July 1, 2012 (Vol. 32, No. 13)

Doris Hafenbradl, Ph.D. BioFocus
Dave Sheppard, Ph.D. BioFocus

Effective Drug Discovery Necessitates New Tools and Technologies that Can Get the Job Done

For many disease areas including oncology, autoimmune/inflammatory diseases, metabolic diseases, neurodegeneration, and viral infections, there is a growing interest in epigenetic targets. Many of these targets are unexplored, and a lack of suitable assay formats and tool compounds, coupled with limited understanding of the relevance of the protein targets to specific diseases and of the cellular phenotypes they can cause, makes progress difficult.

While the know-how around the histone deacetylase (HDAC) family and the DNA methylases is clinically well advanced, tools and technologies for histone methyltransferases (HMTs), demethylases, bromo- and chromo-domains, ubiquitin, and SUMO ligases and deubiquitinases are just emerging. The requirements of developing meaningful assays and generating valuable HTS data for such targets remain challenging.

The number of protein crystal structures for epigenetic targets is increasing through the work of several initiatives. For example, the Structural Genomics Consortium-led initiative aims to develop “chemical probes” that are specifically designed to affect the activity of proteins involved in epigenetic control. These probes could provide potential starting points for drug discovery. However, the availability of tool compounds to understand better the binding modes, conformational changes, and protein flexibility is still very limited.

Despite the lack of specific tool compounds and inhibitors, computational approaches have been valuable. At BioFocus, these computational approaches address entire target families and apply efficient computational compound selection strategies to screening.

Computational Approaches

In order to apply efficient chemogenomics concepts to epigenetic targets, BioFocus has developed EpiRoadmap (Figure 1), an integrated toolbox that enhances the ideas and concepts of the previously developed Roadmap. Various protocols and scripts enable relevant data for an entire target family and their relationships to be compiled rapidly from public sources.

A Java-based platform enables the fast interactive setup of roadmap cartoons for any target family. Rapid browsing through selected members of a target family and target list handling are possible, as is differential analysis of target and off-target binding sites or sub-sites.

Advanced features include similarity searching and clustering for all residues or defined subsets of a drug-binding site. A statistical interface1 to EpiRoadmap enables the generation of un-rooted tree diagrams (as used in phylogenetics) for assessing target similarity relations. Various metrics can be used for the classification approaches.


Figure 1. Screenshot of BioFocus’ EpiRoadmap toolbox

Figure 2 shows the differential analysis of the nearest neighbors of HMT G9a, an enzyme that modifies histone 3 (H3K9). When based on a full-sequence analysis, G9a clusters with only two other family members (Figure 2A). However, the analysis of the S-adenosylmethionine (SAM) binding pocket of G9a provides an entirely different picture, bringing other proteins such as MLL1 into a close relationship with G9a (Figure 2B). Equally, focusing on the substrate-binding site of G9a suggests that MLL2 and MLL3 have high homology with G9a (Figure 2C).


Figure 2. The differential analysis of the nearest neighbors of HMT G9a, an enzyme that modifies histone 3 (H3K9). When based on a full-sequence analysis, G9a clusters with only two other family members (A). However, the analysis of the S-adenosylmethionine (SAM) binding pocket of G9a provides an entirely different picture, bringing other proteins such as MLL1 into a close relationship with G9a (B). Equally, focusing on the substrate-binding site of G9a suggests that MLL2 and MLL3 have high homology with G9a (C).

Assay Technologies

BioFocus has compared and evaluated a number of assay technologies for epigenetic targets. The gold standard remains the radiometric assay format, which allows use of highly relevant substrates. In the specific case of HMTs, BioFocus has used nucleosome preparations, recombinant histone preparations, and other protein substrates or peptides in the radiometric assay format.

For certain members of this target family no peptide substrate has yet been identified, and some protein targets only show enzymatic activity when nucleosome preparations are used as substrate.

Besides the feasibility of assay development, the pharmacological relevance of the screening assays supports the use of endogenous protein substrates. However, there are a number of target family members where homogeneous assay technologies have led to highly comparable screening results. As these technologies are often more cost- and time-efficient they are preferred for primary high-throughput screening (HTS).

In addition to radiometric assays as an HTS method for epigenetic drug discovery for HMTs, histone demethylases, and bromodomains, BioFocus has focused on Almac’s FLEXYTE® fluorescence lifetime technology (FLT) and Caliper’s mobility shift assays. The details of these assay technologies are described in Figure 3 and in Wigle et al.2

TR-FRET, AlphaScreen®, fluorescence polarization (FP), and ELISA-based methods are also available on the market with various advantages and disadvantages concerning reproducibility, specificity, throughput, false-positive/false-negative issues, label limitations, and flexibility of substrate use.

BioFocus has designed and executed case studies for the HMT G9a, a target relevant to oncology and HIV. Further case studies focus on the histone demethylase LSD1, a target expected to be relevant to oncology and bromodomain BRD4, where a number of lead compounds have been published and could be used as templates for the initiation of further ideas (not shown here).

The G9a case study started with the use of computational chemistry tools to select a small number of compounds from the BioFocus compound library, which consists of 900,000 small molecules and about 100,000 natural product samples.

For G9a, extensive information concerning protein structures and tool compound availability allowed us to select compounds using ligand-based as well as docking technologies, covering both the substrate-binding and the SAM-binding sites. Based on this approach, a total of 2,112 test compounds were selected for physical screening.

This small selection was screened three times using i) a FlashPlate® radiometric assay with recombinant full-length histone H3, ii) a FLT assay (as described in Figure 3), and iii) a mobility shift assay (as described in Wigle et al2). Both FLT and mobility shift assays applied a similar peptide substrate. The sensitivity of all three assays was confirmed using S-adenosylhomocysteine (SAH) and sinefungin as reference inhibitors, yielding highly similar IC50 values.

While no common compounds were identified in all three primary screens, we could conclude that the FLT and mobility shift assays primarily generated substrate competitive inhibitors, while the preference for the hit compounds from the radiometric assay was toward the SAM-binding site.

The SAM competitive inhibitors, which were identified by the radiometric and FLT assay, showed a broad range of selectivities against a small panel of HMTs. The most potent hits in the Caliper mobility shift assay all showed substrate competitive behavior, in accordance with the virtual screening source annotation and the mapping to the substrate-binding site.

Our studies have allowed us to evaluate the advantages and limitations of these different technologies in the context of HTS and mode-of-action studies. More details are available in Ahrens et al.3


Figure 3. Principle of FLEXYTE® fluorescence lifetime (FLT) assay platform for epigenetic enzyme targets: The assay employs histone peptide substrates site-specifically labelled with 9-aminoacridine (9AA) fluorophore. The FLT of 9AA is quenched to a short lifetime until the peptide is cleaved by the protease. Protease cleavage is possible only when the lysine/arginine residue of interest is unmodified. Methylation, acetylation, or citrullination by an appropriate epigenetic enzyme prevents protease-induced cleavage, resulting in retention of the short FLT. In contrast, demethylation or deacetylation renders the peptide susceptible to protease-induced cleavage, resulting in an increase in FLT. Hence, epigenetic enzyme activity is reported through changes in the measured FLT of 9AA.

Future Needs

For the future, technologies and tools are urgently required for detailed mode-of-action studies of inhibitors in cellular assays. Understanding the novel epigenetic drug targets, their complex network and interaction, and the relevant pharmacological cellular models is a major challenge. One of the main obstacles in designing the relevant cell-based assays is the current lack of suitable and specific antibodies.

Monitoring methods that can differentiate between mono-, di-, and trimethylation, for example, are urgently needed. Furthermore, technologies that can evaluate multiple modifications on histone proteins (e.g., methylation, acetylation, and phosphorylation events) could help us understand these complex modifications and their relevance to disease.

One technology that is a promising option for detailed epigenetics drug discovery, possibly even for HTS approaches, is the RapidFire®/MS technology (Agilent Technologies). This technology can demonstrate modification on histone proteins in combination with a label-free IC50 determination of inhibitors.4 These and similar technologies will be essential for understanding and evaluating hit and lead compounds in the epigenetics drug discovery arena.

Doris Hafenbradl, Ph.D. ([email protected]), is senior director, biology & natural products, and Dave Sheppard, Ph.D. ([email protected]), is director, computational chemistry at BioFocus. FLEXYTE® is a registered trademark of Almac Group, AlphaScreen® and FlashPlate® are registered trademarks of PerkinElmer, and RapidFire® is a registered trademark of Agilent Technologies.

Citations:

1. R Development Core Team (2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org.

2. Wigle TJ, Provencher LM, Norris JL, Jin J, Brown PJ, Frye SV, Janzen WP. Accessing protein methyltransferase and demethylase enzymology using microfluidic capillary electrophoresis. Chem Biol. 2010 Jul 30;17(7):677-8.

3. Ahrens T, Bergner A, Sheppard D, Hafenbradl D. Efficient hit-finding approaches for histone methyltransferases: the key parameters. J Biomol Screen. 2012 Jan;17(1):85-98.

4. Rye PT, Frick LE, Ozbal CC, Lamarr WA. Advances in label-free screening approaches for studying histone acetyltransferases. J Biomol Screen. 2011 Dec;16(10):1186-95.

Previous articleAdvertorial: DSM Pharma Biotechnology
Next articleEradicating Errors in Third-Gen Sequencing for de Novo Genome Assembly