With so many drug targets in the body being activated by protein hormones, it makes sense that shorter chain peptides might be good starting points for new drugs. But peptides rarely have the optimal properties required to make a successful drug.
At the recent TIDES conference in Boston, Zealand Pharma’s Head of Pharmaceutical Development, Torsten Malmström, asked what it takes to turn a peptide into a drug, and said that often – though not always – a good starting point can be a native peptide. To make a good drug, it will need to have a good half life, stability, aggregation properties, receptor specificity and in vivo efficacy. If it does not already have these, they will have to be designed in.
As well as efficacy at the desired receptor or receptors, other properties are desirable. Is it suitable for daily or weekly (or longer) dosing regimens? Can it be – or should it be – combined with other drugs? Is it suitable for liquid formulation? Is the side-effect profile acceptable? And, of course, does it have efficacy in a relevant disease model?
Dosing frequency, for example, can be altered through careful formulation or device selection, or by changing the peptide in some way. The peptide backbone can be modified; for example, it might be acylated or PEGylated, or the conformation could be fixed, perhaps via stapling. However, care must be taken as such modifications can have an impact on in vitro potency.
‘Once we have defined the starting point, we look to optimise physical and chemical stability, while maintaining efficacy,’ Malmström explained. ‘We will typically go through three, four or even five rounds of optimisation.’ Challenges they frequently face include ensuring that the peptides do not aggregate.
He also advises that it can be a good idea to define the formulation strategy early on in the development process. Attributes that need to be established include the route of administration, the concentration range, the form, such as solution or emulsion, what the diluent should be, whether it is given by injection or infusion. And should the product itself be liquid or lyophilised? And what about storage conditions?
Druggable target space remains small for small molecules and monoclonal antibodies, but cell-penetrating peptides can bridge this gap
According to Francesca Milletti, Head of Chemoinformatics and Statistics at Roche, there has been a real increase in interest in peptides at Big Pharma companies recently. One area of interest at Roche is cell-penetrating peptides, or CPPs. These typically comprise five to 30 amino acid residues, and are generally highly cationic, amphipathic or hydrophobic. ‘Druggable target space remains small for small molecules and monoclonal antibodies, but cell-penetrating peptides can bridge this gap,’ Milletti said.
But there are many challenges that must be overcome if therapeutic success is to be achieved. These include instant cell internalisation, lack of cytosolic delivery, insufficient tissue penetration, proteolytic degradation and poor cell selectivity.
Roche set up a screening platform for cell-penetrating peptides. This uses peptide libraries developed using mRNA or phage display, and includes peptides with a range of physico-chemical properties such as hydrophobicity, charge and polarity. They also looked at what already exists – many cell-penetrating peptides were discovered from related groups of proteins.
One project the company has been involved in is to target IRF5 in inflammation. Dimerisation of IRF5 leads to the activation of promoters of inflammation. The formation of the dimer can be disrupted by inhibition of the interaction between two of the helices in the IRF5 structure. Non-critical residues were mutated to obtain a cell-penetrating peptide, which proved to be active in vitro.
Other examples are eIF4E and eIF4G, which are upregulated in many cancers. Roche found a segment of the protein that is critical for binding, and this was inserted into various cell penetrating peptides, and found to be active. The lead peptides also showed surprisingly good in vitro plasma stability.
Milletti explained that CPPs have primarily been used as research tools thus far, and translation to therapies will require optimal conjugation or grafting to functional peptides, an increase in proteolytic stability, and a reduction in sensitivity to diverse cargoes. However, they could be incorporated into other peptide-based techniques.
Lucinactant (Surfaxin), the lung surfactant product, revolutionised the survival potential of premature babies
Russell Clayton, Senior Vice-President for R&D at Discovery Labs, gave the story of lucinactant (Surfaxin), the lung surfactant product that revolutionised the survival potential of premature babies. ‘The story begins 50 years ago, when President Kennedy’s son Patrick was born at 34 weeks,’ he said. ‘He had a surfactant deficiency causing respiratory distress syndrome, and died just two days later.’ This raised awareness of the condition, and was the catalyst for research into treatments.
Surfactant does not interact with receptors in the lungs – it merely maintains the surface tension between the lungs and the air. Without it, the alveoli will collapse, and much more energy is required to reinflate them. So premature babies who do not have the necessary surfactant frequently develop respiratory distress. Mortality levels dropped following the introduction of the lung surfactant Exosurf in 1990 – before this, a baby born at 29 weeks had only a 50% chance of still being alive the next day. However, this synthetic product did not contain the protein components that make up about a tenth of natural surfactant and are vital for the full effects to be achieved. Exosurf was followed by several animal-derived surfactant products that, while they were both safe and more effective, led to concerns about the exposure of newborn babies to medicinal products from animal sources.
These concerns led to a project to develop synthetic proteins. Surfactant is a complex substance, containing lipids, phospholipids and a handful of proteins, the apolipoproteins SP-A, SP-B, SP-C and SP-D. Of these, Clayton says, SP-B is the most important. A synthetic analogue of SP-B called KL4 was developed by Charles Cochrane at the Scripps Institute in 1991, thinking that the repetition in the amino acid sequence might allow something more simple to be used instead. And, sure enough, KL4 is a 21-amino acid peptide consisting simply of lycine and leucine residues, which mimics the properties of SP-B.
Lucinactant contains this peptide, plus phosphatidylcholine, phosphoglycerol and palmitic acid components. It lowers surface tension at the surface of the lungs, as a result improving lung compliance and gas exchange. It is also bactericidal and modulates inflammatory responses. There were a number of issues with clinical trials – not least that the regulator wanted trials to use Exosurf as a comparator rather than one of the animal-derived products, despite the fact that Exosurf was no longer available in most of the developed world. Ultimately, however, the product gained FDA approval in March 2012, nearly a decade after the trials were completed.
Desirable characteristics
Ignace Lasters, Chief Scientific Officer of Belgian biotech Complix, described his company’s work on Alphabodies, the name it gives to rationally designed small protein scaffolds, with molecular weights in the range 10–12kDa, that use natural building blocks and have desirable characteristics for therapeutic molecules. These can arrange themselves in two different ways: as a single chain or an alpha-helical ‘peptidic’ form.
In its single chain form, the scaffold has a triple-strand, alpha-helical coiled structure, with the three helices arranged in an anti-parallel formation, and connected by short and flexible linker segments. The scaffold is both temperature stable and resistant to proteases, largely as a result of the tight way the isoleucine residues at its layered core are packed. No disulfide bridges are needed to confer stability, but free cysteine residues can be introduced to facilitate derivatisation such as PEGylation. This single-chain format is amenable to genetic engineering, and can be made via recombinant protein expression in common cell substrates, with wide pH stability and also stability to proteases. There is tremendous freedom for sequence variation, Lasters said, while preserving the scaffold’s structural identity.
More than two-thirds of the alphabody’s residues can be altered. Their structure means they can bind to a number of targets to which normal antibody proteins will not, and their compact nature allows more than one antigen binding site to be present on the surface. This enables multiple targets to be bound, which can be difficult for antibody proteins.
Scorpion venom is known to contain ion channel blockers
Lasters believes there are more than 1,500 currently undruggable targets where these structures might have potential. For example, interleukin-23 is a key target in inflammation. The company’s furthest advanced candidate against IL-23 was discovered in 2011, and within a year it had entered preclinical development. It has both potency and selectivity at an undisclosed autoimmune disease target.
It is also possible to convert a single-chain Alphabody structure into a peptidic form, preserving most of its binding capabilities. These peptidic forms are coiled assemblies of three alpha-helices stabilised by non-covalent intermolecular interactions, although they can also be covalently linked. This technique was illustrated via the peptidisation of the IL-23 neutralising Alphabody. It has sub-nanomolar potency in its single-chain form, where the binding site comprises two ‘half sites’ on adjacent alpha-helices with different sequences. These two half-sites were grafted into a single alpha-helical sequence, giving the peptidised homotrimeric mimic, in theory, three identical IL-23 binding sites. The original single-chain structure had 101 amino acids, but the mimicking peptide has just 29 amino acids, yet it still binds strongly in vitro to IL-23, with low-nanomolar binding affinity.
There are at least 80 types of autoimmune disease, which affect an estimated 125 million people around the world
Michael Pennington, President and Chief Operating Officer of Peptides International, based in Louisville, KY, US, described progress advancing ShK peptide into the clinic as a potential treatment for multiple sclerosis and other autoimmune disorders. There are at least 80 types of autoimmune disease, he said, which affect an estimated 125 million people around the world, from multiple sclerosis and rheumatoid arthritis, to Type I diabetes, lupus, inflammatory bowel diseases and psoriasis.
Potassium channels in human T cells offer potential targets to influence autoimmune diseases, notably voltage gated Kv 1.3 and calcium-activated KCa3.1 potassium channels. Both of these have homo-tetrameric functional channels, and if they are activated, autoimmune disease can result. In multiple sclerosis, for example, the T-cell activation cascade is a key part of the disease-causing demyelination process. Disease-relevant autoreactive T cells express the Kv 1.3 high pattern in both MS and Type I diabetes.
A potential lead to block these potassium channels arose from a report that a woman with MS who was stung by a scorpion, experienced an improvement in her symptoms the next day, and became symptom-free about two weeks later. This improvement lasted for a couple of months. Scorpion venom is known to contain ion channel blockers; another source of many potassium channel blocking peptides is the sea anemone. Stichodactyla helianthus toxin, or ShK toxin, is one of these, but it has problems – it is not particularly selective for Kv 1.3.
A programme to synthesise analogues of ShK that might overcome these selectivity issues was started. More than 400 analogues were synthesised and evaluated, and ultimately the peptide ShK-186 was identified as a highly selective Kv 1.3 inhibitor. It has picomolar activity, and has no effect at all in most ion channels, Pennington said.
Mouse models cannot be used to evaluate Kv 1.3 blockers in immune function, as it does not set the membrane potential in mouse T cells; dog, rat, pig or primate models have to be used instead. Therefore it was tested in a rat model, where the rats had relapsing-remitting experimental autoimmune encephalomyelitis. Treated rats had significant improvement in their disease state over those given a saline placebo. Success in this model, and an additional primate model, led to a Phase I trial being started in 2012, run by licensee Kineta.
A source of many potassium channel blocking peptides is the toxin of the sea anemone Stichodactyla helianthus
Interest in macrocycles
There has been plenty of interest in recent years in macrocycles as potential drugs acting at ‘difficult’ targets. They have the potential to work at extended binding sites, where the target is large with multiple interactions, which are not easily addressed by normal small molecule drugs. Yet about 80% of disease targets have this type of extended binding region, Terrett said, and with very many small and compact binding sites already addressed with small molecules, they represent a significant opportunity for drug discovery.
The molecules that Ensemble is working on, termed Ensemblins, fall neatly into the space between small molecules and biological drugs, with better cell permeability and oral bioavailability than biologics, but better affinity for extended binding domains than small molecules. Nature has created very many macrocyclic molecules with a huge range of biological activities, but they did not evolve to treat cancer and inflammation in humans, so they rarely have optimal drug-like properties. These properties can be improved by chemical modification, but this is generally very challenging from a synthetic standpoint.
When creating new macrocycles, it is important to control lipophilicity and manage intramolecular hydrogen bonds if membrane permeability and oral activity are to be achieved. ‘H-bonding can be controlled by judicious N-methylation,’ Terrett said. ‘Can we use this insight to design synthetic macrocycles with drug-like properties?’ As an example, cyclosporin is an orally available cyclic peptide, with 11 amino acids, and seven methylated amides. This N-methylation is frequently significant in nature as it restricts the conformation and removes the hydrogen bonding capability of the amide bond.
An oral drug would be preferable to an injectable antibody, and a macrocyclic peptide-based drug might provide this
Interleukin-17 is an anti-inflammatory target, and several antibody drugs are in development against this target as potential treatments for psoriasis and other inflammatory conditions. The X-ray crystal structure of the dimeric IL17:IL17 receptor complex shows a large protein–protein interaction interface, which has already been shown to be intractable to small molecule discovery efforts. An oral drug would be preferable to an injectable antibody, and a macrocyclic peptide-based drug might provide this.
Ensemble’s starting point was a library of peptide-derived macrocycles made by DNA-encoded chemistry. A mixture of 220,000 compounds is created in a single reaction vessel, with each compound encoded by a unique DNA sequence. This allows the behaviour of each individual molecule within the mixture to be tracked. The library of 3.3 million compounds, in mixtures, was screened for activity, and hits followed up by the company’s medicinal chemistry team. Discrete compounds were made for in vitro and in vivo evaluation, and specific binding to IL17 confirmed for several of these macrocycles, now without the DNA tag. A traditional medicinal chemistry programme was used to improve the properties, with more than 500 discrete macrocycles made in 16 months. By the third generation, the macrocyclic scaffold had a 1000-fold increase in affinity.
The macrocycles potently inhibit IL-17A binding to its receptor, with antibody-like binding kinetics, exhibiting moderate on-rates and very slow off-rates. The lead compounds potently inhibit IL-17A induced cytokine expression in cells, with good selectivity for IL17 over other cytokines. The company has now entered into a development agreement for this programme with Novartis.
Much of this feature is based on the TIDES conference, held in Boston, MA, US in May 2013.