Natural Products and Ion Channel Pharmacology

Productivity crisis in the pharmaceutical industry

The lack of productivity in research and development (R&D) is widely cited as the greatest problem facing the pharmaceutical industry today [1,2,3]. The industry has increased R&D spending at an average annual rate of 12.3% since 1970. In aggregate, R&D spending increased over 80 fold between 1970 and 2008 [4]. However, the dramatic increase in R&D spending has not resulted in a commensurate increase in new therapeutic drugs. In fact, the number of new drugs approved annually by the United States Food and Drug Administration (FDA) has declined since the mid-1990s to levels seen in the 1970s and 1980s, when R&D spending was much lower [3]. One metric of R&D productivity in the pharmaceutical industry is total R&D cost per new molecular entity (NME) approved, where NME is defined as a novel drug not previously approved for marketing in any form or for any other indication. As a consequence of dramatically higher R&D spending and lower NME approvals, the cost per NME is approaching $2 billion, threatening the viability of the pharmaceutical industry [1,2,3,4]. Furthermore, pharmaceutical companies are facing a “patent cliff,” where they are expected to lose $140 billion in annual sales between 2007 and 2016 from patent expirations, with few potential blockbuster drugs in the pipeline to replace them [5].

High pre-clinical and clinical attrition rates for candidate compounds

It is perplexing that productivity has declined in the pharmaceutical industry, especially when one considers that this has happened concurrent with revolutionary advances in science. It has become progressively easier to identify and validate drug targets. But it still remains a significant challenge to find compounds with the selectivity, potency and pharmacological properties of a viable therapeutic drug candidate, as demonstrated by the high attrition rate of compounds both in clinical and pre-clinical testing. One study indicated that only 11% of drugs successfully progress from first-in-man to FDA approval (89% attrition rate), while other industry experts have claimed that the attrition rate is actually greater than 90% [2]. Clinical safety and toxicology account for approximately 30% of attrition in the clinic [2], suggesting that in many cases failures in the clinic are due to a lack of selectivity for the drug's target. A lack of selectivity is also a major problem in pre-clinical attrition. Even after extensive in vitro testing of a promising compound, when applied to animal models, undesirable off-target effects are often discovered, making a candidate compound unsuitable as a therapeutic drug. Although polypharmacology (one drug targeting multiple proteins in concert) has been proposed as a potentially new paradigm in drug discovery, it is still essential for any drug to hit therapeutically useful targets and to avoid interacting with all other proteins that may produce harmful off-target effects [6]. Such off-target activity has contributed significantly to the high pre-clinical and clinical attrition rate of candidate molecules.

Targeting selectivity of natural products vs. random small molecules

An underappreciated fact that may account for much of the off-target activity of failed compounds is that most drug-discovery projects now begin with combinatorial libraries of random small molecules, rather than natural products: the historical source of pharmaceutical drugs (e.g. penicillin, paclitaxel, opiates, etc.). Over the past two decades, concurrent with the productivity decline in the pharmaceutical industry, the major strategy adopted by the pharmaceutical industry to find starting points for medicinal chemistry has been high-throughput screening of combinatorial libraries. This approach has indeed yielded compounds that hit the molecular targets of interest. However, even after more than two decades of combinatorial chemistry, there is only one de novo combinatorial new chemical entity that has been approved by the FDA, and the majority of drugs on the market are still derived from natural sources [7,8,9,10].

While high-throughput screening is a powerful tool for finding compounds from random chemical libraries that interact with the molecular target of interest, one can never screen all other targets that the compound might also potentially interact with, and most of the compounds in large combinatorial libraries have groups suitable for interacting with sites on proteins that are not unique to a single molecular target. Such molecular scaffolds are not necessarily a good starting point for multiple costly cycles of medicinal-chemistry modifications and functional testing, because untested off-target side effects will likely remain even after medicinal-chemistry optimization.

With respect to targeting selectivity, natural products may have a significant advantage. Over millions of years, nature has explored chemical-diversity space extensively with a vast array of organic/biological scaffolds, and we are probably far from having elucidated a significant fraction of these. It is likely that evolutionary forces select against scaffolds that cause a high degree of non-specific interactions with many biological molecules. In a sense, natural products have been field tested by evolution. Structural scaffolds on which many related compounds are based have been under intense selection, both for targeting selectivity and high potency for many millions of years, and therefore off-target properties have been weeded out. We might think of the evolutionary forces that have shaped natural products as being similar to millions of successive rounds of medicinal-chemistry cycles and clinical trials. In contrast to the evolutionary refinement of natural products for potency and selectivity, even a very large combinatorial library is woefully inadequate at exploring chemical-diversity space. Estimates for the possible number of unique organic molecules of reasonable molecular weight (drug like) that could be produced range from 10¹³ to 10¹⁸⁰, while the number of known compounds is on the order of 10⁸ [11]. One estimate of the potential number of chemical structures that could be produced with a molecular weight less than 500, stable in oxygen and water at room temp, and composed of only H, C, N, O, P, S, F, Cl and Br is 10⁶³ [11,12]. Thus, even a very large combinatorial library (e.g. 10⁶ molecules) barely scratches the surface of chemical diversity space, whereas natural products have had eons to explore chemical diversity space to maximize potency and selectivity. Therefore, a natural product found to functionally alter a desirable molecular target for a therapeutic application would be less likely to have off-target complications than a compound from a combinatorial library.

The discussion above is a general one that applies widely to all pharmacological agents, and all molecular targets. However, these considerations are especially important for ion channel pharmacology. Functional ion channel complexes can be grouped into superfamilies that in turn are subdivided into ion-channel families with multiple homologous subunits. In most cases, an individual subunit is then assembled into a tetrameric or pentameric complex; homologs within the same ion channel family can co-assemble to form various heteromeric combinations. This generates an enormous diversity of closely-related ion channel isoforms. The pharmacological agents that are often the most useful are those that can discriminate between the different subtypes in a given ion channel family. Generating the required selectivity to discriminate between closely-related isoforms, together with the necessary potency to make a compound an effective pharmacological agent is a formidable challenge; avoiding other off-target interactions makes this even more difficult. Thus, all of the issues presented above become particularly critical whenever an ion channel is the physiological entity that needs to be targeted.

A concerted discovery strategy for natural products

Although natural products have built-in selectivity advantages, we argue that many of the previous natural-product discovery efforts have been misguided. The bioprospecting approach to natural-products discovery is a chemistry-centric strategy that aims to elucidate the structure of novel bioactive molecules from the natural world [13,14,15]. The historical approach for exploring molecular diversity from plants and animals is to collect a large quantity of tissue (typically 1kg), extract compounds and run tests to determine whether there are any pharmacologically-active components in the extract. A potential problem with this approach is that it only detects compounds that are sufficiently potent, and present at sufficiently high levels, to score on the assay(s) used [13,14]. If one takes a kilo of total animal tissue, many compounds that the animal may have evolved may not be present at sufficiently high concentrations to be detectable.

In contrast to the bioprospecting approach, a focused biology-centric strategy has successfully identified many natural compounds, primarily from venomous animals, that are pharmacologically active on ion channels. The study of venom components was initiated mostly by biologists interested in the physiology of prey capture and defense employed by venomous animals. As a result, it was automatically assumed that the compounds of interest would be synthesized in the venom apparatus of the animals and instead of taking a kilo of a particular snake and extracting all compounds, venom was collected and stored or the venom apparatus was dissected. Thus, the specialized fluid produced or the tissue of the animal that produces the pharmacologically-active compounds serves as the starting material, and not mixed tissue extracts. We suspect that if the standard bioprospecting approach had been taken with spiders or cone snails, and one kilo of total tissue were first extracted, since the venom duct of these animals is only a very small fraction of the total body mass, none of the pharmacologically-active compounds targeted to ion channels would have been detected. Although this seems almost a trivial factor, it is a crucial consideration for the discovery efforts for natural products targeted to ion channels.

The purification of α-bungarotoxin from the venom of the Taiwanese banded krait by Chang and Lee [16] marked the historical inception of ion-channel pharmacology, establishing venom as an important source of pharmacological agents targeted to ion channels. The demonstration that α-bungarotoxin targeted the skeletal muscle subtype of nicotinic receptors was a keystone in molecular neuroscience as well, since the toxin made the purification and definition of the first functional ion channel complex possible, and ultimately led to the characterization of the ligand-gated ion channel superfamily. Since that historic work, a large number of investigators have examined snake venoms for additional components targeted to ion channels.

In more recent decades, the expanding exploration of other animal venoms has led to these becoming sources of novel ion channel targeted pharmacological agents. Cone snails, spiders, scorpions and sea anemones all have provided natural products that have pharmacological effects on different types of ion channels. The range of venom components that target various ion channel subtypes has been steadily expanding, and compounds from venom remain a major resource for future discovery. In fact, recent mass-spectrometry profiling of crude venoms has suggested that the diversity of venom components may have been substantially underestimated previously; approximately 1000 different masses, representing unique compounds, have been detected in the venom of a single species of spider and cone snail [17,18,19]. With ∼700 species of cone snails and 41,000 species of spiders, the potential diversity of venom components in these species alone is enormous. One estimate of the total number of distinct peptides in all venomous animals is 20 million; only a tiny fraction have been discovered and characterized [18].

The pace of venom-peptide discovery historically has been dictated by bioassay-guided purification of venom components, followed by Edman sequencing. Recently cDNA cloning of venom-peptide transcripts [20,21,22,23] and de novo sequencing of venom peptides by tandem mass spectrometry is accelerating the pace of discovery substantially [18,24]. In one study employing a novel mass-spectrometry approach, 31 full peptide sequences were obtained from only 7% of the crude venom from a single cone-snail venom duct [25].

Prialt: a case study in natural-product selectivity

One success story in utilizing venom components for therapeutic drug discovery is the discovery of a peptide, ω-conotoxin MVIIA (Fig. 1), from the marine cone snail, Conus magus. When this peptide was originally purified from venom, it was assayed for biological activity by injection into fish and mice. When injected into the periphery of fish, it caused a delayed neuromuscular paralysis, which was not evident in mice. However, when it was injected intracranially in mice, it caused shaking [28,29]. These results pointed to a species difference in gene expression that ultimately culminated in a critical insight in natural product drug discovery: although orthologous genes (same gene separated by speciation events in evolution) may be highly conserved structurally across species, they may have a different tissue-specific expression pattern in different species. ω-Conotoxin MVIIA targets very selectively the N-type calcium channel (Ca_v2.2) [28,30,31], which is expressed in the motor neurons of fish (hence the peptide is paralytic to the cone-snail's fish prey) but is primarily expressed in the central nervous system in mammals, and is not the primary Ca channel subtype at the neuromuscular synapse. So, when the peptide is administered to the spinal cord of humans, it produces analgesia, rather than paralysis [26]. This difference in expression pattern allowed the peptide to be developed as a drug for intractable neuropathic pain, which was approved for marketing (as Prialt) by the FDA in 2004 in the United States and in Europe in 2005. One notable aspect of the pre-clinical drug development was that many analogs of the natural peptide were made in the attempt to improve the potency or selectivity, but ultimately the natural peptide, without a single modification, was the molecule that was taken to the clinic [26]. Because Prialt is a peptide, and does not cross the blood-brain barrier, it must be administered intrathecally to be delivered to N-type calcium channels in the spinal cord, which limits its potential market as an analgesic drug. However, before the discovery of Prialt, the N-type calcium channel was not known to mediate pain signaling. Prialt effectively validated the N-type calcium channel as a pain target. Consequently, various pharmaceutical companies are now pursuing the development of small-molecule inhibitors of the N-type calcium channel that may penetrate the blood-brain barrier when administered orally. Notably, other venom peptides that target ion channels in the peripheral nervous system for therapeutic applications should not require intrathecal administration.

In the case of Prialt, an advantageous feature of targeting an ion channel directly was that no tolerance to the drug develops in a clinical setting [27]; in contrast, because morphine and other opiates are agonists of G-protein coupled receptors (GPCRs), the down-regulation that occurs when a GPCR is continuously exposed to agonist limits the efficacy of morphine when pain persists over an extended period of time. This feature of drugs with ion channel targets makes them more attractive for development, particularly for chronic conditions.

ShK peptide: ion-channel drug discovery beyond the nervous system

Since the discovery of ω-conotoxin MVIIA (Prialt) in the 1980s, ion channels have proven to play many important roles outside the nervous system. In the immune system, effector memory T cells (T_EM) utilize the voltage-gated potassium channel Kv1.3 in a signaling pathway that results in activation and proliferation in response to antigen presentation. Activation and proliferation of T_EM cells is a hallmark of several autoimmune diseases, such as multiple sclerosis, rheumatoid arthritis and psoriasis. Inhibitors of Kv1.3 have a demonstrable antiproliferative effect on T_EM cells, while sparing naïve and central memory T cells (T_CM) [32,33].

Much effort has gone into discovering and/or engineering highly potent and selective inhibitors of Kv1.3 as potential therapeutic drugs for autoimmune disease. A highly potent peptide inhibitor of Kv1.3, known as ShK (Fig. 1), was discovered from the Caribbean sea anemone Stichodactyla helianthus [34] [35]. ShK inhibits proliferation of T_EM cells via block of Kv1.3 with a K_d of 10 pM [36]. Although ShK is selective for Kv1 channels, it also has high affinity (K_d, 28 pM) for the neuronal Kv1.1 channel [37], which could be a barrier to clinical development of the natural peptide, although no side effects were observed with ShK treatment in an experimental model of multiple sclerosis [38]. Over the past decade, George Chandy and co-workers have generated 380 analogs of ShK to improve targeting specificity and in vivo stability the peptide. This effort has resulted in analogs with greater than 100-fold selectivity over Kv1.1 with improved stability [39]. An analog of ShK is expected to enter human clinical trials this year [40]. While small molecule inhibitors of Kv1.3 are attractive for the treatment of MS, they are presently less potent and less selective than the best ShK analogs [33].

Venomous vs. non-venomous animals as potential sources of ion-channel targeted drugs

Natural products comprise a rich resource for the continuing discovery of pharmacologically-active compounds that target ion channels. As illustrated above, the major group of natural products that dominate the pharmacology of ion channels is peptides/polypeptides. The discovery process has been skewed towards peptides from venoms. Non-peptidic natural products targeted to ion channels remain mostly unexplored, but these should become much more accessible in the coming decades.

The emerging field of chemical ecology (the study of chemicals mediating interactions between living organisms) should eventually elucidate many novel biologically active compounds. Interactions between animals are only beginning to be elucidated at a mechanistic and molecular level. Notably, it is not just venomous animals that produce chemicals to affect the behavior of their potential predators, prey or competitors. In the marine environment in particular, there is persuasive evidence for animals producing diverse compounds to affect the behavior of the other animals around them [22,41,42]. The critical difference is that venomous animals produce the compounds of interest in a well-defined anatomical structure. For most non-venomous animals, what is not established is where and when and how such compounds are produced.

Thus, we believe it likely that in the marine environment, pharmacologically-active compounds that have evolved to modulate the behavior of another animal are targeted to ion channels, since these are the obvious molecular targets that would strongly affect behavior. Such compounds are mostly unexplored. The key problem with pharmacological discovery from non-venomous animal sources is that because the basic biology remains mostly unknown, the specialized anatomical structure(s) where the interesting chemicals are likely to be produced have generally not been directly investigated. However, compounds that have evolved to affect animal behavior would seem prime candidates as ligands for specific ion channels, as will be discussed in the next section.

This raises yet another fundamental scientific issue: if an animal is not venomous, then the compounds produced to modulate behavior of a different animal must be able to reach the physiological target molecules without the specialized delivery systems that venomous animals have evolved. In many ways, the problem is similar to the problem of getting through the gut barriers or the blood/brain barrier encountered in human pharmacology. What makes these issues critical for the next decades is that it may well be that there is a world of “drugable compounds” already produced by these animals, but because of historical biases that are a consequence of how discovery in the pharmaceutical industry has evolved, these have hardly been accessed. At present however, with rapid advances in molecular biology and biotechnology, if these compounds are being produced by any animal, they should, in principle, become accessible.

Biotic interactions and ion-channel-targeted natural products

One reason for optimism in the future of natural products as drug leads are the early indications that a vast trove of yet undiscovered compounds exist that affect the function of ion channels. In the few systems studied, the underlying biochemistry and pharmacology of biotic interactions has proven to be much more sophisticated than could ever have been predicted a priori. We discuss one specific case from cone snails to illustrate this. We suggest that this specific predator/prey interaction will prove typical of the multi-layered mechanisms that underlie an interaction between organisms that has arisen through natural selection. Another way of viewing the overall message of the biology discussed next is that for interactions between organisms to be effective in the real world, a sophisticated molecular strategy is required. Evolution is apparently a stringent taskmaster: for one animal to change the behavior of another using chemical compounds demands that natural products with novel and surprising pharmacology be generated.

The specific example of a biotic interaction that is reasonably well-understood at a biochemical and pharmacological level is between a fish-hunting cone snail predator and its fish prey. All cone snails are venomous and predatory, and a significant number (∼100 species) hunt fish as their primary, if not exclusive prey. While there is molecular phylogenetic evidence for ∼6 distinct clades of fish-hunting cone snails [43], two of these are particularly well understood, both with respect to how they capture their prey, as well as with regard to their evolutionary history. Although the two clades, comprising ∼25 different species, evolved fish-hunting independently, their strategy for capturing fish is convergent at a general physiological level.

These two clades of fish-hunting cone snails are usually regarded as subgenera of Conus: Chelyconus, comprising fish-hunting species in the new world, and Pionoconus, a clade of fish-hunting species dispersed throughout the IndoPacific, from Hawaii and Easter Island, to the East African coast. The biogeography and recent molecular phylogenetic evidence suggests that the evolution of piscivory in Chelyconus and Pionoconus were significantly separated both in time (by several million years) and space (new world vs. IndoPacific) [21,44,45,46].

Both groups of fish-hunting Conus species evolved a harpoon-shaped radular tooth that serves both as a hypodermic needle for venom delivery as well as a harpoon for tethering the envenomated fish. The fish prey, depending on the site of injection and the amount of venom that the snail is able to deliver, struggles briefly, but invariably quickly becomes tetanically paralyzed with very stiff fins, and is reeled in towards the false mouth of the snail (in which pre-digestion takes place). After an initial period of tetanic immobilization, the fish sometimes recovers and will struggle to free itself, but in most cases, the second stage, an irreversible neuromuscular block, occurs [29,47,48,49]. As a result, even when the fish is able to free itself loose from the harpoon because it recovered quickly from the original tetanic immobilization, it invariably becomes paralyzed and characteristically cone snail species in both clades will do a quick search of the neighborhood to find the immobilized fish. Thus, a fish able to break free in an aquarium setting is still devoured.

In all species in these two clades of fish-hunting cone snails, fish envenomation evokes two distinct physiological endpoints. First is the very rapid tetanic immobilization (“excitotoxic shock”); the effect is equivalent to being tasered. Second is a complete block of neuromuscular transmission (“neuromuscular block”). In both clades of fish-hunting cone snails, multiple venom components cause each of these effects (See Table 1). We refer to groups of toxins that act together for a common physiological endpoint as “cabals” [29,47,48,49]. The venom components that cause the very rapid onset of a tetanic immobilization are called the “lightning-strike cabal”. Those venom components that cause a complete block of the neuromuscular transmission are known as the “motor cabal”.

What is striking is that the two clades of fish-hunting cone snails have independently evolved both a lightning-strike and a motor cabal. At a gross pharmacological level, the molecular targets of the two cabals overlap, and in many cases are identical. However, what provides clear molecular evidence for convergent evolution is that the gene superfamilies that have been recruited to provide these analogous pharmacological functions are often divergent between the two clades, but they are homologs for all of the species within each clade. Furthermore, in both cases, a large number of venom components comprise each cabal — e.g., for both sets of fish-hunting cone snails, a combination drug strategy has evolved. The critical result is that the targets of all of the major venom components that comprise these cabals are ion channels.

Future perspective

We presented an overview of the present state of the field of ion channel pharmacology in the first section of this article. A rationale is given for why natural product discovery is likely to be a more promising approach than combinatorial chemical synthesis combined with high-throughput screening, presently the norm in the pharmaceutical industry. We also presented our view that the traditional approach used for natural product discovery, “bioprospecting,” has likely not detected the great diversity of natural products that target ion channels. Two specific examples of peptides from venoms (cone snail and sea anemone) illustrated that natural products targeted to ion channels do indeed have considerable biomedical potential. We have focused on venoms from the marine environment to help us make the case for the general discovery paradigm that we are proposing. However, it is noteworthy that there is enormous untapped potential for pharmacological discovery both from venomous- (cone snail, sea anemone, scorpion, spider, snake, etc.) [18] and non-venomous animals, encompassing both peptides and non-peptidic organic compounds, across marine and terrestrial habitats.

This overview provided the rationale for why a broad natural product discovery initiative should greatly accelerate progress in ion channel pharmacology. What would seem to be a required basic-science component of this initiative would be better understanding of biotic interactions between animals, particularly when animals use a chemical strategy to change the behavior of their prey, predators or competitors.

Venomous and non-venomous interactions that affect behavior are likely to be especially prominent in the marine environment, and the case is made that the mechanistic basis of each individual biotic interaction will involve natural products targeted to ion channels with great selectivity. This will require a more incisive knowledge of the biology of the animals producing such compounds, and of the chemical ecology of the marine environment. This should provide the raw material for drug leads targeted to ion channels in the coming decade. However, it will require a paradigm shift, abandoning the predominant strategy for natural product discovery at the present time and instead adapting a new interdisciplinary approach to discovery that incorporates phylogenetics and organismal biology with chemical ecology and medicinal chemistry.

Executive Summary

Natural-product discovery is critical for rapid progress in ion channel pharmacology.

• Low R&D productivity is widely cited as the biggest problem facing the pharmaceutical industry. The R&D cost per new drug has increased to billions of dollars, accompanied by high pre-clinical and clinical attrition rates of candidate compounds, which in many cases can be attributed to lack of targeting selectivity.

• Natural products, which have been refined by evolutionary forces for potency and selectivity, may have distinct advantages in targeting selectivity over random small molecules.

• The bioprospecting approach to natural-products discovery was misguided and has probably failed to identify many of the most potent and selective compounds that mediate biotic interactions.

Proof-of-principle for natural products targeted to ion channels as drug leads.

• Two specific examples illustrate that natural products targeted to ion channels are suitable for drug development: Prialt is an FDA-approved drug that is chemically equivalent to ω-conotoxin MVIIA (peptide). An analog of ShK, a sea anemone peptide, is expected to enter clinical trials this year for multiple sclerosis and other autoimmune diseases.

Chemical ecology: a cornucopia for the ion channel pharmacology of the future.

• The discovery of natural products that target ion channels has been skewed toward venoms in the past.

• Chemical ecology should elucidate many novel compounds that mediate biotic interactions between prey, predators and competitors, especially in the marine environment. Such natural products should have built-in drug-like properties and many will likely target ion channels.

• One example that illustrates the potential diversity of natural products targeted to ion channels from the marine environment is from fish-hunting cone snails. Many distinct venom components act in concert for prey capture or defense in the “lightning strike cabal” and “motor cabal.”

Why natural products will prove to be more diverse and selective than ligands for ion channels developed in the past.

• The fact that biodiversity leads to chemical and pharmacological complexity is illustrated by the divergence of venom components between different clades of fish-hunting cone snails.

• Venom components that act in concert for prey capture or defense have evolved to be highly selective, apparently because 1) this allows a compound to reach its target faster and 2) without interference upon other targets or circuitry perturbed by other venom components.

• We argue that the “cabal” strategy (multiple compounds acting in concert to mediate a biotic interaction) will be much more general and widespread in biology than is currently anticipated, leading to the discovery of many novel, highly selective, ion-channel targeted compounds.