Pooled screening for antiproliferative inhibitors of protein-protein interactions

Abstract

Protein-protein interactions (PPIs) are emerging as a promising new class of drug targets. Here, we present a novel high-throughput approach to screen inhibitors of PPIs in cells. We designed a library of 50,000 human peptide-binding motifs and used a pooled lentiviral system to express them intracellularly and screen for their effects on cell proliferation. We thereby identified inhibitors that drastically reduced the viability of a pancreatic cancer line (RWP1) while leaving a control line virtually unaffected. We identified their target interactions computationally, and validated a subset in experiments. We also discovered their potential mechanisms of action, including apoptosis and cell cycle arrest. Finally, we confirmed that synthetic lipopeptide versions of our inhibitors have similarly specific and dosage-dependent effects on cancer cell growth. Our screen reveals new drug targets and peptide drug leads, and it provides a rich data set covering phenotypes for the inhibition of thousands of interactions.

Figure 1. Overview of library design and identification of cancer-specific inhibitors

(A) A library of 50,549 peptides containing interaction motifs was designed, synthesized on oligonucleotide arrays and cloned into a lentiviral library. (B) A dropout screen was performed with expression of the peptides after puromycin selection and cell collection of four time points in triplicates (at 0, 10, 15 and 24 days for RWP1 cells and 0, 9, 14 and 23 days for HEK283T cells). (C) Effect on cell viability of each peptide was quantified using high-coverage sequencing, with the read count of each peptide being an indication of the viability of cells expressing it. (D) The target of each peptide was identified using structural and bioinformatics methods and subsequently validated using pulldown and Förster resonance energy transfer (FRET) experiments.

Figure 2. Identification and experimental validation of cancer-specific inhibitors

(A) Venn diagram of identified peptides leading to dropout/increase specifically in RWP1 (cancer) or HEK293T (normal) cells, or leading to common dropout/increase. (B) Experimental validation of the effect of peptides on cell viability in single-lentivirus infection experiments. Red bars represent cancer (RWP1)-specific peptides and blue bars represent normal (HEK293T)-specific peptides. Blue dashed line indicates average cell viability of HEK293T-specific peptides in cancer (RWP1) cells. Meanwhile, red dashed line indicates average cell viability of RWP1-specific peptides in normal (HEK293T) cells. Statistical analysis was performed using Student’s t-test (2-tails) by comparing the cell viability of each RWP1-specific peptide against those of HEK293T-specific peptides (upper panel). Comparison of cell viability of each HEK293T-specific peptide against those of RWP1-specific peptides was also performed (lower panel). * P-value < 0.05. Experiments were done in triplicate. Data represent mean values ± s.d.

Figure 3. Characterization of peptide-target interactions

(A) Schematic view of the identification of interactions between peptides and targets. Peptides are either mapped directly to structures of protein interactions in the PDB or are mapped to known domain-motif interactions where both the linear motif and the protein domain involved in the interaction are known. (B) Evaluation of peptide-induced disruption between target and source protein in pulldown experiments. In all four tested cases, the expression of the peptide leads to disruption of the interaction. Also, the peptide is pulled down by its putative binding partner. Experiments were done in triplicate. Full gel images in Supplementary Fig. 11. (C) FRET assay for the determination of binding of peptides to targets. Strong FRET signal is seen, indicating direct in vivo interactions between the peptide and its putative binding partner. White scale bars represent 15 μm. (D) shRNA-induced cell viability rescue experiment. Cell viabilities are measured with cells stably expressing shRNA or transfected with the peptide. For the rescue experiment, cells stably expressing shRNA are transfected with the peptide. * P-value < 0.05. Experiments were done in triplicate. Data represent mean values ± s.d.

Figure 4. Mechanism of action of cancer-specific inhibitors

Downstream effect of peptide (A) ETFSDLW (inhibiting P53/MDM2). As is documented, disruption of this interaction leads to a decrease in degradation of P53, thereby increasing P53 levels, triggering apoptosis.. Experiments were done in triplicate. Full gel images in Supplementary Fig. 11. (B) TEGPDSD (inhibiting p53/SFN). Inhibition of this interaction appears to lead to decrease in cyclin B1 and cdc2 levels (and an increase in cyclin A), suggesting it may trigger cell cycle arrest. Blue circle represents source protein of peptide, and green square represents targets of peptide. Red line represents interaction between cancer-specific inhibitor and target. Experiments were done in triplicate. Full gel images in Supplementary Fig. 11. (C) Effect of peptide TEGPDSD on cell cycle. Fluorescence-activated cell sorting (FACS) analysis confirms that it leads to cell cycle arrest in G2. Experiments were done in triplicate. Data represent mean values ± s.d. (D) Effect of synthetic lipidated peptides on cell viability. Synthetic GNG4, MUS81 and INCENP derived lipopeptides show strong dosage dependent effects on cell viability in RWP1 cells, while scrambled versions do not affect cell viability. Experiments were done in duplicate. Data represent mean values ± s.d.