By Peter Halkjaer-Knudsen
Sr. VP. Alliances & Business Development
iNovacia

Scientists in the research department of Biovitrum in Stockholm, Sweden, took a novel approach to forming a spinoff company late in 2005, effective in May of 2006. They saw an opportunity to initiate not a management buyout, but rather a scientist buyout, whereby all scientists became part owners of a company whose roots go back to the discovery department at Pharmacia. iNovacia became a fully independent company, also based in Stockholm, with a name inspired by combining “innovation” with “Pharmacia,” where it all started.

Today, iNovacia is a clinical research organization involved in early drug discovery, from assay development/HTS to late-stage lead optimization. We have invested in technologies that give early and structured insights into targeted biological/biochemical processes, by looking at both the chemistry and biophysics of the molecules that participate in these processes. These technologies consist of conventional biochemistry and molecular pharmacology combined with general biophysical tools that may be applied in all steps from assay development to lead optimization.

The importance of understanding the vast number of biological processes affected by novel pharmaceutical interventions has increased dramatically during the last decade; these processes are important both from discovery and safety/regulatory standpoints. In addition, significant increases in costs along the development chain of a candidate drug have spurred the development of more effective ways to select, as early as possible, candidates with the best potential of surviving all the way to market.

For many target proteins, the chemical and biophysical processes that affect “survivability” may be mapped for various key events by applying biophysical tools. In our company, we focus on ways of optimally applying different biophysical tools in the drug-discovery process, based on experience gained over the years within the mother companies, Pharmacia and Biovitrum.

Fine Tuning Assays
In assay development, it is critical to know how the protein behaves under assay conditions, and also to collect data relevant to physiological conditions. Many target proteins have great structural complexity—that is, they have several functional domains or are multimeric. Such proteins may behave very differently depending on their fold or quaternary structure (monomeric, dimeric, multimeric etc). Other proteins are difficult to formulate in a suitable assay format due to protein aggregation problems, or the protein’s activity may be hampered due to strongly bound fortuitous ligands emanating from the recombinant protein production process.

Biophysical tools are helpful in addressing these problems. For example, dynamic light scattering (DLS) in a 384-plate format has been successfully applied in buffer optimization studies to identify conditions most likely to prevent target aggregation or stability problems¹. Non-denaturating electrospray ionization mass spectrometry (ESI-MS) reveals details of quaternary structures and detects any fortuitous ligands bound to the target. Protein NMR, analytical ultracentrifugation, and microcalorimety can be applied when a more in-depth understanding of the folding state and binding characteristics of the target protein is required.

Pinpointing Libraries
Exposing the target protein to a chemical library that has the potential to yield “druggable” hits is as important as using a qualified assay in an HTS campaign. The library needs to be large enough to cover a relevant volume of the druggable chemistry space, and the space may vary from target to target. Therefore, we have invested in predefined subsets of the screening library, focusing on specific protein families. In many cases, public or proprietary information is available on the structure-activity relationships of a specific target. In these cases, iNovacia applies logistics that enable us to cherry pick custom screening sets from a collection of approximately 300,000 compounds. The selected compounds must be designed structurally to be recognized by the target protein, and they also must meet stringent criteria for hit- and lead-likeness. In 2006, all compounds were made from powder, giving a very fresh start to the discovery effort. We believe it is very unusual to have a library where more than 90% of the compounds are more than 90% pure!

In addition, we have a compound library containing small and highly soluble compounds specially designed for fragment-based screening, an approach in which our group has a good deal of experience, primarily using our 600 MHz Hi-res NMR². The design of the fragment library is intended to generate hit/lead-like starting points with high-ligand efficiencies, amenable for rapid expansion for quickly exploring the SAR, and building solid IP (intellectual property).

Qualifying Hits
As the list of hits evolves from the HTS, dose response curves are generated and the hits are qualified for identity, purity and stability. However, as with the challenges in assay development, there are many potentially expensive pitfalls that can be avoided. The IC50/EC50 curves may look interesting in terms of potency, but the mechanism of interaction between the hit and the target may not be desirable. Filtering out such dead-end hit series immediately after HTS will save resources otherwise engaged in trial and error, which instead can be focused on ensuring that the hit series acts through sound mechanisms, and on specifically developing the series³.

Although all current knowledge has been put into the design and integrity of the chemical library used in HTS, we must be prepared for surprises that need to be identified and acted upon, both to filter out dead ends and to provide important information for building an early SAR for prompt hit expansion. One such problem is the formation of aggregates by the hit alone or in combination with the target protein. These aggregates may inhibit the function of the target protein, something that is not revealed by inspecting the IC50/EC50 curves. A DLS analysis of the hits is an effective early flag for such problems.

Other potentially unwanted events are chemical modifications of the target protein, either as oxidation of amino acid residues, irreversible binding to the target protein, or a high degree of unspecific binding. Even if the hits look good in terms of chemical, biophysical, and ADME hit criteria, medicinal chemists still must understand the nature of the hits’ ligand/target interactions in order to understand their SARs and prioritize them. In the investigation of “good” hits and their analogues, protein NMR binding studies, as well as ESI-MS and differential scanning calorimetry (DSC), are applied both to nail down unwanted events and to cluster different chemical series that compete for the same binding site, and thus have overlapping pharmacophores. In the end, all these efforts are aimed at promptly giving the project team a set of chemical starting points for lead optimization with maximum potential for survival and freedom for expansion.

With such a small group (only 40 people), it is great to have all our expertise—classical HTS, fragment-based screening, biophysical characterization of targets, hits, and hit-target interactions⁴—gathered around the lunch table every day. It makes for extremely efficient decision making in, for example, selection of technologies to be employed. Using the tools described above with maximum efficiency, we are able to progress soluble targets from validated assay to lead in six to eight months, mainly because several of the processes can be staggered so that several teams work simultaneously on the same hit series. Insoluble targets may still take a bit longer.

All of this is possible because the biophysical characterization of the hit series is set up to display the weaknesses of each series; therefore, if one series goes through with no or few comments, then all teams can focus simultaneously on developing it into lead status, thereby avoiding most dead ends and wasted resources.

References
1. Dimethyl Sulfoxide related Effects in Protein Characterization.Tjernberg A, Markova N, Griffiths WJ, and Hallén D. J. Biomol. Screen. 2006; 11:131-137.
2. Structure-based screening as applied to human FABP4: A highly efficient alternative to HTS for hit generation.van Dongen M, Uppenberg J, Svensson S, Lundbäck T, Åkerud T, Wikström M, Schultz J. J. Am. Chem. Soc. 2002; 124: 11874-11880. See also Commentary in Highlights section of Nature Reviews Drug Discovery by Peter Kirkpatrick. NRDD 2002;1: 841.
3. Mechanism of action of pyridazine analogues on protein tyrosine phosphatase 1B (PTP1B).Tjernberg A, Hallèn D, Schultz J, James S, Benkestock K, Byström S, Weigelt J. Bioorg. Med. Chem. Lett. 2004;14: 891-895.
4. The devil is still in the details–driving early drug discovery forward with biophysical experimental methods. Lundqvist T. Curr. Opin. Drug Discovery 2005; 8: 513-519.