There has been a significant progress in the systems biology field during the last decade. Sudhir Sahasrabudhe of Prolexys Pharmaceuticals explains how his company\'s application of proteomics technologies has resulted in the identification of novel disease targets, drug candidates and biological markers predictive of drug response.
The Human Genome Project left the industry with an enormous amount of data, but little additional knowledge of human health and disease states. The term proteomics is used to describe the next quantum leap in understanding the participation of proteins, the products of the genome, in health and disease. The study and mapping of the human proteome (or proteomics) is the critical link between the data generated from the Human Genome Project and the use of that information to solve problems in human health and disease.
Proteomics encompasses more than just identification and quantification of proteins. It also includes their localisation, modifications, interactions, activities and, ultimately, their biological function and therapeutic potential (see Figure 1). The company a protein keeps, or its interactome, provides key insights into a protein's role in health and disease.
Prolexys Pharmaceuticals has developed an integrated technology platform for protein characterisation, mapping of protein interactions, identification of novel therapeutic targets, target validation, and drug discovery using:
Prolexys also uses this knowledge base and technology platform to identify novel therapeutic targets and develop new chemical entities to treat cancer.
The HyNet yeast two-hybrid (Y2H) method of discovering binary protein interactions is based on the use of two structural components of a yeast transcription factor (see Figure 2). The two components are the DNA binding domain (BD) and the transcriptional activation domain (AD). Both of these domains are required at the promoter of the reporter genes for transcriptional activation and yeast survival. Proprietary yeast strains are used as a host for cloned human protein fragments from diverse human tissue types fused with either the BD or AD. Plasmid libraries are constructed that encode human protein fused to either of the two domains to create bait and prey hybrid proteins. Bait proteins are fused to the BD and this hybrid binds directly to the promoter of reporter genes.
In contrast, the prey proteins are fused to the AD. These hybrids are recruited to the promoter only if the human protein portions of the bait and prey proteins bind each other. Libraries of baits and preys are created in haploid yeast of opposite mating type. Mating between these haploids results in diploids with bait protein, prey protein and the reporter genes. If the bait and prey interact, a complex is formed that localises the AD to the reporter genes causing their transcription. Yeast colonies grow only if the reporter genes are active, which is the basis of this assay for human protein interactions.
A unique feature of the company's HyNet strategy is a random approach to bait selection. Most other efforts use a directed screening approach in which specific proteins of interest are used as baits. The random approach that Prolexys employs eliminates the need to specifically clone baits and characterise them for their suitability in the Y2H assay. This random approach in HyNet has the additional advantage that it affords the possibility to discover unknown proteins and their interactions within the human proteome.
To compliment the random approach to bait selection, specific proteins of therapeutic or disease biology interest are selected for use in directed Y2H screens. As an illustration of the breadth of information that can be obtained from the high-throughput Y2H process, we have assembled world's largest database of human protein interactions (120,000 non-redundant interactions) covering approximately half of the human genome. More recently we elaborated the protein interaction map of malarial parasite Plasmodium falciparum (Nature 438:103-107, 2005) in collaboration with Dr Stan Fields (University of Washington, Seattle, USA) to enable identification of pathways and targets suitable for therapeutic and prophylactic intervention.
This high-throughput platform is supported by automation, robotics and an elaborate Laboratory Information Management System (LIMS). Figure 3 depicts a flow chart of the actual HyNet process, listing each individual process step. A process of this complexity requires rigorous sample tracking and recording of process and results data. This is achieved by using the LIMS system custom designed to support the workflow.
HySpec is a proprietary, directed process of protein expression, purification and high throughput analysis of interactions based on pull down experiments and mass spectrometry. The HySpec process is conducted as follows :
Identifying the mode of action and evaluating the specificity of new lead compounds are essential and critical steps in the preclinical drug development process. However, in many cases the mode of action of active compounds cannot be easily assessed in complex cell-based screening systems.
At Prolexys, we have developed a chemi-proteomics approach to identify drug targets from human cell lysates. Compounds are immobilised to a solid support via special linkers to reduce steric hindrance. This process is compatible with a variety of immobilisation chemistries. This method is well integrated into our sensitive HySpec analysis process and allows for rapid and simple screening of small molecules to determine their specific targets and, consequently, the mode of action (see Figure 5). This process can also identify secondary targets that could lead to undesirable side effects. The chemi-proteomics process allows the identification of not only the primary targets of a drug, but also the proteins interacting with the target. Therefore, this method can provide pathway information to put the primary target into a broader context. This combination of target and pathway information can be used to prioritise targets for continued drug development.
We have used a combination of HyNet and HySpec to identify novel protein-protein interactions, or protein-drug interactions (see Figure 6) as a starting point for therapeutic intervention in the area of cancer. The combination of these two workflows was used to elaborate the beta-catenin pathway. Colon adenocarcinoma (colon cancer) is believed to be initiated and maintained because of aberrant beta-catenin pathway activity. Prolexys has identified several small molecule drug candidates that antagonise novel protein targets in this pathway.
This review will describe one example in which the chemi-proteomics technology was used to identify the drug target, and eventually small molecule compounds that are selectively toxic to tumour cells with aberrant RAS pathway activity. Analogues of erastin series are RAS- and RAF- pathway selective anti-tumour agents. Erastin was discovered in the laboratory of Dr Brent Stockwell at MIT/Whitehead using a synthetic lethal screening approach. Using a human cell line engineered to co-express four oncoproteins, a high-throughput screen was executed and nine compounds were identified that selectively killed these cells over their non-tumourigenic counterparts.
One of these compounds, erastin, acts solely on human cells co-expressing small T antigen (ST) and an activated H-RAS oncogene. The chemi-proteomics approach was employed to identify protein(s) that selectively bind to erastin. Erastin acts through target of erastin (TOE) protein and thereby causes the appearance of oxidative species in cells with oncogenic RAS or RAF. Erastin is effective in numerous tumor cell lines. Use of erastin revealed that oncogenic RAS is associated with increased TOE protein levels in tumor cells and that TOEs are potential cancer drug targets. Medicinal chemistry efforts at Prolexys have led to the identification of a potent small molecule PRLX 93936 with highly desirable pharmaceutics properties.