Guide to virtual screening: application to the Akt phosphatase PHLPP

Abstract

We present an example-based description of virtual screening (VS) techniques used to identify new regulators of the Akt phosphatase PHLPP (PH domain Leucine repeat Protein Phosphatase). This enzyme opposes the effects of two kinases, Akt and PKC, which play a major role in cell growth and survival. Therefore, PHLPP is a potential therapeutic target in pathophysiologies where these pathways are either repressed, such as in diabetes and cardiovascular diseases, or over-activated as in cancer. To the best of our knowledge, no PHLPP inhibitors have been reported so far in the literature. In this study, we used a combination of chemical and virtual screening techniques that led to the identification of a number of inhibiting compounds with diverse scaffolds. These compounds bind PHLPP and inhibit cell death when tested in cellular assays. We employed GLIDE docking software to screen a library of more than 40,000 compounds selected from the NCI open depository (250,000 compounds) by similarity searches. We compare the efficiency at which we determined binding compounds from the chemical screen, and compare enrichment factors of the virtually discovered compounds over chemical screening.

Fig. 1.

Docking scheme: A 3-dimensional grid file and a compound library are necessary for docking. The grid file is derived from a PDB file or from a homology model. The homology model is then modified to add or remove ions or ligands in the active site, and fill in missing residues. The compound library is narrowed via similarity searches. Then GLIDE using the funneling scheme is employed.

Fig. 2.

PHLPP2 phosphatase domain model structure. (a) The entire phosphatase domain homology model with the approximate area which formed the docking grid for PHLPP2 lightly shaded, the rest of the phosphatase domain is darkly shaded, the two Mn²⁺ ions are represented as dark orbs in the center of the image, and the dashed lines represent the region enlarged in subsequent panels. (b, c) Docked poses of two hit molecules, which were verified as low μM inhibitors of PHLPP2, compound 45586 (b), and compound 134145 (c), both in stick representation (a full color image is available online or from the authors).