Development of a Compound Interaction Screen on a photoactivatable Cellulose Membrane (CISCM) to identify drug targets

Abstract

Proteins represent the main target class of bioactive compounds.1 The identification of a drug’s protein target is crucial to understand drug function on a molecular level. Major progress has been achieved in both the construction of large scale screening libraries of structures with drug-like properties as well as in high-throughput methods to screen the bioactivity of those substances against a phenotype of interest. Such phenotypic screens allow to identify hit compounds that alter the disease phenotype in a desired manner.2,3 However, the following deconvolution of the mode of action (MoA) and identification of corresponding protein targets remains a key challenge in drug discovery.2,4 Here we developed a novel target identification (target-ID) approach, which combines the power of photoaffinity labeling (PAL) and quantitative affinity purification mass spectrometry (qAP-MS). This compound interaction screen on a photoactivatable cellulose membrane (CISCM) allows us to probe the interaction of many proteins with multiple compounds in parallel without compound derivatization or structure-activity relationship (SAR) studies being required. To achieve this, we equipped cellulose membranes with a diazirine PAL-probe and photoimmobilized a selection of wellstudied bioactive compounds with ultraviolet (UV) light. The resulting photocrosslinked compound cellulose array was then incubated with protein extract and specific targets were identified via quantitative affinity purification mass spectrometry. For the natural structure analogues cyclosporine A (CsA), tacrolimus (FK506) and sirolimus (rapamycin) this reliably identified known protein interactors, while the known targets of small fragment-like compounds, such as (S)-thalidomide, lenalidomide, methotrexate and metformin, could not be detected. As another limitation, the target tubulin beta (TUBB) could not be identified as a specific interactor of the pharmaceuticals vinblastine and paclitaxel. The observed results were nearly unaffected by the increase of the spacer-arm length, separating the diazirine from the cellulose surface, and by photocrosslinking the compounds isolated from the cellulose membranes with subsequent immobilization of resulting photocrosslinking products. Physisorbed CsA on diazirine-modified non-UV irradiated cellulose membranes could identify the known interactors Peptidyl-prolyl cis-trans isomerases A and F (PPIA, PPIF), whereas physisorbed CsA, sirolimus and FK506 on non-functionalized cellulose membranes could not identify any of their known protein interactors. In summary, we developed a novel compound interaction screen that can rapidly screen for drug targets in a parallel fashion without prior derivatization of the drugs. CISCM reliably identifies the targets of natural products and current limitations of our method are due to the known limitations of diazirine-based photocrosslinking strategies and affinity purification mass spectrometry.