June 1, 2008 (Vol. 28, No. 11)

New Methodologies Seek to Find, Map, and Ultimately Disrupt These Processes

Most proteins including mutated, disease-causing ones do their work as part of multiprotein complexes. Thus, sites of protein-protein interactions would seem to be optimal targets for drug discovery. Of course, this is not as simple as it sounds; researchers have been reluctant to target protein complexes as the average protein-protein interface is quite large (approximately 1,100 Å2) and relatively flat.

That said, a number of technologies have been developed to find, map, and ultimately disrupt protein-protein interactions, and many are beginning to bear fruit. Some examples that were presented At CHI’s recent conference on “Protein-Protein Interactions as Drug Targets” are discussed more fully in this article.

Donny Strosberg, Ph.D., professor of the department of infectology of the Scripps Research Institute (www.scripps.edu), studies protein-protein interactions in the context of host pathogen relations. His lab is currently focused on the hepatitis C virus (HCV), an RNA virus.

The HCV core protein homodimerizes to form the structural element of the viral nucleocapsid. Inhibiting this dimerization should thus block formation of the infectious particle. Moreover, the HCV core is highly conserved across all genotypes; this makes it an attractive drug target.

Dr. Strosberg presented his laboratory’s identification of a 15-residue, core-derived peptide that inhibits dimerization of the N-terminal region of the core. Dr. Strosberg said that this is the first time the core dimerization has been inhibited. This peptide also blocks infectious HCV release from host hepatoma cells.

The peptide was identified using two approaches: an amplified luminescent proximity homogenous assay (ALPHA screening) and a homogenous time-resolved fluorescence assay (HTRF). ALPHA screening is based on donor and acceptor beads with differently tagged proteins; if the proteins interact, they bring the beads together to generate an intensely amplified signal.

HTRF depends on a fluorescence-resonance energy transfer between antitag antibodies labeled with Europium cryptate and XL-665 (allophycocyanin). In the case of HCV, Dr. Strosberg noted, the ALPHA screen had “much better sensitivity than HTRF.” Another advantage is that “the beads are always reusable.” However, he was quick to point out that interactions are “very different between one pair of proteins and another” and that, therefore, “several techniques each complement the other.”


At Prolexys Pharmaceuticals (www.prolexys.com), the emphasis is on mapping protein-protein interactions to identify and exploit novel targets for drug discovery in cancer and cardiovascular disease. The nonprofit Huntington’s Foundation immediately appreciated the utility of Prolexys’ human interactome database and therefore gave a grant to Sudhir Sahasrabudhe, Ph.D., CSO of the firm, to identify therapeutic agents in Huntington’s disease and other neurodegenerative diseases.

“The nonprofits are getting savvy,” Dr. Sahasrabudhe noted, in that they recognized the value of the extensive, accurate database of human protein-protein interactions generated at Prolexys. The dataset was built using the company’s HyNet system, a highly automated, high-throughput process based on random yeast 2-hybrid searches.

Huntington’s disease is a fatal neurodegenerative disease that is inherited in an autosomal dominant fashion. The wild-type huntingtin protein (Htt) has, so far, an undetermined role in development. In Huntington’s disease, there is an expansion of the glutamine-encoding GAG tract of the Htt1 gene. This expanded span of glutamines obviously changes the shape of the protein, which disrupts its normal interactions and can cause it to interact with inappropriate partners. The mutant protein eventually causes cell death.

The huntingtin protein interactome was compiled using HyNet as well as HySpec, which consists of affinity pull downs followed by mass spectroscopy. The hypothesis was that genetic modifiers of neurodegeneration would be enriched among proteins that interact with mutant Htt.

Interestingly, there was little overlap of interacting proteins between the datasets generated by the two methods. HyNet identified more proteins involved in protein turnover, signal transduction, and transcription, whereas HySpec primarily identified proteins involved in metabolic processes. Dr. Sahasrabudhe also performed a yeast-based, high-throughput screen to identify small molecules that prevent huntingtin protein aggregation.


David Mann, Ph.D., lead senior scientist at Infinity Pharmaceuticals (www.infi.com), discussed his work on the identification of a highly potent and selective Bcl-2 antagonist at the same meeting. In collaboration with Novartis, he used diverse oriented synthesis (DOS) to generate a relatively small library of complex compounds.

Dr. Mann explained that when chemical libraries were introduced roughly 20 years ago, they were “targeted to inhibit kinases” and were, therefore, often large but chemically not very interesting. DOS is “different because at the core, structures are more complex and natural product inspired. Structures are more rigid, and there is more stereochemistry.”

Libraries are thus composed of 10,000 rather than 100,000 molecules, but they are of a set—they are similar at the core. The idea was that there would be “more unique and specific hits from a high-throughput screen and a better lead to start” designing drugs.

This was certainly the case for Bcl-2. the founding member of a prosurvival subfamily of proteins, and one of the most common targets for PPI inhibitors. It was discovered in B-cell lymphoma, where its overexpression allows cancerous transformation by preventing normal apoptosis. Bcl-2 and proteins like it have since been found to be overexpressed in many human cancers. Its antiapoptotic function is regulated by heterodimerization with other family members.

ABT-737 is one of the most successful Bcl-2 inhibitors identified to date. It is a mimetic of BAD, a native Bcl-2 binder. ABT-737 binds to Bcl-2 with a Ki in the 10 to 100 µM range. Using DOS, Dr. Mann’s group got initial hits with Kis

These compounds are also highly specific to Bcl-2; unlike ABT-737, they do not even bind to the structurally similar protein Bcl-x. Interestingly, they bind in the same pocket of Bcl-2 as ABT-737 does but in the opposite direction.

Infinity is no longer using the DOS platform to screen for new drug targets, but scientists at Novartis are continuing to work on a number of libraries.


Scientists at Polyphor (www.polyphor.com) spent the last decade developing protein epitope mimetics (PEM) technology to discover and optimize therapeutically valuable inhibitors of protein-protein interactions. At the conference, Daniel Obrecht, Ph.D., discussed how the company’s method generated a potent CXCR4 inhibitor.

PEM molecules aim to act as small molecules. They are fully synthetic, peptide-like molecules that mimic b-hairpins because b-hairpins often contain residues essential for protein-protein interactions. Once identified, a chosen b-hairpin is stabilized on a proprietary template. Then, focused libraries are designed around it using a multiparallel synthesis technology, and promising mimetics are optimized to the desired biological and pharmacological properties using the required design variables.

Michael Altorfer, Ph.D., CFO at Polyphor, noted the many advantages of its approach. First, he said, the “constant methodology leads to rapid improvement, eliminates repeated process development, and reduces the lead-throughput time and necessary resources. The high degree of automation increases throughput, reduces throughput time, and facilitates scalability.

“The versatile methodology permits rapid optimization of ADMET properties and selectivity, the main reasons for small molecule attrition.” Moreover, this technology is “applicable to a wide range of targets and therapeutic applications” and has the potential to “yield a new class of drugs combining the advantages of both small molecules and biopharmaceuticals.”

PEM molecules have been particularly effective at targeting GPCRs. Few small molecule antagonists of these receptors are known, perhaps because of their large ligand-binding domains. CXCR4 is a coreceptor, along with CD4, for the entry of T-cell line-tropic (X4) HIV-1 into T cells.

One of the first CXCR4 antagonists to be described is polyphemusin II, from the American horseshoe crab (Limulus polyphemus). Using Polyphor’s parallel combinatorial synthesis, Dr. Obrecht’s group developed a mimetic based on polyphemusin II but with significantly improved plasma stability, selectivity, and pharmacokinetic properties.

This CXCR4 inhibitor is currently in Phase I trials, and Dr. Altorfer is hopeful that “if these results are positive, other applications such as multiple dose treatment, e.g., neutropenia and leukemia, can be envisaged.”

These different technologies all take various tracks toward achieving similar ends—interrupting protein-protein interactions. The ends, though, are not really similar; each interaction is unique and thus might require a unique structure or mechanism to disrupt it. All of the existing technologies and maybe more are therefore necessary, and hopefully all of them will be successful.

Previous articleInvestigators Uncover Role of Two Tumor Suppressor Genes in Aging
Next articleApplied Biosystems and BioTrove to Offer Synergistic Mass Spec Technologies