizing pulse. Typical results testing the effect of different treatments using this protocol were illustrated with superimposed traces as it is shown in Fig. 3e, where the time-course I Nav inhibition by AE was analyzed by depolarizing the neurons from − 100 mV as described, after 10-12 voltage steps in control external solution (black traces) AE was superfused for 5-10 s (green traces) and the effect on I Nav was monitored. Finally, AE was washed and I Nav recovery followed for several seconds (traces in black). As this is shown in the bottom graph, application of 0.1 mg ml −1 of AE produced I Nav inhibition with a τ 50 of 2.88 ± 0.03 s (Fig. 3f, green circles; 5 neurons), while smaller outward currents activated at this potential were not significantly affected (yellow circles; most likely corresponding to voltage-gated K + channels). Recovery of inhibition required, in most cases, a 30-45 s wash. During I Nav inhibition produced by AE, no change in the current inactivation rate was observed. For example, in a group of neurons (n = 6) the control time constant of inactivation was of 0.97 ± 0.48 ms, while after 3 s in the presence of AE (0.1 mg ml −1 ), time in which more than 50% of I Nav inhibition was regularly reached, this parameter was of 1.13 ± 0.7 ms. These values did not show a statistical difference.
All the experiments described above used AE, the homogenate of B. annulata. Although it is known that this method successfully discharges nematocysts, the venom obtained with AE clearly may include substances from other anemone tissues in addition to the nematocysts content per se (e.g. Ref. 20 ) To confirm that effects were elicited by nematocysts content, a procedure for nematocysts isolation was done followed by controlled discharge through mechanical means. Figure 4a shows the sea anemone homogenate visualized using phase contrast microscopy. This sample contained a large number of nematocysts, together with symbionts and the sea anemone tissue itself. Comparatively, using a modified Bloom's method 42 , and settling periods of 24-48 h (Fig. 4b) provided with nematocysts free samples of most of other animal tissue, also significantly reduced the number of symbionts. An amplification of these preparations showed the typical morphology of loaded nematocysts containing twisted tubules in their lumen (arrows, Fig. 4c). The purified nematocysts samples obtained after 24-48 h were then pressed to 6.2 × 10 7 Pascals using a French press. The extract obtained will be referred to as nematocysts extract (NE). The pressed preparations were then visualized under the microscope (Fig. 4d,e) to quantify the number of loaded and unloaded nematocysts after 24-48 h (Fig. 4f). Under the microscope, nematocyst discharge was visualized as an empty capsule associated with a large ejected tubule (asterisks and arrows, respectively in Fig. 4e). Figure 4f is a quantification made by triplicate of undischarged (gray bars) or discharged (green bars) nematocysts, in non-pressed aliquots compared with pressed aliquots from the same anemones (f) Quantification of total (black columns), undischarged (gray columns), and discharged nematocysts (green columns) (per μl), for both non-pressed and pressed preparations in samples obtained by the Bloom's method either after 24 h (light green) or 48 h (dark green) (mean ± S.D.; *p < 0.05 for comparisons between pressed vs. non-pressed). (g) Typical effect of AE (1 mg ml −1 ) on neuronal I Nav response at − 20 mV (n = 6) using the same protocol as in Fig. 3. (h) A group of anemones from the same batch was processed by the Bloom's method and pressed either after 24 or 48 h, and the corresponding extracts were assayed on the neuronal I Nav activated at − 20 mV (light and dark green, respectively). In gray is shown the lack of effect of the non-pressed 48 h sample. Individual data points in (g,h) represent the mean ± S.D. from 4 to 5 neurons. www.nature.com/scientificreports/ samples. Quantification showed that in both 24-h and 48-h groups the efficiency of total nematocysts recovery increased, and that a small proportion of non-pressed nematocysts were already discharged. More importantly, it revealed that pressing led to a significant nematocysts discharge. For instance, from a total of 1190 ± 144 nematocysts per μl after 48 h of incubation, the pressed sample rendered 1096 ± 130 discharged nematocysts per μl; thus, most nematocysts were discharged after being pressed. The time-course and potency of the effect for both AE and NE on neuronal I Nav , using the protocol as in Fig. 3d-f, is shown in Fig. 4g,h. First, 1 mg ml −1 AE inhibited I Nav by 77.9 ± 12.6%, while NE (1 mg ml −1 ) obtained from the same anemones batch, incubated for 24 or 48 h, inhibited I Nav by 56.8 ± 5.2% (light green circles) and 69.4 ± 4.4% (dark green circles), respectively (Fig. 4h). Both NE samples showed lower inhibition amplitude with a slower time-course compared to AE; however, the inhibitory effect was similar. The 48-h non-pressed (gray circles) sample presented low activity on I Nav (peak inhibition of 9.2 ± 4.7%). Altogether, these data strongly suggested that the effect on I Nav was due to a substance or substances contained in the nematocysts of B. annulata.
As a first approach for separation and purification of the toxins responsible for the effects, chromatography was performed with a Sephadex G-50 M column that allows substances separation based on their molecular weight. Chromatographic pattern from AE separated three peaks that were identified as FI, FII and FIII fractions (Fig. 5a). The bioassay of these fractions (Fig. 5b) on neuronal I Nav indicated that 1 mg ml −1 FI did not affect I Nav , while FII was a potent fraction reaching inhibition values of 99.82 ± 0.47%. FIII also showed activity reaching an inhibition of 79.5 ± 25.9%, although its effect exhibited greater dispersion. The inhibitory time-course by FII was like that observed with AE and NE (Fig. 5b).
Since FII and FIII elicited I Nav inhibition, a chemical analysis using MALDI-TOF mass spectrometry was done as this is shown in Fig. 5c. Both fractions showed peptides with molecular masses between 3 and 12 kDa, however, FII was more complex in composition than FIII. Importantly, the molecular masses observed were in the range of those reported for sea anemone neurotoxins (3-7 kDa) 6 . The presence of apparently equivalent components, in FII (e.g., m/z 5914.34 and 12,382.10) and FIII (m/z 5911.56 and 12,368.31) (Fig. 5c), might be an explanation for similar effects in both fractions. The graph shows I Nav inhibition time-course produced by FII (1 mg ml −1 ; 11 neurons), while FI did not produce a significant effect (4 neurons; mean ± S.D.), the column graph resumes the inhibition produced by each fraction (1 mg ml −1 ). (c) MALDI-TOF-MS spectrum obtained from the FII and FIII fractions obtained from AE gel filtration chromatography, the numbers indicated the mass for the main signals detected. In (d) FII was applied in distinct concentrations and a dose-response curve (in pink) was adjusted to data points indicating an EC 50 of 80 ± 16 μg ml −1 . (e) FII was tested on the neuronal action potential generated by depolarizing steps (35 ms, 0.75 nA) every 500 ms; three decreasing FII concentrations are illustrated. Traces in black (10 superimposed traces in each case) corresponded to control action potentials before FII superfusion, whereas red traces were obtained during FII application. 0.5 mg ml −1 potently inhibited the action potential, a concentration 10 times lower still blocked completely the action potential in few seconds, while 0.01 mg ml −1 FII was ineffective.  (Fig. 5d). Also, FII was used to analyze in cultured neurons its effect on action potential generation. For this, constant current pulses (0-65-0.85 nA) were applied to generate action potentials every 1-1.2 s at 30% above threshold amplitude, to avoid failures in impulse generation. Then, different solutions containing 10 μg to 1 mg of FII were applied for 10 s (n = 4, Fig. 5e). Complete action potential inhibition was observed in a dilution as low as 0.05 mg ml −1 after 3-4 s of superfusion, while a robust inhibition effectiveness was observed from 0.1 to 1 mg ml −1 . FII produced a fast action potential blockage, and the duration was not affected during FII application. At all concentrations tested, the recovery of action potential amplitude was complete after washing.
Tetrodotoxin (TTX) is a potent Na v channel blocker 43 , the molecular identity of the TTX binding site on the channel is well known 44 . To know whether NE inhibition effect was produced acting on the same or similar TTX binding site, extract was tested in HEK293 cells expressing either hNa v 1.6 or hNa v 1.5 channels. As it is well known, TTX acts differentially on distinct I Nav subtypes; hNa v 1.6 is a TTX-sensitive channel, whereas hNa v 1.5 is essentially a TTX-insensitive channel 45 . I Nav currents in HEK cells were recorded similarly to those in neurons; thus, current responses were activated by depolarizing the cell periodically to values around the peak current. As is illustrated in Fig. 6a,c, 1 mg ml −1 NE superfusion inhibited the current elicited by hNa v 1.6 opening as strongly as the blockage produced by 50 nM TTX (Fig. 6b,c). Remarkably, 1 mg ml −1 NE also strongly inhibited the current generated by hNa v 1.5 channel activation (Fig. 6d,f). As expected, in these same cells the I Nav current was not blocked by TTX (Fig. 6e,f). Also, comparing the rates of inhibition and recovery of NE and TTX on the hNa v 1.6 current, NE seemed to act faster than TTX, and 155 ± 34 μg ml −1 NE inhibited 50% of I Nav (Fig. 6g). Like that observed with AE, NE (1 mg ml −1 ) either boiled for 15 min or incubated for 24 h at 37 °C, fully retained activity (Fig. 6h). inward current (I s ) associated with an increase in membrane conductance in mRNA-injected oocytes (Fig. 2be). This response was generated in both neurons and oligodendrocytes, although the amplitude response was clearly weaker in the latter cell type, as illustrated in Fig. 7a. In oocytes, the estimated E rev corresponded with the flux of Cl − ; thus, the I-V relationship for the inward response was built for neurons (Fig. 7b). We also found that the E rev for the inward current response in external solution containing 146 mM Cl − was of − 8.8 ± 3.5 mV, while in solution with reduced Cl − concentration (36.5 mM) the E rev shifted to a potential of + 16.2 ± 4.3 mV, and the inward current amplitude became larger; altogether, these results suggest that inward current response was mainly carried by Cl − . The main Cl − channel in neurons is GABA A R. To explore a possible GABA A R involvement in the response elicited by AE, the extract was coapplied in neurons with either 30 μM bicuculline or picrotoxin, an antagonist and a blocker of the GABA A R, respectively. In neurons (Fig. 7c-e; 5-7 neurons), bicuculline inhibited to 19 ± 18.1% the current elicited by 30 μM GABA, whereas picrotoxin blocked around 50% of the GABA response (to 48.2 ± 7.6%). In the same neurons, bicuculline and picrotoxin inhibited the current activated by 0.5 mg ml −1 AE to 57.5 ± 24.4% and 52.2 ± 12.8%, respectively (Fig. 7e). Contrary to what was observed in the GABA-generated current, bicuculline had a more variable effect on the inward current elicited by AE (see Fig. 7c), although picrotoxin presented a similar blocking effect either with GABA or AE stimulation (Fig. 7d). The FI-FIII fractions obtained by chromatography also showed differential potency on neurons, thus 1 mg ml −1 FI was ineffective (7 neurons) to generate the inward response, 1 mg ml −1 FII generated inward responses of 1128 + 763 nA (13 cells), while FIII generated weak responses of 170.3 + 112 nA (7 cells), suggesting that the effect was elicited for a substance mainly contained in fraction FII.
To confirm a direct action on GABA A R by AE and NE, receptors were expressed in oocytes. Expression of the most common neuronal GABA A R, comprised by α1β2γ2 subunits, provides sensitivity to GABA superfusion (Fig. 8a). Oocytes responding to 10 μM GABA were also sensitive to 0.5 mg ml −1 AE superfusion, generating smooth inward currents that opened rapidly after beginning the application and stayed activated until GABA or AE were washed out. To make the application of AE more efficient, the extract was applied through a jet of solution ejected close to the oocyte surface. Extract solution jet application was a reproducible method of stimulation that consumed minimal amount of venom (Fig. 8a). The AE jet application method did not generate any response in native oocytes (Fig. 8b); however, in oocytes expressing the neuronal GABA A R, the delivery of either AE or NE prompted a rapid inward response that, as expected, had an E rev of − 22.8 ± 2.6 mV (Fig. 8c). GABA A Rs composed of other subunits expressed in oocytes were also activated by AE and by NE applied in similar manner; thus, receptors comprised by α3β2γ1 subunits typically expressed in oligodendroglial cells 46 and homomeric ρ1 GAB A R 47 were also operated by the extracts. The α3β2γ1 receptor has an EC 50 for GABA of 80 μM, with a threshold for the response close to 1 μM (Fig. 8d, red traces). The typical neuronal receptor expressed in www.nature.com/scientificreports/ the oocyte membrane has an EC 50 around 100 μM with a threshold also close to 1 μM, while the homomeric GABA A R ρ1 has an EC 50 of 1.2 μM, with a threshold within the nanomolar range (Fig. 8d, green traces). Using the same conditions and micropipette for jet application, responses to AE had an expected amplitude for the respective GABAAR expressed; thus, α3β2γ1-receptor responses were lower in amplitude, while ρ1-receptor responses were consistently larger (Fig. 8e). These results suggested that extracts could contain GABA at a concentration within the micromolar range; thus, AE samples were analyzed using a GABA-specific ELISA test (with nanomolar detection sensitivity). However, AE and NE samples (3 different extracts) that activated robust responses in oocytes expressing either the α1β2γ2-or the α3β2γ1-receptor reported a GABA concentration of 9.97 ± 1.8 nM in the ELISA test, which by itself did not explain the observed amplitude responses. This strongly suggested that B. annulata venom contained a potent GABA A R agonist different to GABA.

Discussion
More than sixty toxins affecting Na v channels have been isolated from sea anemones, most of them cause a delay in Na v channel inactivation 48,49 . The results described here show two novel toxic effects exerted by substances contained in the nematocysts of the sea anemone B. annulata, each one acting on fundamental membrane proteins of the mammalian nervous system: the Na v channel and the GABA A R. The recognition of species of interest, due to specific effects of their toxins and/or by their venom potency, is undoubtedly an advantage to www.nature.com/scientificreports/ direct the search for new bioactive principles. Here we show that B. annulata venom contains substances that act on the nervous system of both vertebrates and invertebrates, as seen by the effects on crabs and on membrane proteins of vertebrates, specifically those from the nervous system of mammals (including human). The novelty of the effects described here, clearly indicates the importance of studying the potential bioactivity of cnidarian families that remain unexplored.
In several experiments of this study, nematocysts of B. annulata were first isolated using a modified Bloom's method 42 , and then subsequently discharged in a controlled manner, applying pressure conditions that did not affect the symbiont structural integrity, as previously reported 50 . This increased the certainty that the effects observed were generated by compounds contained in the nematocysts, since extracts that were not pressed had reduced activity compatible with the basal discharge observed by nematocysts quantification.
Na v and GABA A R channels are crucial in neural functions and effects on them could explain by themselves the venom lethality 30,51,52 . Na v channels are responsible for nerve impulse generation, while GABA A Rs are important effector molecules of GABA, the main inhibitory neurotransmitter in the nervous system. Until now, most neurotoxins isolated from sea anemones (e.g., ShK, BgK, CgNa, and CGTX II) affect K v or Na v channels 13,[53][54][55][56][57][58][59] . The peptide neurotoxins that act on Na v channels are widely studied and have been classified into four types 6 . It is known that the action mechanism of many toxins from anemones involves their binding to open Na v channels, causing a delay in channel inactivation and an increase in action potential duration 57,60,61 . However, different to this mechanism, our results showed that B. annulata venom contains a substance that inhibits the Na v current response at the whole voltage range tested, without affecting both the inactivation rate, and apparently neither the voltage dependency of activation. Also, I Nav inhibition more probably caused a rapid and dose-dependent blockage of the action potential generation in cortical neurons with no effect on its duration, suggesting that the inhibitory effect of B. annulata venom was more compatible with a blockage of the Na v channel pore, although a complete definition of the inhibition mechanisms involved will require of further experiments. The chemical identity of the toxin(s) remains to be elucidated; however, it would correspond to a heat-stable molecule of more than 3000 Da. These features agree with the results of mass spectrometry analyses of FII and FIII that showed the presence of peptides in the expected range. Thermostability of anemone toxins has been reported in several cases and, this characteristic has been associated with the stability provided by the presence of disulfide bridges in its structure (e.g., Refs. 6,51,62 ). It is also evident that its action mechanism was not identical to that of TTX, because B. annulata venom inhibited both the human Na v 1.6 and Na v 1.5 channel current responses, which clearly differ in their sensitivity to TTX 45,63 . It is still possible that B. annulata bioactive substance, described here, shares an overlapping binding site with TTX, and some small change in Nav1.5 is enough to interfere with TTX and not with that of B. annulata.
A different binding site for pore blockage in Na v channels is that of the cone snail toxin GIIIA 64 , more experiments will be necessary to establish its possible participation in the B. annulata venom effect. The functional study of new molecules that affect Na v channels in mammals remains limited. Hence, the discovery of new template molecules that could act in a more specific and differential manner is of general interest for the design and development of novel molecular tools (e.g. Ref. 65 ). Furthermore, Na v channels regulation are important not only for nervous system physiology but also because of their possible role in various diseases including cancer [66][67][68][69] .
A second molecule in the venom, acting as a GABAergic agonist, also represents a new action mechanism for sea anemone toxins, and its rapid action opening the channel in the millisecond range, as well as the complete recovery upon wash, resembles the action of the full agonist muscimol produced by the fungus Amanita muscaria 70 . Several natural substances act on GABA A R and many are drugs that have powerful effects on the nervous system (e.g., picrotoxine, β-carbolines). Also, peptides that inhibit GABA A Rs have been previously described in marine animal venoms in at least two species, Laticauda semifasciata and Conus imperialis 71 . In the case of B. annulata venom, the active substance operated three GABA A R types with different molecular identities (either neuronal or glial heteromeric receptors, as well as the ρ1 homomeric receptor); thus, this suggested that the venom contained concentrations in tens of the micromolar range of GABA. The expression of the GABAergic signaling system has been documented in cnidarians where it has been related to nematocyst discharge regulation, among other functions 72 . However, results of ELISA assays with sensitivity to nanomolar GABA concentration showed that the venom contained GABA in around 10 nM. This indicated that the molecule responsible for GABA A R activation was of a different nature since this concentration was too low to activate either the neuronal or the oligodendroglial receptor. Alternatively, the venom could contain a potent positive modulator that would increase the GABA A R sensitivity to GABA.
Nevertheless, our results also suggested that GABA A Rs were not the only receptors activated by the venom. Thus, testing the venom on cortical neurons revealed that currents produced in some cells, persisted even after its co-application with bicuculline, a potent competitive antagonist of the GABA site. This finding could also be explained by a different competition mechanism between bicuculline with the active substance as compared with GABA. On the other hand, the importance of a molecule in the venom that depresses the activity of the central nervous system seems to be clear, and it is striking that this mechanism apparently is not common among the venoms studied so far, since combining it with a powerful Na v channel blocker molecule would result in a fast-acting bioweapon.
The advancement of robust biochemical and molecular techniques will provide detailed information about the identity of substances contained in the venom of many species that remain unexplored, including cnidarians. This study showed that the common Caribbean B. annulata is a source of novel neurotoxins that potently target essential proteins for the functioning of the mammalian central nervous system.

Conclusion
The number of marine species that synthesize bioactive molecules studied is far from being exhausted, these molecules are widely diverse, and their molecular targets are of great interest in terms of basic mechanisms and for the development of new therapeutic and experimental tools. The use of heterologous expression models for the recognition of target molecules, such as the expression in X. laevis oocytes, represents a useful method to recognize relevant effects of the venoms. Here, oocytes expressing mammalian brain ion channels and membrane receptors revealed effects of the B. annulata venom on two of the most important proteins for the functioning of the nervous system, Na v channels and GABA A Rs.

Materials and methods
Animal ethics declarations. All animal studies complied with the ARRIVE guidelines. All the animals were treated with the utmost care to minimize their suffering and were handled in accordance with the Guide for the Care and Use of Laboratory Animals by the USA National Institutes of Health, as well as the Local Guidelines on the Ethical Use of Animals for Experimentation at the Universidad Nacional Autónoma de México. B. annulata sample collection was carried out through permit no. PPF-066/20 from Dirección General de Ordenación Pesquera y Acuicultura, Quintana Roo, México.
Sea anemone collection and venom extraction. Specimens of Bartholomea annulata were collected from the reef lagoon at Puerto Morelos, Quintana Roo, México (20° 50′ 45″ N 86° 52′ 08″ W; 20° 50′ 55″ N 86° 52′ 52″ W) by SCUBA diving (Fig. 1). The venom extracts were prepared using two different methods, as follows: (1) Sea anemone macerate to obtain AE was carried out as described elsewhere 29 . Briefly, whole sea anemone bodies were homogenized in deionized water to perform the nematocysts discharge. Rupture of nematocysts was monitored by microscopy until most nematocysts were discharged. The extract obtained was centrifuged at 3.2 × 10 3 ×g at 4 °C for 10 min, the supernatant was decanted, and the same procedure was applied one more time. Finally, the supernatant was dialyzed using a pore size limit of 3000 Da, lyophilized and frozen until it was used. (2) Nematocyst's isolation and discharge to obtain NE. To avoid contamination of the extract with other cells or tissues, sea anemones were washed with filtered (0.45 µm) sea water to remove sand and small stones 57 , then they were exposed to six freeze-thaw cycles. Sea anemone bodies were immersed in two volumes of distilled water kept at 4 °C and stirring every 10 min for 2 h (modified from Ref. 42 ). Samples were filtered with a 100-µm-mesh and the filtered material was allowed to settle for either 24 h or 48 h. After the settlement period the sediment was washed three times, first with phosphate buffer solution (PBS, containing in mM: 2.74 NaH 2 PO 4 , 7.2 dibasic Na 2 HPO 4 ) plus 0.01% Triton, and then twice with PBS alone. The precipitate was visualized under a microscope to monitor the number of undischarged nematocysts. Then, discharge of these organelles was carried out applying a pressure of 6.2 × 10 7 Pa using a French press cell disruption homogenizer (SLM-AMINCO, Urbana, IL, USA). It has been reported that symbionts are affected applying a higher pressure of 8.3 × 10 7 Pa 50 . Finally, samples were centrifuged at 10 5 g for 1 h at 4 °C, and the supernatant containing the venom was dialyzed using a pore size limit of 3000 Da, lyophilized, and kept at − 30 °C. Toxicity test. Both extracts were tested in Ocypode quadrata crabs, collected from the Puerto Morelos beach. Aliquots (50 μl) of either extract dissolved in PBS, or PBS alone as a control, were injected through the third walking leg 29,33 . Size exclusion chromatography. The AE was fractionated using Sephadex G-50 M. The column was equilibrated with PBS. Samples dissolved in the same buffer were eluted to a flow rate of 300 μl min −1 , with monitoring at 280 nm, and 5-ml fractions were collected. The fractions were concentrated with reduced pressure ("rotary evaporator"), dialyzed using a pore size limit of 3000 Da, and lyophilized.
Mass spectrometry. Fractions FII and FIII were chemically analyzed using MALDI-TOF mass spectrometry. 5 μl of a saturated solution of sinapinic acid (> 99.0% purity for MALDI-MS, Sigma, St. Louis, MO, USA) were added to 5 μg of lyophilized of each fraction. 1 μl of this solution was deposited onto the MALDI plate and allowed to dry at room temperature. The spectrum was recorded on linear positive mode on a mass spectrometer (Microflex Bruker Daltonics, Bremen, Germany) equipped with nitrogen laser λ = 337 nm and a 20 kV acceleration voltage. mRNA and cDNA. The poly(A)-mRNA was purified from adult Wistar rat brains; for this, mRNA was extracted from the whole brain using the guanidinium/phenol/chloroform method followed by oligo(dT)-cellulose chromatography 73 . The mRNA pool used here was purified from 19 different extractions from the same number of brains, each extraction was analyzed for its quality by electrophoresis and the ratio of absorbance 260/280 nm, as well as for its expression capacity of ion channels and membrane receptors into oocytes from different frogs. It was dissolved to 1 ng nl −1 in water and stored at − 80 °C until its use for expression using Xenopus oocytes.
The cDNA coding sequences for α1, α3, β2, γ1, and γ2 subunits of the GABA A R were obtained as previously reported 46 from neural tissue, while ρ1 subunit coding sequence was donated by Dr. Ataúlfo Martínez-Torres 47 . All subunits were amplified, and each fragment obtained was cloned into pXENEX1 vector at the Ncol, BamHI, and NotI sites, then plasmids were linearized with the Hind III enzyme and used as templates for cRNA in vitro synthesis using the T7 mMESSAGE mMACHINE kit following the standard protocol (Ambion Invitrogen, Two or more days after injection, the membrane currents were recorded using the two-electrode voltageclamp technique 38 . Microelectrodes (1 MΩ) were inserted into the oocyte and transmembrane currents were monitored using an Axon GenClamp 500 B (Molecular Devices, San Jose, CA, USA) amplifier. Signals were digitized and stored using an analog-to-digital convertor (Axon DigiData 1200; Molecular Devices) and specialized software (pClamp v9; Molecular Devices) 75 .
Unless otherwise stated, oocytes were held at − 60 mV and continuously superfused (10 ml min −1 ) with NR. The venom and distinct neurotransmitters (serotonin, 5 HT; γ-aminobutyric acid, GABA; acetylcholine; glutamic acid) were applied regularly by superfusion. To activate voltage-dependent ionic channels, mainly Na v channels, a protocol of voltage steps was applied to build the current-voltage (I-V) relationships. Voltage-step protocols were applied while oocytes were held at − 100 mV and membrane potential was changed from − 80 to + 40 mV in steps of 20 mV for 250 ms, current responses were measured either at the steady-state or at the peak of the inward current, for I s response or I Nav , respectively.
The method of drug application by superfusion regularly requires of a high sample volume (10 ml per min of application) for oocyte recording. Here, a much more efficient delivery system for the extracts was used (Fig. 8). This method consisted of applying the extract sample, either AE or NE (in NR solution), through jets of solution from a micropipette placed near the surface of the oocyte (approximately 100 μm), while maintaining the general flow of the external medium at a minimum level of 0.5 ml min −1 ; then the extract contained in the micropipette was ejected by applying small pulses of pressure through the opposite end. The pressure (regularly 10-20 PSI) was controlled through a manometer as well as by modifying the pulse duration. In this manner small volumes of test solution were applied. Regularly jets of few nanoliters (25-100 nl) were enough to activate current responses in oocytes expressing GABA A Rs.
Electrophysiology and culture of neural cells and HEK293 transfected cells. Primary culture of neurons was obtained from E18 Wistar rat embryos according to previously described procedures 76 . Briefly, neurons derived from cortical lobes were resuspended in B27 neurobasal medium plus 10% Fetal Bovine Serum (FBS) and then seeded onto poly-l-ornithine-coated (30 µg ml −1 ) 24-well plates bearing 12-mm-diameter coverslips at 1 × 10 3 cells per well. The medium was replaced by serum-free B27-supplemented Neurobasal medium 24 h later. The cultures were essentially free of astrocytes and microglia and were maintained at 37 °C and 5% CO 2 . Cultures were used at 7-12 days in vitro (DIV).
Primary cultures of oligodendrocytes derived from optic nerves of 12-day-old Sprague-Dawley rats were obtained as described previously 77,78 . Cells were seeded on 24-well plates bearing 12-mm-diameter coverslips coated with poly-d-lysine (10 mg ml −1 ) at a density of 10 4 cells per well. Cells were maintained at 37 °C and 5% CO 2 in Sato medium 70 and recorded from 1 to 2 DIV.
Whole-cell patch-clamp recordings of either neural or HEK293 membrane ionic currents were performed with an Axon 700B amplifier (Molecular Devices). Cells were constantly perfused with normal external solution (NES, containing in mM: 135 NaCl, 5 KCl, 2 CaCl 2 , 1 MgCl 2 , 10 HEPES, and 5 glucose adjusted at pH 7.3 with NaOH). The pipette was filled with normal internal solution (NIS, containing in mM: 5 NaCl, 130 KCl, 1 CaCl 2 , 1.8 MgCl 2 , 10 HEPES, 10 EGTA, 0.2 Na-GTP, and 2 Mg-ATP at pH 7.3 (KOH). Data were digitized at 5 kHz and low-pass filtered at 0.5 kHz. Currents were recorded at distinct holding membrane potentials depending on the experiment as specified in each case, digitized, and stored for analysis using the analog-to-digital converter Digidata 1400 (Molecular Devices) and pClamp10 software (Molecular Devices). Current-voltage (I-V) relationships were regularly built by changing the membrane potential from − 100 to + 40 mV in 10-mV steps of 250 ms while cells were held at − 100 mV. In some instances, cells were held at a desired membrane potential while a test solution was superfused or applied using the jet method, and the peak currents generated were I/V plotted. Time-courses of the effect on I Nav were followed in cells held at − 100 mV and then depolarizing steps (20 ms) eliciting maximal current (usually to − 20 mV or − 10 mV) were applied periodically every 0.75-1 s. Usually, the control current was monitored for 10 s and then a test solution was superfused for 5-10 s to record the effect on I Nav, this was followed by a washing period with NES.
In one set of experiments the action potential was recorded in neurons under current-clamp, in these cases depolarizing current pulses (0.65-0.85 nA) of 35 ms were applied to generate action potentials every 500 ms, after a recording period control, different dilutions of FII were applied to test their effect on the nerve impulse.