When thinking of toxic compounds, researchers often have in mind the famous citation of Paracelsus (Philippus Theophrastus Aureolus Bombastus von Hohenheim): "The dose makes the poison" or in its more complete version "All things are poison and nothing is without poison. Solely the dose determines that a thing is not a poison." [1]. Although this 16th century concept is now debated, Paracelsus is still recognized as one of the fathers of "Toxicology" as it is envisioned today.

Medical drugs and toxins are good illustrations of this adage. On the one hand, medical drugs are often perceived as non-poisonous because of their potential benefit for humans. However, it is now well known that widely used drugs like Paracetamol℘ or Ibuprofen℘ can be highly toxic at relatively moderate doses [2]. On the other hand, toxins are generally perceived as dangerous for humans because of their possible accumulation in the alimentary chain or their use as biological weapons (e.g., anatoxin-a). However, following Paracelsus, lowering the dose might allow to make them non-toxic. Solving the issue of toxicity would pave the way to turn them into medicine. Indeed, in addition to their use as an invaluable source of ligands for studying structural or functional properties of their molecular targets [3], these molecules are now increasingly interesting to researchers for their use as medicine or cosmetic products [4].

The word 'toxin' was first introduced by Ludwig Brieger as a name for poisons made by infectious agents [5]. These biological poisons allow the organisms to survive difficult environmental situations where the toxins are advantageous for prey capture or defense [6]. Plant toxins (e.g., nicotine) often function as protection against certain animals. In animals, toxins have similar defense potential and are also used to capture prey. Toxicity might, however, be less directly connected with environmental situations as in the case of fish and shellfish that become poisonous after feeding on toxic plants or algae.

Toxins are nowadays usually defined as poisonous substances produced by living organisms including bacteria, microalgae, plants or fungi. We will use here this definition and therefore restrict ourselves to natural substances affecting an animal.

The knowledge of toxins acting on the ligand gated ion channels (LGIC) is dispersed and not homogeneous. Some toxins were isolated and chemically characterized, but poorly studied on the LGIC afterwards. Others found very wide applications and are frequently used in research or even as therapeutic intervention. This in-homogeneity could be due to the fact that the study of toxins acting on LGIC is the intersection of two relatively separated fields: toxinology on one hand and the study of the LGIC on the other hand. Another explanation is that many of the toxins described below were discovered and characterized (and then forgotten?), before the diverse LGIC had been identified. We therefore decided to construct a list of toxins targeting the LGIC that would be as exhaustive as possible.

This compilation of toxins targeting the LGIC should be useful as toxins constitute a relatively unbiased—in term of chemical space covered—source of ligand structures that can be used: (i) as a source of inspiration for drug design, as much as hits identified from high-throughput screening experiments; (ii) in structure/activity relationship studies; (iii) in virtual screening studies. The structure of a representative member of each family of the ligand-gated ion channels has been very recently solved by X-ray crystallography making these studies timely.

Nicotinic receptors are arguably the most well-known LGIC to date. This is probably due to their very early discovery and the large number of toxins blocking them. Historically, the pharmaceutical knowledge of the cholinergic system has emerged with the “discovery” of curares by Spanish explorers in South America during the 16th century. Indeed, curares were used there by local tribes for hunting [1]. It was found during the 19th century that curares block the synaptic transmission at the level of the neuromuscular junction [13] therefore paralyzing the prey. Similar usage of curares have also been reported in Africa [14], and Malaysia [15]. Toxins had an invaluable contribution to the emergence of the notion of LGIC as nAChR where first defined as the “nicotine and curare receptive substance” [16].

Another historically significant toxin targeting the nAChR is α-Bungarotoxin (reviewed in [17]). It was first used to isolate the nAChR [18,19]. The venoms of marine cone snails represent a rich combinatorial-like library of evolutionarily selected, neuropharmacologically active peptides called conotoxins (Figure 4, Table 1) that target a wide variety of receptors and ion-channels [20]. The subtype specific snake α-neurotoxins and cone snail α-conotoxins are still widely used to probe receptor structure and function in native tissues and recombinant systems [21].

Nicotine is a highly toxic alkaloid proposed to serve as an insecticide protecting Tobacco plants [22]. It is the prototypical agonist at nicotinic cholinergic receptors (Figure 5) in comparison to muscarinic receptors [23]. Tobacco extract was used as an insecticide for centuries [24], perhaps as early as 1690 [25]. The effect relies on the presence of nicotine and anabasine (see Section 5). Moreover, nicotine and anabasine were still in use in the early 20th century (see below). Nicotine is also important medically because it is thought to be responsible for tobacco addiction through the stimulation of α4β2 nAChR on dopaminergic neurons of the ventral tegmental area (VTA) [26].

β-Erythroidine is isolated from the coral tree Erythrina crista-galli. It acts as a competitive antagonist of nicotinic receptors (it targets α4β4, α4β2 and α3β2* receptors and has weak affinity for the muscle type and α3β4) [56]. At high concentrations, they are noncompetitive blockers of possibly all nicotinic receptors subtypes [57,58,59]. Erysodine is a structurally related Erythrina Alkaloid acting on α4β2 and α3β2* nAChR. Both compounds are weak binders to α7 nAChR explaining that they are used to discriminate between different nAChR subtypes although they are relatively non-selective.

Methyllycaconitine (MLA), is extracted from Delphinium species [60] and is a potent and highly selective α7 nAChR antagonist [61,62]. MLA is largely used for this property as a pharmacological tool in research [63]. MLA together with additional alkaloids in Delphinium species (nudicauline, 14-deacetylnudicauline, barbinine and deltaine) have also been found to act on nAChRs blocking the neuromuscular junction which may be related to the Delphinium species involvement in cattle poisoning [64,65].

The toxins of the GABA receptors are, as for the nAChR, of different categories (agonists, antagonists and allosteric modulators) as illustrated by the examples described below.

α-thujone is extracted from the wormwood Artemisia absinthium and is found in absinthe [66]. It is a negative allosteric modulator of GABA-A receptors (Figure 6) resulting in convulsant activity [67]. α-thujone also antagonizes 5HT3 receptors [68].

Bicuculline, isolated from Dicentra cucullaria, is a competitive antagonist of GABA-A receptors causing convulsions [66].

Muscimol is an agonist extracted from Amanita muscaria partly responsible for the toxic effect of the mushroom [66].

Picrotoxin is a non-competitive antagonists isolated from Menispermaceae. Binding modes for picrotoxin have been proposed in the ion channel [69,70,71].

Strychnine is found in the seeds of the Strychnine tree (Strychnos nux-vomica). Strychnine causes muscular convulsions and eventually death through asphyxia or sheer exhaustion. It is used as a pesticide, particularly for killing small vertebrates such as birds and rodents. Strychnine participates in the pharmacological differentiation of receptors responsive to glycine. Indeed, some NMDA receptors are also activated by glycine but are strychnine insensitive [72].

The serotonin receptors have very few known toxins: Conotoxin GVIIA (σ-conotoxin; a large 41 amino-acids conotoxin [73]) and d-tubocurarine.

d-Tubocurarine is a mono-quaternary alkaloid obtained from the bark and stems of Chondrodendron tomentosum. d-Tubocurarine blocks nAChRs at the neuromuscular junction but also acts on serotonin receptors [74]. d-Tubocurarine is the archetypal curare. As neuromuscular blockers, curares can be used as skeletal muscle relaxants and were indeed introduced in anesthesia in 1942 [75]. Curares are active only by injection. They are harmless if taken orally because curare compounds are too large and too highly charged to pass through the lining of the digestive tract to be absorbed into the blood. This explains how they can be used to kill prey that will be later ingested.

Ageltoxin (agatoxin) are arylamine toxins (the α-agatoxins) found in the venom of the spider Agelenopsis aperta. They paralyze insects by blocking glutamatergic neuromuscular transmission [76]. Ageltoxins are thought to be non-competitive channel blockers specific for NMDA receptors [76,77].

Conantokins are found in the venom from Conus fish hunting snails [78]. The conantokins (G, L, R and T) form a class of peptides that inhibit competitively NMDA receptors [79,80]. Interestingly conantokins possess a large number of γ-carboxyglutamic acid residues (Figure 7) [81]. One of the γ−carboxyglutamic acid residues is thought to participate in the selectivity of conantokin G [82].

Quisqualic acid is isolated from the seeds of Quisqualis indica. Quisqualic acid is an agonist at AMPA receptors [83]. Quisqualate used to be the prototypical ligand of AMPA receptors, which were therefore called quisqualate receptors. However, this name has been abandoned as quisqualate also acts at metabotropic glutamate receptors [83].

Kainic acid was first isolated from the red alga Digenia simplex, where it might play a defense role, [84] and is also found in other algae [85]. Kainic acid is the prototypical agonist defining the Kainate subtype of glutamate receptors. The toxin is associated with human poisoning through the consumption of mussels that eat the algae.

Purotoxin are P2X receptors modulators which were isolated from the venom of the wolf spider Geolycosa sp [86]. Later, purotoxin has been isolated and shown to be a non-competitive antagonist of the P2X3 receptors [87]. It was found to be a close homologue to other spider toxins of unknown function and a more distant homologue to toxins known to bind to other receptors [87]. However, none of the receptors tested, besides P2X3, were sensitive to purotoxin, which makes it to date the only toxin specific of P2X receptors [87].

The nAChR has the largest number of known toxins (Sections 2 and 5). It is probably not surprising that nAChR also has the largest number of toxins used medically. Interestingly, both agonists and antagonists are used differently to what is observed for most of the other LGIC.

None of the toxins targeting the other pentameric LGIC (GABA, Glycine, Serotonin) have been used in medical practice. Furthermore, none of the toxins targeting these receptors has entered clinical trials.

For GABA receptors, a patent proposes the use of either the agonist muscimol or the antagonist bicuculine for the treatment of myopia. Numerous, synthetic compounds targeting the GABA receptors are on the market, the most significant being barbiturates and benzodiazepines. These compounds act by increasing the mean open time of the ion channel.

Among synthetic compounds, anesthetics (e.g., propofol) potentiate Glycine receptors. However, propofol also acts on GABA receptors. Currently, no drug targets specifically Glycine receptors [110]. The amino-acid taurine, found in energizing sodas is an inhibitor of the Glycine receptor [111].

Synthetic antagonists targeting the Serotonin receptors are used as antiemitic drugs [112]. Similarly, natural compounds targeting the Serotonin receptors have antiemitic properties, e.g., ginger extracts and delta-9-tetrahydrocannabinol from Cannabis sativa.

Toxins targeting the NMDA receptors (Figure 7, top) have neither been used medically nor tested clinically. Synthetic compounds targeting the NMDA receptors in clinical use are mainly channel blockers (Ketamine, Memantine, Amantadine) but also antagonists (Felbamate) [113]. Partial agonists have also been tested clinically (GLYX-13 and Acomposate).

Conantokins G and T have been noticed to display antinociceptive [114,115] and anti-epileptic properties [4,116]. Some conantokins demonstrate receptors subunit selectivity, which makes them attractive drug candidates [117].

Domoic acid containing algae are used as vermifugal agents in Japanese traditional medicine [118].

Toxins targeting the AMPA receptors (Figure 7, middle) have neither been used medically nor tested clinically. The only drug on the market targeting AMPA receptors is the potentiator Aniracetam (in Italy and Greece). Drugs targeting the AMPA receptors that entered clinical trials were either antagonists (e.g., E2007/perampanel advanced in phase III) or positive allosteric modulators.

Kainic acid is used in Chinese and Japanese traditional medicines as an anthelmintic to treat ascarialis [118]. Kainic acid is the only toxin targeting the kainate receptors (Figure 7, bottom) used as a treatment, and none have been tested clinically.

No drug targeting P2X receptors is on the market yet [119] and only very few clinical trials have been performed on that target (CE-224535 and GSK-1482160, which are not toxins). This observation can probably be explained by the very recent discovery of this family of LGIC. However, P2X receptors are attracting a lot of interest from pharmaceutical companies as shown by the significant number of patents filled recently [120]. All of the compounds proposed to target P2X receptors are antagonists. In agreement with this observation, among toxins targeting the P2X receptors (Figure 8), Purotoxin has been proposed for the treatment of pain based on the observation of antinociceptive activity in animal testing [87].