The logic seems sound: If you want to design an efficacious drug, then you need to understand the target well. Yet the use of structure based drug design still has its distractors, here Dr Joanne McCudden extolls the virtues of the process
The value of Structure Based Drug Design (SBDD) in the drug discovery process was recently highlighted in work by UCB and Domainex1 which led to the successful discovery of novel inhibitors of Mitogen-activated protein kinase kinase (MAPKK, also known as MEK), which is over-active in many human cancers.
SBDD is the use of three-dimensional structural information about a drug target protein, obtained through X-ray crystallography or NMR methods, to either: a) identify small molecule compounds able to inhibit the function of the target (hit identification); and/or b) elaborate the identified hit within the constraints of the target’s binding pocket, in order to optimise compound potency and selectivity.
The availability of structural information about the drug target, i.e. being able to see how a ligand binds to a receptor or enzyme – and thus having a rational basis for understanding and improving binding interactions of a drug compound – can pay significant dividends when optimising the potency and specificity of a novel drug series.
Although there are still detractors, SBDD has been used successfully in the design of a number of successful drugs2. For instance, it was used in the discovery of Xalkori, a therapy launched by Pfizer last year to treat a rare and intractable form of lung cancer. It also has played a significant role in the development of an Alzheimer’s disease treatment by Eli Lilly & Co, an antibiotic designed by GlaxoSmithKline that is currently in clinical trials and a Sanofi anti-coagulant in the final stages of development.
In the recent case study reported by UCB and Domainex, the research groups worked together to develop a novel system to study the three-dimensional structure of MAPKK1. Although a crystal structure of this kinase had been reported in the public domain, attempts to reproducibly resolve the structure of MAPKK-inhibitor complexes were unsuccessful, and so alternative approaches were sought.
X-ray crystallography studies to resolve the structure of a protein requires large amounts of high quality, soluble protein. Frequently, particularly for mammalian proteins, recombinant expression of full-length proteins is problematic. As an alternative, the expression of functional sub-constructs or domains of a protein is an effective approach, so long as appropriate termini for the truncation can be identified. The standard method for identifying the appropriate termini for such an expressible sub-construct is to use bioinformatics to identify putative domain boundaries. However, this approach is often unsuccessful as bioinformatics cannot predict such boundaries with precision, so that expressible sub-constructs are typically found by trial and error.
SBDD can be a very efficient approach to drug discovery, allowing focussed design of novel compounds that specifically bind to the relevant binding pocket of a target protein
One approach to solving this problem is to randomly fragment a target gene in order to screen large numbers of potentially useful sub-constructs or domains. For example, Domainex is able to use its patented Combinatorial Domain Hunting (CDH) technology3 to produce very large unbiased finely-sampled gene-fragment libraries of up to 100,000 domains. In the Meier study, this technology identified the only domain construct, selected from a library of ~4,000 candidates, which enabled reproducible MAPKK crystallisation1.
The Meier study illustrated that bioinformatics methods can suggest domain boundaries which are not optimal for expression. In the case of the UCB-Domainex work, the initial (bioinformatically-derived) construct was poorly expressed, whereas the fragment of MAPKK which was identified using gene-fragmentation methods produced high expression levels. This example further validates the gene-fragmentation approach – and specifically the CDH technology – as a solution to the challenge of rapidly identifying highly expressible and crystallisable protein domain sub-constructs.
CDH was first reported by Reich et al. in 20063, as a means to screen up to 100,000 gene constructs for use in protein assay development and/ or structural studies. This technology has been further optimised since that initial publication, but essentially consists of the following steps:
- The drug target gene is recoded using proprietary algorithms, which optimises the gene for expression in E. coli and for the fragmentation. The recoded gene is then synthesised.
- To generate random fragments of the target gene, PCR amplification of the resynthesised gene is completed in the presence of deoxyuridine. This leads to the random incorporation of uracil in place of thymine.
- A mixture of DNA repair enzymes designed to recognise sites where uracil is present is used to create blunt-ended double-strand breaks in the DNA.
- Fragmented DNA of a relevant size for the protein region of interest is then selected and cloned into proprietary vectors that allow expression of all potential open reading frames which incorporate a poly-histidine tag sequence.
- Typically a library of 20,000-100,000 E. coli clones is plated, grown, and probed for the expression of soluble poly-histidine fusion protein. Typically about 750 colonies expressing high-levels of protein are selected and incubated at a larger scale. Soluble extracts from these cultures are partially purified and level of expression, solubility and stability of each protein fragment is analysed.
- Typically a number of clones expressing recombinant proteins which are of the correct size, soluble, and antigenically correct will be identified. These clones are end-sequenced to confirm their identity and location on the parental gene sequence. At this stage, protein fragments able to bind to a known ligand (which could be a small-molecule probe, or a peptide) can be selected for and used to develop binding or activity assays for use in small-molecule hit identification.
This approach has been used for a wide range of drug targets, including supporting Domainex’s in-house oncology research into the design of lysine methyltransferase and TBK1/IKKε inhibitors. In the case of the UCB-Domainex collaboration, the protein domains that were identified using the CDH approach were subsequently used to resolve the structure of MAPKK and a number of MAPKK-inhibitors co-structures; for the support of the UCB’s SBDD effort.
SBDD can be a very efficient approach to drug discovery, allowing focussed design of novel compounds that specifically bind to the relevant binding pocket of a target protein. However, this approach is reliant on the availability of structural information, which can be challenging to obtain for many targets.
A number of groups are working to develop technologies that will facilitate the production of suitable proteins for structural studies for a wider range of exciting drug targets. Protein-protein interactions are a particularly challenging area for drug discovery, and many researchers believe that these targets will often only be tractable through access to structural information. To that end Domainex has developed a variant of its technology, called CDH2, which allows the identification of heterodimeric domain complexes4. Technologies such as these will allow drug researchers to remain at the forefront of scientific discovery.
1. Meier C, et al. (2012) Engineering human MEK for structural studies: A case study of combinatorial domain hunting. Journal of Structural Biology; 177:329-34.
2. Rockoff J (2012) Drug Discovery gets an Upgrade. The Wall Street Journal; 16th April
3. Reich S, et al. (2006, October). Combinatorial Domain Hunting: An effective approach for the identification of soluble protein domains adaptable to high-throughput applications. Protein Science, 15(10), 2356–2365.
4. Maclagan K, et al. (2011, March) A combinatorial method to enable detailed investigation of protein-protein interactions. Future Med. Chem., 3(3), 271-282.
Author: Dr Joanne McCudden
Dr Joanne McCudden is a biochemist and business development professional, with over 13 years of experience in life sciences. Joanne joined Domainex in 2010 and is responsible for the company’s licensing and revenue generation activities