Sneaking a peak at evolution's recipe book

In the last year, Perth-based Phylogica has announced a number of milestones in its plan to revolutionise the world of novel therapeutics and knock antibodies off their perch as the favoured route for pharmaceutical companies targeting certain diseases.

One of the biggest milestones was the signing of a research collaboration and licence agreement with Johnson & Johnson Research, the Australian arm of one of the biggest pharmas in the playground. While Phylogica is not able to disclose the disease area the collaboration is targeting, the company's CEO, Dr Stewart Washer, says it is hugely important validation of the potential of the technology.

That technology involves a range of peptides called phylomers, which are able to bind to target proteins and block their interactions. The company sees huge potential in developing alternatives to antibody-based therapeutics, a multi-billion dollar sector.

The J&J deal came hot on the heels of another international partnership, this time with Irish company Opsona Therapeutics, which itself is involved with another Big Pharma player, Wyeth Pharmaceuticals.

Phylogica's partnership with Opsona is in the booming field of toll-like receptors (TLRs) and T-regulatory cells, linked to a number of inflammatory diseases. This is a typical agreement structure in which Phylogica will out-license its peptide compounds to Opsona and receive customary milestone payments and royalties on any sales.

Closer to home, Phylogica and two of its other collaborators, the University of Melbourne and Murdoch University, recently announced the discovery of drug candidates that block key pathways in rheumatoid arthritis, a field currently dominated by antibody-based therapies.

"We have found a number of phylomer binders against a panel of validated targets in rheumatoid arthritis, including the classical target tumour necrosis factor (TNF)," Washer says. "When you consider that just one of the current antibody drugs to target TNF (J&J's Remicade) has sales of approximately $2.5 billion last year, we're very excited by the potential of pursuing this target."

Under the collaboration, the candidates will be tested in various laboratory models by Murdoch and the Telethon Institute for Child Health Research, before beginning experiments in animal models by Melbourne.

Phylogica also has partnerships with Melbourne's Baker Heart Research Institute for cardiac disease, Sydney's Garvan Institute for diabetes and the Neuromuscular Research Institute for stroke, and the McComb Foundation in Perth for deep tissue burns.

Recent trials with the McComb Foundation, directed by burns pioneer Dr Fiona Wood, show Phylogica's new drugs are able to significantly accelerate wound healing following an acute burns wound, an important factor in reducing scarring. The company's vice president of technology development, Adjunct Associate Professor Paul Watt, says in this trial the phylomers are being added topically, straight on to the wound.

"What's attractive about this application is that if you've got a full-thickness burn, it is relatively easy to deliver a peptide because there is no epidermis to get through," Watt says.

"We are getting amazing protective effects in terms of wound healing. In a burn there is an acute inflammation response which can actually block the wound healing. So if you can settle down that inflammation response you get much better healing, and that's what we have found."

Shapes and folds

Phylogica was only established in 2001 and publicly listed in 2005, but its technology was incubated over many years at the Telethon Institute for Child Health Research in Perth, which is a major shareholder in the company. Another shareholder and Telethon's partner in the venture is the Fox Chase Cancer Centre in Philadelphia.

Paul Watt was a driving force in establishing Phylogica after many years of research, including a doctorate from Oxford in molecular biology and postdoc work at Harvard in DNA topology, before returning to Oxford for a second postdoctoral fellowship in genome stability and cancer research.

This was where he was first introduced to the power of yeast genetics and the tools it offers for validating targets and understanding the functions of human proteins.

"I was in cancer research and the problem was there were no tools to validate - as there still aren't at the moment - potential targets, especially intracellular targets, at the protein level.

"I wanted to find a tool that would enable me to surgically disrupt particular protein interactions without disrupting others."

In traditional drug discovery the only tool available to survey different protein structures has been antibodies, he says, but antibodies have certain drawbacks. They are unable to be used inside cells themselves because they are too large and have a tendency to fall apart as they are only joined together by disulphide bonds, which are destroyed in the reducing environment inside cells.

"What I wanted to do was find a blocking reagent to validate these cancer targets that had the structural diversity of antibodies but also get around some of the negative features of antibodies - the problems with delivery, the problems with prohibitive royalty stacks. Antibodies are owned by multiple companies so you end up paying a lot of royalties."

The final problem with antibodies is the cost of manufacture, as they have to be made in cell-based bioreactors. Phylogica's phylomer peptides, on the other hand, are chemically synthesised.

Phylomer libraries

Drug development is only half of the story behind Phylogica, however. What really lies at the heart of its technology is its libraries of millions of distinct phylomers, which are sourced from archaea, an ancient microbial phylum which split off from conventional bacteria about three billion years ago.

Phylomers are fragments of proteins derived from these microbes, which have hugely divergent protein structures. What makes phylomers so different is all about shape, Watt says.

"Proteins are really only assembled from a manageable repertoire of different folds or shapes. Originally they thought there might be as few as 1000 but now they think there might be about 3000. That's still a manageable amount.

"Our largest library contains 260 million different phylomers from 25 fully sequenced genomes. From that number we have thousands of representatives of each of the basic fold classes. That means that we have so much structural diversity that we get better hit rates than antibodies, or anything else I'm aware of, for blocking protein interactions."

According to Watt, it's not actually the size of the library that matters in drug discovery, it's what you do with it. "It's not about how many different molecules you're screening, it's about how many different structures you are screening," he says.

"If you are screening a whole lot of similar molecules with uninteresting structures then that's not going to get you hits. What matters is how much structural diversity you have in your library, not how much numerical diversity there is.

"There are companies that have larger peptide libraries than ours, but we doubt whether they have anything like the structural complexity of ours, because our libraries have been through the filter of evolution for a plethora of different structures to provide a richer source of blockers.

"And that's what we've found - we've found that when we screen our phylomer libraries for blocking protein interaction targets, for example of disease interactions involved in inflammation, we find that we get hit rates that are at least one hundred-fold and up to a thousand-fold better than the hit rates with random peptides. That's because random peptides rarely fold into any interesting structures, whereas we are taking folds that have been selected by evolution.

"If you take peptides that come from parts of proteins, then evolution has already selected for particular sequences that can adopt and fold into stable conformations that have worked biologically before. So rather than play fair, we prefer to cheat from evolution and sneak a peak at evolution's successful recipe book for how you make a peptide structure that has a biologically relevant shape."

Some of these shapes may have particularly high thermal stabilities having evolved in thermophilic bacteria which can live at temperatures exceeding 95 degrees, he says.

Protein interference for target validation

Watt sees his phylomers as an alternative not only to antibodies but as a tool to complement short interfering RNA-based (siRNA) technology as well. Watt describes siRNA as a very powerful primary survey instrument which has inherent limitations.

"Everyone imagines that siRNA is going to solve all the problems in target validation, but if you think about it siRNA is a means of taking away the message of a particular gene, and through that it takes away the protein it encodes. But there are a lot of proteins that aren't regulated at the transcriptional level.

"If you take away the message then you are not really silencing the protein, which might be very stable, so you don't get much of a phenotype."

But Watt believes that's not the only issue: there's another key question of the specificity attainable with RNAi, aside from issues of off-target silencing or unintended induction of an interferon response.

"For example, if hypothetical protein A interacts with a protein B in disease and a protein C in normal cells, if you take away protein A with siRNA then you have blocked the normal functions of A as well as its the disease-associated functions. It's not very fine-tuned and can therefore often be uninformative.

"In these cases a true 'protein-interference' technology like that using phylomers can play a role. In this case you can block the disease interaction specifically." Immunogenicity

When talk turns to blocking protein-protein interaction by introducing foreign substances into the body, the dreaded issue of immune response is always raised. Just as mouse monoclonal antibodies are foreign, so are these synthetic peptides.

Watt doesn't see this as a major issue with phylomer-based therapeutics, as most do not contain significant amounts of MHC2, the major histocompatibility epitopes that generate an immune response.

Phylogica has contracted Scottish company Accuro Biologics to model a large set of phylomer sequences for immunogenic potential and found that only a small percentage were predicted to raise a significant immune response. "In our hands, phylomers haven't turned out to be particularly immunogenic in animal models either," Watt says.

"The reason for that is not that they differ in any way from any other protein fragment, except that they are small. The vast majority of the phylomers don't contain any significant MHC2 epitopes and the simple reason is that their sequence is so small they are statistically unlikely to contain such epitopes.

"That's another advantage of being small - you can slip under the radar screen to avoid recognition by the immune system. There are foreign peptides that come from toxins from venomous creatures that are also not immunogenic, for the same reason that they are small and they don't happen to contain MHC2 epitopes.

"Just because you are foreign doesn't mean you generate an immune response. And just because you are human doesn't mean that you don't generate an immune response - some people have a response to insulin, another peptide, even to human insulin."

Further applications: diagnostics

While the potential of phylomers is most obvious in therapeutics, there may be some diagnostic applications down the track. One application area that Watt says could be very powerful due to the enhanced hit rates of these millions of structure-rich phylomers is in the field of protein microarrays.

"If you were to put 50,000 phylomers onto a protein chip, there are enough specificities that you'd expect to have multiple hits to any given target and that's quite a powerful thing. What it means is that without having to know anything about a target spectrum you can expect to find hits, so you can imagine the diagnostic applications."

He envisages a future in which chips would be available in which clinicians could look up a diagnostic profile as there would be enough phylomers binding to a set of targets to provide a signature. Using proteomics tools, designers of diagnostic kits would also be able to very rapidly survey the phylomer library for new biomarker signatures.

"You can't really do that with current antibody libraries because they are biased to a particular class of target where there are plenty of antibodies raised. There simply aren't enough specificities in the current antibody libraries to be able to survey the human proteome with any prospect of obtaining a hit to any given target.

"Whereas if you have more structural diversity in a library as you would with phylomers, you'd expect that in a reasonably modest sample of 50,000 different phylomers on a couple of protein chips, you might expect to have a hit to virtually all the protein targets that you probe such an array with.

"That's very powerful concept for multi-disease diagnostic applications, the identification of new biomarkers for disease as well as for very rapid target validation. It's an area we are only beginning to look at."