Potent macrocycle inhibitors of the human SAGA deubiquitinating module

Abstract

The Spt-Ada-Gcn5 acetyltransferase (SAGA) transcriptional coactivator contains a four-protein subcomplex called the deubiquitinating enzyme (DUB) module that removes ubiquitin from histone H2B-K120. The human DUB module contains the catalytic subunit ubiquitin-specific protease 22 (USP22), which is overexpressed in a number of cancers that are resistant to available therapies. We screened a massive combinatorial library of cyclic peptides and identified potent inhibitors of USP22. The top hit was highly specific for USP22 compared with a panel of 44 other human DUBs. Cells treated with peptide had increased levels of H2B monoubiquitination, demonstrating the ability of the cyclic peptides to enter human cells and inhibit H2B deubiquitination. These macrocycle inhibitors are, to our knowledge, the first reported inhibitors of USP22/SAGA DUB module and show promise for development.

Keywords: USP22; chromatin; deubiquitinating enzyme; drug discovery; macrocyclic peptides; transcription; ubiquitin.

Figure 1 –

RaPID cyclic peptide library screening: A) Schematic of Random nonstandard Peptide Integration Discovery platform used for targeting the SAGA DUB module for binding species. B) Sequences for each peptide selected from RaPID screen for further evaluation as inhibitors. Each peptide contains the indicated D- or L-amino acid at the N-terminus covalently linked to the C-terminal cysteine sulfhydryl group.

Figure 2 –

Inhibition of human DUB module activity inhibition by cyclic peptides. A-F) Each indicated cyclic peptide was tested for its ability to inhibit Ub-AMC cleavage in the presence of increasing amounts of the peptide into a reaction with a fixed amount of DUB (200 nM) and Ub-AMC (1 μM). K_i values were fitted using the Morrison approximation with a fixed K_m value of 35 μM as determined in Figure 3A. G) Inhibition of human DUB module activity on H2B ubiquitinated nucleosomes. Effect of each cyclic peptide (1 uM) and DMSO control on hDUB module cleavage rates on ubiquitinated nucleosomes. H) Quantitation of the disappearance of the H2B-Ub band,

Figure 3 –

hD1 is a non-competitive inhibitor requiring the cyclized form for optimal USP22 targeting. A) hDUBm cleavage rates were measured at multiple substrate and hD1 concentrations to determine inhibitor equilibrium binding constant (K_i) and mode of inhibition. The data were best fit to a non-competitive model of enzyme inhibition, with a measured K_m of 35 μM, k_cat of 2.8 s⁻¹, and K_i of 180 nM. B) Cleavage rates of 5 μM Ub-AMC by 200 nM hDUBm in the presence of the indicated concentration of either linear, scrambled, or cyclic hD1 peptide sequences fitted using IC₅₀ analysis. IC₅₀ values were calculated as 270 nM for the cyclic hD1, 1 μM for the linear version, and >1000 μM for the scrambled hD1 sequence.

Figure 4 –

Small angle x-ray scattering (SAXS) analysis of the DUB module in the presence and absence of cyclic peptide. A) Scattering intensities of 1 mg/mL (6 μM) hDUBm alone (black), hDUBm in the presence of 7uM hD1 (red), and theoretical scattering intensities calculated from the yDUBm crystal structure (PDB ID: 3MHH, cyan). The right y-axis describes raw measured intensities (upper curve), while the left y-axis is a plot of I*s values (Kratky plot). B) Frequency distribution (P(r)) of possible vectors between surfaces on the scattering sample. Experimental curves from hDUBm (black) and hDUBm plus hD1 inhibitor (red); calculated curve for yDUBm (cyan). D_max is indicated by a colored arrow corresponding to each sample. C) Electron density maps derived from SAXS data of hDUBm alone (gray) and hDUBm in the presence of hD1 (red). Dashed line indicates D_max. Correlations between the peptide-bound and apo complexes indicated by pseudocoloring (scale below). The yeast DUB module structure (green) was used to independently orient both density maps.

Figure 5 –

hD1 is specific for USP22. A) DUBProfiler assay of the effect of hD1 on a panel of 44 human DUBs. Cleavage of a ubiquitin-rhodamine(110)-glycine substrate was compared in the presence and absence of 1 μM hD1. B) Inhibition of H2B DUB complexes as a function of hD1 concentration. Cleavage rates of 5 μM Ub-AMC were assayed for 75 nM USP22/ATXN7L3/ENY2/ATXN7 (hDUBm), 100 nM USP27x/ATXN7L3/ENY2, or 1000 nM USP51/ATXN7L3/ENY2. Rates were normalized to the uninhibited reaction in each case to determine the fraction of apparent activity. IC₅₀ values of hD1 were 100 nM for USP22, 3 μM for USP27x, and >250 μM for USP51.

Figure 6 -

Cyclic peptide inhibitors of hDUBm increase cellular H2B-Ub levels. A) HEK293T cells incubated with 5 μM of each cyclic peptide for 2 hours. Histone proteins were extracted, and samples analyzed by western blot with anti-H2BK120ub antibody. Bottom: anti-histone H3 loading control. B) Quantitation of western replicate results indicates significant enrichment of H2B-Ub when normalized to H3 abundance. Bars labeled with two asterisks indicate p < 0.01 (hD1 and hD5), and a single asterisk indicates p < 0.05 (hD2, hD3, hD4, and PR-619) when each measurement is compared with untreated cells.