Botulinum neurotoxins (BoNTs) are produced by anaerobic bacteria of the genus Clostridium and are the most poisonous biological substances known [1,2]. BoNTs are responsible for botulism, a neuroparalytic syndrome characterized by flaccid paralysis, originating in the peripheral nervous system, and by dysfunctions of the autonomic nervous system [3,4,5,6]. In food borne botulism, the most common form of botulism, BoNTs are ingested with food in which spores have germinated and the organism has grown. Food-borne botulism has accompanied mankind since the beginning (see [7] for extensive review). At the end of the 18th century, some well-documented outbreaks of sausage poisoning in Württemberg, Southern Germany, prompted early systematic botulinum toxin research. Between 1817 and 1822, the German poet and district medical officer Justinus Kerner published the first accurate and complete description of the symptoms of food-borne botulism. He did not succeed in defining the suspected biological poison that he called “sausage poison” or “fatty poison”; however, he developed the idea of a possible therapeutic use of the toxin. In 1895, Emile Pierre van Ermengem, Professor of bacteriology at the University of Ghent, following a botulism outbreak after a funeral dinner with smoked ham in the small Belgian village of Ellezelles, discovered the pathogen anaerobic gram positive bacterium Clostridium botulinum. The bacterium was so called because of its pathological association with the sausages (Latin word for sausage = botulus).

It was not until the middle of 20th century that BoNTs were used as therapeutic drugs in medicine. Research began when Dr. Edward J. Schantz and his colleagues were able to purify botulinum toxin serotype A (BoNT/A) into crystalline form. In 1953, the physiologist Dr. Vernon Brooks discovered that the injection of small amounts of BoNT/A into a hyperactive muscle blocked the release of acetylcholine (ACh) from motor nerve endings, causing temporary “relaxation”. In the 1960s, the ophthalmologist Dr. Alan B. Scott started treating monkeys with BoNT/A, theorizing that its muscle-relaxing effects might help in the treatment of crossed eyes (or strabismus). In the 1980s, Alan Scott introduced the use of BoNT/A as an alternative to conventional surgery to treat human strabismus. After these pioneering studies, therapeutic applications of BoNT/A were extended to a wide variety of neurological disorders associated to spasmodic muscle contractions, for example, the spasmodic torticollis, blefarospams, facial emispams, and dystonia [8]. Nowadays, the clinical indications for BoNT/A are constantly growing, ranging from treatment of overactive skeletal and smooth muscles, to management of hypersecretory and painful disorders such as migraine, trigeminal neuralgia and the myofascial pain syndrome [9,10,11].

This review is focused on studies in which BoNTs, mainly BoNT/A, were used as pharmacological treatment against pain in animal models.

From the plethora of toxins produced by Clostridium botulinum, only toxins whose action is restricted to the nervous system, such as BoNTs, are strictly defined as neurotoxins. Other toxins, such as botulinolisin, toxin-C2, exoenzyme-C3, which target different type of cells, mostly non-neuronal, are not included in the BoNTs family. Seven different serotypes of BoNTs have so far been characterized (A-G), and these serotypes are active on many different types of vertebrates [3,4]. BoNTs are secreted as multimolecular complexes together with non toxic accessory proteins (emagglutinins) that do not contribute to the paralyzing effects of BoNTs but protect them from the passage through the acidic and proteolytic environment of the stomach after the ingestion of spore-contaminated food [12,13].

BoNTs are proteins of about 1,300 amino acids and consist of three domains of similar size (50 kDa). The NH₂-terminal domain, which is named L-chain (L = light) domain, is a zinc endopeptidase that represents the catalytic domain expressing the protease activity. The other two domains, which are covalently bound to form the H-chain (H = heavy), are the central domain, responsible for the membrane translocation of the L-chain into the neuronal cytosol, and the COOH-terminal domain, which consists of two equally sized subdomains, responsible for the neurospecific binding [14,15,16].

The cellular action of BoNTs occurs as a four-step mechanism: (i) binding of BoNTs on the neuronal presynaptic membrane, via interaction with gangliosides, synaptic vesicle protein 2 and/or synaptotagmin, depending on the serotype [16,17,18]; (ii) internalization of BoNTs via endocytosis of the BoNTs-receptor complex inside the neurons; (iii) translocation of BoNTs L-chain from endocytosed vesicle to the neuronal cytosol; and, finally, (iv) zinc-endopeptidase activity on cellular targets [13,14,25]. Intracellular targets of BoNTs are three proteins involved in neuroexocytosis of the neurotransmitter synaptic vesicles. These proteins are: SNAP-25 (synaptosomal associated protein of 25 kDa), VAMP (vesicle associated membrane protein), also called synaptobrevin, and sintaxin. All these proteins are involved into assembly of the SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) protein core complex, which is fundamental for correct docking and fusion of neurotransmitter vesicles with neuronal membranes [13,14,15,16,17,18,19,20]. The peculiar characteristic of BoNTs resides in their high affinity for one of the three SNARE proteins: SNAP-25 is cleaved by BoNT/A (the serotype most commonly used in clinical practice), /E and /C; VAMP/sinaptobrevin is the target of BoNT/B (another serotype used in clinic), /D, /F and /G; and sintaxin is cleaved only by BoNT/C. The cleavage of one of these proteins is sufficient to prevent the correct assembly of the SNARE core complex and the consequent fusion of synaptic vesicles with the neuronal presynaptic membrane, thus inhibiting the neurotransmitter release [19,20]. This effect is reversible, and the duration of action is dependent on the serotype. Further details on binding, internalization and mode of action of the different BoNTs serotypes can be found in other reviews [21,22,23,24,25,26].

Studies on the pathophysiology of botulism revealed that neuromuscular paralysis is due to selective inhibition of evoked ACh release from cholinergic nerve endings at the skeletal neuromuscular junction [1]. This canonical effect of BoNTs gave the opportunity to use them as therapeutic agents in a variety of neurological disorders due to hyperfunctionality of cholinergic terminals [8]. However, BoNTs cannot be considered exclusively as ‘cholinergic’ toxins. Though they act preferentially on nerve terminals between motoneurons and muscle fibers, BoNTs can also block the neural transmission at other peripheral synapses, cholinergic or not [9]. This has indirectly been proved also by symptoms observed during botulinal neuromuscular paralysis, where autonomic nervous system dysfunction (parasympathetic, i.e., cholinergic, and sympathetic, i.e., adrenergic and noradrenergic) coexists with neuro-motor alterations [4,13].

Extensive in vitro and in vivo studies demonstrated that BoNTs (mainly BoNT/A) are able to block the Ca²⁺-evoked release of neurotransmitters, other than ACh [27]. As for examples, in vitro studies demonstrated that BoNTs inhibit various neurotransmitters’ release, including glutamate, GABA, aspartate, catecholamine, and noradrenaline, from cerebral synaptosomes, chromaffin cells, central neurons and hippocampal slice cultures [28,29,30,31,32,33,34,35,36,37,38]. In vivo studies, mainly using microdialysis techniques, demonstrated that BoNTs block both dopamine and monoamine release under various conditions [39,40,41,42]. Moreover, BoNTs block glutamate release also from non-neuronal cells, such as astrocytes [43,44,45]. Finally, there are many evidences in literature that BoNTs are able to block the release not only of classical neurotransmitters but also of neuropetides, such as Substance P (SP) [46,47,48,49] and calcitonin gene-related peptide (CGRP) [50,51,52,53,54,55]; neuropeptides whose role in pain modulation is known. These latter findings gave strong input in searches for alternative therapy against trigeminal neuralgia and various types of headache and migraine—neurological diseases where SP and CGRP release often act as a trigger point to exacerbate disease attacks. In these pathologies, BoNT/A probably exerts analgesia by blocking the neuropeptides’ release at the nociceptive nerve endings. Some encouraging results have been obtained in humans [56,57]. The effect of BoNT/A in tension-type headache, chronic daily headache, and/or migraine is not further considered in the present review.

BoNT/A is effective in relieving pain in spastic and non-spastic muscle conditions in humans. This analgesic effect has generally been attributed to muscular relaxation. Notwithstanding, there are reports in the literature stating that patients experience pain relief shortly after BoNT/A treatment [58,59], i.e., before of any muscle-relaxing action of the toxin, or that the pain relief is still maintained after muscle power returned to normality [60]. These results suggest that the analgesia attributed to BoNT/A may be due to more complex mechanisms than the simple muscular relaxation. In such cases, the pain relief cannot be ascribed to abolition of those factors, such as muscle tone, excessive muscle contraction and spasms, which are due to muscle hyperactivity [61]. In the next sections, we will discuss the evidences from animal models in favor of an analgesic action of BoNTs, with particular emphasis on pain not directly related to abnormal muscle hyperactivity, such as inflammatory and neuropathic pain. Animal model studies have been mainly performed with BoNT/A and /B and have been conducted by using the 150 kDa purified toxin or equivalent commercial products.

In the following, we will indicate BoNT/A, or /B, indifferently from which preparation has been actually utilized.

Generally, pain is divided into nociceptive and pathological pain. Nociceptive pain is caused by the sustained activation of peripheral nociceptors in response to peripheral tissue injury and can be classified as somatic or visceral pain according to where the pain arises. The extent of nociceptive pain normally reflects the extent of tissue damage and it disappears when the origin of nociceptive stimuli disappears. On the other hand, pathological pain often evolves toward a chronic condition in which pain exceeds the extent of tissue damage. Independently from the organs or tissues involved, chronic pain may be a consequence of either inflammatory processes (for example, rheumatoid arthritis in humans) and/or injury to the nervous system, both peripheral (PNS) (for example, postherpetic neuralgia) and central (CNS) (for example, spinal cord injury) nervous system, which is commonly known as neuropathic pain. A common feature of pain-related mechanisms is the activation of sensitization processes, plastic changes in PNS and CNS (adaptative or pathological) that lead to an enhanced response and/or decreased threshold to nociceptive stimuli. For example, under neuropathic pain conditions, both hyperalgesia (an exaggerated pain perception in response to a painful stimulus) and allodynia (the perception as painful of a stimulus that does not usually provoke pain) are commonly developed. In basic science, many animal pain models have been developed in order to mimic human pain conditions. In the next paragraphs of this section, an overview of the studies involving the use of BoNTs in animal pain models will be presented and relevant results discussed.

All the evidence in favor of analgesic properties of BoNT/A in a wide variety of pain models pose some interesting questions about the mechanism of action. In the recent years, two main points emerged about the effects of botulinum neurotoxins: the first one is that these molecules are effective not only in peripheral but also in the central nervous system. The second important point is that part of the effects induced by BoNT/A administration are observable distant from the site of injection.

Under therapeutic treatment of muscle hyperactivity conditions, BoNT/A is locally applied, with little or absent diffusion, and its paralyzing action remains confined to the nerve-muscle junction, close to the injection site. This assumption constituted a “dogma” for many years of the use of these neurotoxins in human therapy. Some experimental evidence obtained in recent years from basic scientific research challenge this dogma and raise concern that while most of the effects are localized close to the injection site, BoNT/A can also act at distant sites. Actually, the possibility that BoNT/A could reach the CNS by retrograde transport was already suggested many years ago by experiments with radiolabeled BoNT/A [112,113]; however, it was observed that the retrograde axonal transport was so slow that the toxin was likely to be inactivated before it reached the cell soma [114].

More recent studies support the retrograde transport of BoNT/A. In particular, Antonucci et al. [115] demonstrated that BoNT/A may be retrogradely transported by central neurons and motoneurons and then transcytosed to afferent synapses. In their very elegant work, the authors presented three pieces of evidence in favor of axonal migration of BoNT/A. First, they showed that after a unilateral intrahippocampal injection of BoNT/A, cleaved SNAP-25 was detectable also in the contralateral untreated hemisphere. Second, after injection of BoNT/A into the optic tectum, cleaved SNAP-25 appeared also in synaptic terminals within the retina. Since the natural target of BoNT/A is the neuromuscular junction, in a third experiment, authors chose to test the spread of toxin in the facial motoneurons projecting to the whisker muscles. After BoNT/A injection at the center of the whisker pad, cleaved SNAP-25 was detected in the facial nucleus, confirming a possible migration of toxin also along motoneurons. Summarizing, after BoNT/A injection into central and peripheral regions, truncated SNAP-25—the target protein of SNARE complex selectively cleaved by BoNT/A—appeared not only at the injection site but also in distant regions projecting to the infusion area. The retrograde spread was blocked by the microtubule depolymerizing agent colchicine, pointing to an involvement of microtubule-dependent axonal transport. In another research, when BoNT/A was injected in the whisker pad area, Kitamura et al. [105] observed a strong inhibition of the increased KCl-evoked neurotransmitter release from trigeminal ganglion neurons after infraorbital nerve constriction.

Similarly to Antonucci et al. [115], the results reported in another study [116] may not be completely accounted for without considering a possible axonal transport of the toxin distant from the injection site. Bach-Rojecky and Lackovic [116] used the model of unilateral injection of acidic saline solution into gastrocnemius muscle as an animal model of muscle hyperalgesia. In this model, both ipsi and contralateral hyperalgesia are detectable, with contralateral hyperalgesia being considered as centrally mediated secondary hyperalgesia. BoNT/A, subcutaneously injected into ipsilateral hindpaw, exerted antihyperalgesia both on ipsi- and contralateral hindpaws. The effect on both sides was evident on day 5 and was of similar intensity. Interestingly, when colchicine was injected into ipsilateral sciatic nerve, one day before the ipsilateral BoNT/A injection, the antihyperalgsic effect of BoNT/A on both sides was not observed. On the contrary, if colchicine was injected into contralateral side, opposite to the site of pain induction and BoNT/A injection, it did not prevent the antihyperalgesic effects of BoNT/A on both sides. Finally, when BoNT/A was applied intrathecally, the bilateral hyperalgesia was also reduced. Altogether, these results cannot be explained without the involvement of the central nervous system and without the assumption that BoNT/A is transported from the site of injection. Bilateral effects after unilateral BoNT/A injection were also observed in paclitaxel-induced neuropathy in rats [73], and in streptozotocin induced diabetic neuropathy in rats [107].

Even if an easy generalization of the results derived from animal studies to humans is not recommended and a number of factors (anatomical differences, doses, volume of dilution, etc.) have to be taken into account (see [117]), these studies represent a crucial step in the comprehension of botulinum neurotoxin transport mechanism.

Considering all the results reviewed, we suggest a possible model of action of BoNT/A on pain transmission, as in the scheme depicted in Figure 3. In this scheme, two pain conditions are considered: inflammatory, such as that induced by chemicals (formalin, capsaicin, etc.) or neuropathic pain derived from nerve injury (CCI and spinal nerve ligation or transection). BoNT/A may be administered peripherally by subcutaneous or intraplantar injections, or centrally by intrathecal or intracerebroventricular injections. Depending on the site of injection, BoNT/A may differently exert its effects. By inhibiting the release of neurotransmitter and/or neuropeptides from nociceptive endings, peripherally injected BoNT/A may reduce directly the peripheral sensitization and indirectly the central sensitization, (Figure 3, point 1). In the same way, intrathecally injected BoNT/A may inhibit the release of neurotransmitters and/or neuropeptides from central terminals of nociceptive afferents, reducing the central sensitization (Figure 3, point 2). Finally, peripherally injected BoNT/A may be retrogradely transported along axons of peripheral nerves, thus acting also on central sensitization processes (Figure 3, point 3). This retrograde transport may be responsible for the displacement of peripheral injected BoNT/A at the level of the spinal dorsal horn where it exerts a possible inhibition of spinal release of neurotransmitters. Moreover, in a neuropathic pain model, the retrograde transport of BoNT/A, after being peripherally injected at the level of the paw, may be involved in the observed regenerative processes at the level of injured nerve [91] and the inhibitory effects at level of DRG and spinal cord [103].

Whatever the mechanism, the analgesic effects observed, as well as the lack of deleterious side effects, imply that, while experimental research is going to continue, the potential use of BoNT/A for therapy is strongly supported.

In summary, BoNTs are zinc metalloendoproteases that exhibit extraordinary specificities for proteins involved in the neurotransmitter release process, whose toxicity makes them responsible for animal and human botulism. However, these extremely poisonous molecules can become useful therapeutic agents in a number of expanding applications in human medicine, including modulation and alleviation of pain conditions. This is particularly evident for one of the BoNTs serotypes, the serotype A. One of the reasons for this success is the evident advantage of using a drug with a prolonged duration of action, thus allowing a long interval between treatments, as demonstrated by all the studies reviewed in this present review.

Another important aspect is that a single injection of BoNT/A is able to induce long-term analgesic effects in many different pain models, from inflammatory to neuropathic. The indications for BoNTs therapy will, no doubt, continue to expand and ongoing efforts to elucidate BoNTs’ mechanisms of action for reducing pain will provide an essential foundation for developing future therapeutic strategies. Although BoNTs may benefit pain syndromes and can theoretically be administered to treat many pain conditions, its use appears restricted to peripheral administration and the possible lethal consequences of systemic administration of the toxin limit its potential for clinical trials. It should be recognized that peripheral administration is also not completely devoid of risk and needs further investigations, with particular reference to the central action. Much evidence has been presented for retrograde transport of BoNT/A distant from the site of injection and more in depth studies will be necessary to characterize the mechanisms of this retrograde transport. Among others, as suggested by Caleo et al. [118], an important aspect that has to be better clarified is whether the central effects depend on the dose used as well as on density of innervations and levels of expression of toxin receptors.

In an attempt to bypass the risks due to toxicity, a desirable characteristic for future BoNTs products would be an increased specificity for pain conditions. This may be achieved by replacing the native binding domain with another protein to re-direct the light chain to a different nerve or cell [119]. Such modifications may enable BoNTs to treat pain without engendering weakness of nearby muscles. In this context, an innovative class of biopharmaceuticals obtained by recombinant techniques has been recently proposed [120,121,122,123,124,125]. These recombinant proteins, also named as “targeted secretion inhibitors”, incorporate the light chain fragment of BoNT/A expressing the endopeptidase activity fused with a protein that selectively binds to specific neuronal or non-neuronal cells. A promising example of these proteins is the recombinant protein obtained by fusion of the light chain of BoNT/A with lecitin from Erythrina cristagalli, which binds to galactose-containing carbohydrates selectively present on nociceptive afferents in the central and peripheral nervous system. This recombinant protein has been demonstrated to be effective in inhibiting the release of SP and glutamate from DRG neurons in culture [122] and to relieve pain symptoms in different pain models [123].

Since BoNT/A is actually under clinical trials for treatment of pain in various types of headache and migraine conditions [57], the translation of encouraging results from preclinical studies in inflammatory and neuropathic animal pain models, to clinical treatments of chronic pain in humans can be considered a crucial step for human health. However, more in depth researches are still necessary to better establish the exact mechanism responsible for analgesic effects of BoNT/A. This will improve our knowledge about this relevant thematic of research and ongoing efforts to elucidate BoNTs’ mechanisms of action for reducing pain will provide an essential foundation for developing future therapeutic strategies.