Neuropeptide signalling systems - An underexplored target for venom drug discovery.

Neuropeptide signalling systems are ubiquitous throughout the animal kingdom and regulate many fundamental physiological processes. They typically contain multiple receptor subtypes with complex signalling cascades that remain poorly understood due to a lack of receptor subtype-selective ligands. Because of their high level of conservation, endogenous neuropeptides, and in particular venom neuropeptides, represent a source for novel neuropeptide receptor ligands with varying pharmacological properties. Known venom neuropeptides are likely to be the tip of the iceberg and technological advances in mass spectrometry and sequencing technology combined with the development of novel bioinformatics tools promise to facilitate the identification of venom neuropeptides and characterisation of neuropeptide signalling systems. However, many challenges remain including a limited venom species diversity and identification of PTMs. Nevertheless, neuropeptides recruited into animal venoms offer a viable source for novel pharmacological tools and therapeutic leads to study the role of neuropeptides in health and disease.

Neuropeptides are implicated in multiple physiological processes and, consequently, numerous disease processes [ 47 ]. Disease areas where neuropeptides play a key role and where neuropeptide signalling systems are a validated therapeutic target or are actively being investigated as a therapeutic target include obesity, heart failure, epilepsy, sleep disorders, autism and depression [ 47 – 55 ]. Both neuropeptide analogues and small molecules targeting neuropeptide receptors have been developed for therapeutic treatments ranging from cardiovascular disorders, diabetes, pain, and labour to gastrointestinal disorders. Table 1 provides an overview of approved drugs targeting neuropeptide signalling systems. A range of other neuropeptide targets are also under investigation [ 47 ]. There are clear examples of the clinical importance of neuropeptides including opioid analgesics, insulin treatment for type 2 diabetes mellitus, oxytocin to induce labour, and parathyroid hormone to treat osteoporosis ( Table 1 ). However, many trials targeting neuropeptide signalling systems have been disappointing in part due to lack of efficacy, receptor dimerization, unexpected side effects, and dosage-dependent effects [ 41 , 47 , 56 , 57 ]. One potential cause of lack of efficacy is the complexity of neuropeptide signalling, which displays temporal variations and signalling redundancy, but also a lack of knowledge on signalling mechanisms [ 41 , 47 , 58 ]. Consequently, it is no longer the expectation that a single neuropeptide will cure a single disease and there is much scope for targeting neuropeptide signalling systems for therapeutic treatment in a wide range of diseases [ 41 ]. Neuropeptides, and in some cases neuropeptide precursors, are also valuable biomarkers and diagnostic probes to aid disease diagnosis and prognosis ( Table 1 ). As the various roles of neuropeptides in (patho)physiological processes are elucidated, their value as biomarkers and diagnostic markers is often assessed. Neuropeptide signalling systems represent a great opportunity for the development of therapeutics, in particular for peptide therapeutics. An in-depth understanding of the mechanistic roles of neuropeptides in health and disease is essential to translate their therapeutic potential into drug targets, therapeutic leads, diagnostics and biomarkers. Although technical advances within next-generation sequencing and mass spectrometry have greatly facilitated neuropeptide discovery and characterisation of their signalling systems [ 59 – 62 ], most neuropeptide signalling systems remain poorly understood [ 14 ]. In addition to their complex signalling pathways (multiple G protein-dependent and -independent pathways), neuropeptides have numerous characteristics that make them difficult to identify using mass spectrometry and sequencing technology, especially because of the high occurrence of PTMs, and the existence of isoforms [ 62 ]. Identification based on mass is further complicated by (i) varying and typically low in vivo concentrations (up to 1,000-fold lower than classical neurotransmitters); (ii) lack of enzymatic fragments, requiring highly-sensitive proteomic detection methods; and (iii) rapid in vivo degradation [ 62 ]. Consequently, the therapeutic potential of many neuropeptide systems remain unrealised [ 41 ]. Subtype selective ligands that can selectively inhibit or stimulate neuropeptide receptors are key pharmacological tools to dissect their functions in health and disease [ 58 ]. These ligands also constitute therapeutic leads. Such pharmacological toolboxes already exist for some neuropeptide signalling systems. For instance, bradykinin acts via two receptors, B1, B2, and receptor-selective agonists and antagonists have been developed for both receptors [ 63 , 64 ].

Venom and neuropeptide research employ comparable strategies and technologies to identify peptides and face similar challenges including low peptide concentrations, PTMs, variable size, high degree of structural diversity and a large number of isoforms [ 60 , 62 ]. A combined approach of mass spectrometry, sequencing technologies (i.e. transcriptomics) and bioinformatics (termed ‘Integrated Venomics’) has been very successful for both venom characterisation [ 77 , 102 , 140 , 145 ] and neuropeptide identification [ 29 , 59 , 154 ]. Mass spectrometry is very sensitive, fast, allows de novo sequencing and enables identification of PTMs, although more material is required compared to transcriptomics [ 60 , 62 ]. For animals with no or only a partial reference genome, transcriptomics in particular has greatly accelerated characterisation of animal venoms and identification of neuropeptides [ 140 , 155 ]. Advances in sequencing technology and mass spectrometry combined with in silico prediction and bioinformatic tools has facilitated research in both areas [ 59 , 60 , 150 , 156 ]. PTMs include the precursor protein cleavage sites that produce the mature peptides as well as residue modifications and can only be partially predicted from the nucleotide sequence [ 151 ]. Although advancements in mass spectrometry have increased selectivity and sensitivity, PTM determination of low abundance peptides remains challenging and requires sample preparation and separation methods such as liquid chromatography or capillary electrophoresis to maximise sensitivity [ 59 , 62 ]. This is particularly relevant for venom characterisation of animals that produce very small amounts of venom. D-amino acids, a PTM identified in spider, mollusc and mammalian venom peptides, as well as mollusc neuroexcitatory peptides and crustacean neurohormones, are notoriously difficult to detect due to a lack of a sequence change or mass defect [ 157 ]. New mass spectrometry methods have been developed that measure the distinct molecular fragmentation patterns among peptide diastereomers with tandem mass spectrometry (MS/MS) [ 59 ]. Computational methods are necessary to filter, process, store and integrate the multifaceted information generated in mass spectrometry and sequencing studies. In addition, bioinformatic tools have been developed to facilitate neuropeptide and venom peptide identification including prohormone prediction tools (NeuroPID [ 158 ]), precursor protein cleavage site prediction tools (NeuroPred [ 159 , 160 ], ProP [ 161 ], SignalP [ 162 ]), and peptide databases (neuropeptide database NeuroPep [ 2 ] and venom peptide databases [ 163 – 166 ]). The advancement and development of novel bioinformatic tools is critical to meet the computational demands of high-throughput mass spectrometry and sequencing technology which produce ever larger and more complex datasets [ 60 ]. Identification of a peptide, be that a neuropeptide or venom peptide, using mass spectrometry or polynucleotide sequencing technology does not provide information on the peptide target. After being identified with an ‘omics’ approach, peptides are typically synthesised and their activity characterised. Fortunately, synthesis of peptides (under 50 residues) is relatively straightforward using solid-phase peptide synthesis and the use of selective amino acid protecting groups allows directed disulfide-bond formation, which is important for characterising disulfide-rich peptides displaying potentially several disulfide bond isomers [ 167 , 168 ]. In addition, recombinant expression, native chemical ligation, semisynthetic methods, and ligase/enzymatic ligations enable synthesis of longer and more complex peptides [ 100 , 169 , 170 ]. Subsequent testing on high-throughput assays such as fluorescent plate readers which measure Ca 2+ , Na + , K + signalling (such as Fluorescence Imaging Plate Reader, FLIPR), electrophysiology, or single cell assays enables assessing of activity at various receptors [ 171 , 172 ]. However, in vitro characterization of venom peptides provides limited insight into their function, and often unexpected effects are observed once administered in vivo [ 173 ].

Endogenous neuropeptides from animals throughout the animal kingdom represent a solid starting point for discovery. Neuropeptide receptor sequence conservation means that there are numerous occurrences of neuropeptides from one phylum activating the corresponding receptors in another phylum. For example, vertebrate substance P is able to activate a Drosophila tachykinin receptor expressed in Xenopus oocytes [ 65 ], and mammalian neuropeptide Y peptides can activate the Drosophila neuropeptide F receptor, the invertebrate neuropeptide Y receptor orthologue [ 66 ]. Small sequence variations in ligands and receptors across species result in diverse pharmacological profiles including differences in selectivity, affinity and function (e.g. agonism, antagonism, and biased signalling) [ 67 , 68 ]. This diversity can be exploited to benefit human health as neuropeptides identified from different organisms represent a rich source for molecular probes and therapeutic leads to dissect neuropeptide signalling in health and disease. This hypothesis has been underpinned by multiple studies, including new oxytocin and vasopressin ligands identified from insects [ 69 , 70 ], salmon calcitonin used to treat hypercalcemia [ 71 ]; and pramlintide (based on rat amylin) to reduce blood glucose levels in type 1 and type 2 diabetes mellitus patients [ 72 ]. Another, and maybe even richer, source for ligands targeting neuropeptide signalling systems is animal venom. Venoms are highly complex mixtures mostly made up of proteins and peptides (~90%) that disrupt normal biochemical and physiological processes in the prey/predator for the benefit of the venomous animal [ 73 – 76 ]. Venom peptides have an array of targets including receptors and enzymes affecting biological functions in various locations including the CNS, PNS, blood and muscle tissue [ 77 ]. Importantly, venom components have undergone strong natural selection for in vivo stability, potency, selectivity and delivery to engage in precise molecular interactions exogenously to warrant their survival through defence and predation [ 73 – 76 , 78 , 79 ]. Many of these features are also highly desirable in drug development, thus venoms represent an advantageous drug discovery starting point compared to combinatorial or DNA/RNA display libraries [ 78 , 80 – 83 ]. Indeed, venoms are a highly successful source for drug leads with five venom-derived peptide drugs on the market, including Exenatide (Byetta ® ) and Lixisentide (Adlyxin, Lyxumia) from the Gila monster to treat type 2 diabetes [ 84 , 85 ]; Eptifibatide (Integrilin ® ), an antiplatelet agent from the rattlesnake Sistrurus miliarius barbourin [ 86 ]; and Bivalirudin (Angiomax, Angiox) from the leech Hirudo medicinalis , which is used as an anticoagulant agent [ 87 ]. Furthermore, conotoxins from the marine predatory cone snail have transformed neuropathic pain research with one conotoxin drug approved to treat severe chronic pain (Ziconatide, Prialt ® ) [ 88 ]. Spider toxins have also provided outstanding therapeutic leads for stroke, pain, and epilepsy through their actions on ion channels Nav1.1 [ 89 , 90 ], Nav1.7 [ 91 , 92 ], and acid-sensing ion channels 1a (ASIC1a) [ 93 ], respectively. Chlorotoxin, isolated from the venom of the scorpion Leiurus quinquestriatus and initially characterised as a voltage-gated chloride channel ligand [ 94 ], was subsequently identified as a potent and selective matrix metalloproteinase-2 receptor ligand and is currently undergoing clinical trials to visualise glioblastomas [ 95 ]. Finally, venom peptides and their use as receptor-subtype selective pharmacological probes have revolutionised our understanding of many human receptors and links to disease [ 96 – 102 ]. For instance, bradykinin-potentiating peptides, originally identified in the venom of the snake Bothrops jararaca , led to the development of a new class of hypotensive drugs, angiotensin converting enzyme inhibitors, including captopril, for the treatment of high blood pressure and heart failure [ 103 ]. Venom proteins and peptides are a result of gene duplication and selective expression of an endogenous protein/peptide in the venom gland, a process called neofunctionalization [ 73 , 75 , 76 , 104 – 107 ]. Neuropeptides are no exception, and at least 176 different neuropeptides from 16 different neuropeptide families have been found in the venom gland transcriptomes or venoms of 107 different species, including spiders, scorpions, insects, centipedes, cone snails, octopus, monotremes, snakes, lizards and fish ( Table 2 ). There exist several examples where venom neuropeptides have demonstrated their therapeutic potential. Exendin-4, a GLP-1 receptor agonist isolated from the venom of the Gila monster ( Heloderma suspectum ), was used to develop Exenatide as well as Lixisenatide for the treatment of type 2 diabetes mellitus [ 85 , 108 ]; contulakin-G, a neurotensin homologue isolated from Conus geographus venom is under investigation for the treatment of acute and chronic pain [ 109 ]; natriuretic peptides found in snake venoms are used to understand human natriuretic peptide receptor signalling [ 110 ]; sarafotoxin, identified in the venom of the snake Atractaspis engaddensis, acts on endothelin receptors, has been used to understand endothelin receptor signalling and aided development of endothelin receptor antagonists for the treatment of cardiovascular disorders [ 111 ]; conopressin-T, a vasopressin-like peptide identified in the cone snail Conus tulipa venom, contributed towards the development of novel vasopressin receptor antagonists that are used for the treatment of hyponatremia [ 112 ]. Although venom neuropeptides have been utilised for their therapeutic potential, very little is known on the physiological role venom neuropeptides play in predation and defence. Venom natriuretic peptides and sarafotoxins likely contribute to prey capture by targeting the cardiovascular system and affecting vasodilation or vasoconstriction and heart contractility [ 110 , 113 , 114 ]. Cono-insulins reduce blood glucose levels in fish which results in sluggish behaviour facilitating capture [ 115 ]. Pain and inflammation are common responses to envenomation, often for defensive purposes [ 116 ] and for envenomation of mice by the wasp Polistes lanio lanio these effects were attributed to tachykinin-like peptides acting on the neurokinin-1 (NK1) receptor [ 117 , 118 ]. Conorfamide-Sr2, an FMRF-amide related peptide isolated from worm-eating Conus spurius venom, was hypothesized to have a defensive function as it had mild paralytic effects in the limpet Patella opea [ 119 ]. In addition, based on the neuropeptide families they belong to, venom neuropeptides are implicated in various physiological processes including cardiovascular functions (natriuretic peptide [ 120 ], crustacean cardioactive peptide [ 121 ], RFamide [ 122 ], neurotensin [ 123 ]), metabolism (insulin [ 124 ], glucagon-like peptide-1 [GLP-1] [ 125 ], RFamide [ 122 ]), reproduction (oxytocin [ 16 ]), fluid homeostasis (natriuretic peptide [ 126 ], vasopressin [ 44 ], bradykinin [ 127 ]) and pain (tachykinin [ 128 ], bradykinin [ 25 ], enkephalin [ 129 ], RFamide [ 122 ]). In addition to venom neuropeptides with high homology to the human neuropeptides, venoms also contain peptides with very little homology to neuropeptides that potently bind to neuropeptide receptors. Venom peptides are generally characterised by well-conserved disulfide bond frameworks with loops that are highly mutable, thereby providing an evolutionary easily amendable scaffold able to adapt to a wide range of targets [ 76 , 130 ]. A good example of this is mambaquaretin-1 isolated from the green mamba ( Dendroaspis angusticeps ), which binds allosterically to the vasopressin V2R (GPCR) and protects against renal cyst development, making it an interesting therapeutic lead for the treatment of polycystic kidney disease [ 131 ]. This was unexpected, since Mambaquaretin-1 has a Kunitz-fold, which is a well-characterised scaffold with three disulfide bonds, two antiparallel beta strands and a short alpha helix [ 131 , 132 ] that is commonly found in venoms and known to target ion channels and proteases [ 110 , 131 , 133 , 134 ]. Other examples of venom peptides that have no homology to neuropeptides but target neuropeptide receptors include hypotensin-I isolated from the scorpion Tityus serrulatus and which is a B2 bradykinin receptor agonist [ 135 ]; d-ctenitoxin-Pn1a from the spider Phoneutria nigriventer that targets opioid receptors [ 136 ]; t-CnVA and conorphins from cone snail venom that target the somatostatin and opioid receptors, respectively [ 137 , 138 ]; and Crotalphine isolated from the venom of the snake Crotalus durissus terrificus that targets the k-opioid receptor and causes analgesia in rats [ 139 ]. The overall chemical and functional diversity of venom toxins is what renders venoms such interesting sources for the discovery of novel drug leads, diagnostic probes and pharmacological tools, which could accelerate our understanding of neuropeptide signalling in humans, if explored properly. The number of venom neuropeptides listed in Table 2 is likely to represent only a small fraction of the total number of neuropeptides recruited into animal venoms as a result of limited characterisation. The broad phylogenetic range of venomous animals and number of known venom compounds (~6,000) versus the number of known venom neuropeptides suggests there are many more venom neuropeptides to be discovered [ 76 , 140 ]. One explanation for the low number of neuropeptides is that only a limited number of venoms have been characterised to date and there is a bias towards certain animal lineages at the expense of others [ 140 ]. For example, snakes and cone snails have been heavily studied in contrast to invertebrates such as centipedes, cephalopods (octopus, squid and cuttlefish) and cnidarians [ 140 , 141 ]. In fact, it is hypothesised that less than 1% of the total number of venom proteins and peptides have been characterised [ 77 ]. Another reason for the low number of known neuropeptides in animal venoms may have to do with the methods employed to characterise them. The majority of neuropeptides listed in Table 2 were identified by transcriptomics. The transcriptome is the sum of all mRNA molecules in one cell or a population of cells [ 142 ]; it provides information on transcript expression level, transcriptome dynamics across different tissues or conditions, and is not dependent on an existing genomic sequence. Transcriptomics, often combined with proteomics, is the most common venom characterisation method and allows a non-biased and non-targeted approach to characterise venom glands [ 140 , 143 – 145 ]. However, it has also revealed high intraspecific variation due to diet, geographical location, age, and gender which necessitates care when interpreting results [ 146 – 148 ]. The correct and efficient use of bioinformatics is essential to be able to process and integrate the large amount of data generated in a transcriptomic study [ 149 ]. Annotation of transcriptome sequences relies on sequence comparison to known, annotated proteins and peptides in databases such as UniProt and the National Center for Biotechnology Information (NCBI) database [ 150 ]. The short length of neuropeptides and high sequence diversity of neuropeptide precursors makes them difficult to identify using sequence homology methods such as the NCBI Basic Local Alignment Search Tool (BLAST) [ 61 , 151 ]. Consequently, it is likely that a large number of neuropeptide homologues in venoms have not been detected in ‘omics’ data. Therefore, it seems reasonable to expect identification of a higher number of neuropeptides in animal venoms using a targeted approach. This seems to have been the case in cone snails where neuropeptide-specific search in the venom gland transcriptome of Conus victoriae led to the identification of five different neuropeptides [ 152 ], in contrast to the two neuropeptides previously identified in C. victoriae [ 153 ].