Articles pubs.acs.org/acschemicalbiology
The Commonly Used PI3-Kinase Probe LY294002 Is an Inhibitor of BET Bromodomains Antje Dittmann,† Thilo Werner,† Chun-Wa Chung,‡ Mikhail M. Savitski,† Maria Fal̈ th Savitski,† Paola Grandi,† Carsten Hopf,† Matthew Lindon,§ Gitte Neubauer,† Rabinder K. Prinjha,§ Marcus Bantscheff,*,† and Gerard Drewes*,† † ‡ §
Cellzome GmbH, Meyerhofstrasse 1, 69117 Heidelberg, Germany Molecular Discovery Research, GlaxoSmithKline Medicines Research Centre, Stevenage SG1 2NY, United Kingdom Epinova DPU, GlaxoSmithKline Medicines Research Centre, Gunnels Wood Road, Stevenage SG1 2NY, United Kingdom S Supporting Information *
ABSTRACT: A commonly used small-molecule probe in cell-signaling research is the phosphoinositide 3-kinase inhibitor LY294002. Quantitative chemoproteomic profiling shows that LY294002 and LY303511, a close analogue devoid of PI3K activity, inhibit the BET bromodomain proteins BRD2, BRD3, and BRD4 that comprise a family of targets structurally unrelated to PI3K. Both compounds competitively inhibit acetyl-lysine binding of the first but not the second bromodomain of BET proteins in cell extracts. X-ray crystallography shows that the chromen-4-one scaffold represents a new bromodomain pharmacophore and establishes LY294002 as a dual kinase and BET-bromodomain inhibitor, whereas LY303511 exhibits anti-inflammatory and antiproliferative effects similar to the recently discovered BET inhibitors.
C
analogue LY303511, including cancer cell line sensitization to TRAIL-induced apoptosis12 and the inhibition of NF-κB activation13 and IFNβ14 production in LPS-stimulated macrophages. Here, we report that LY294002 and LY303511 bind to a family of targets structurally unrelated to PI3K, the BET bromodomain proteins, and that LY303511 induces antiinflammatory and antiproliferative effects similar to the recently reported BET inhibitors.15,16 We17 and others18 have explored immobilized derivatives of LY294002 for the affinity purification of PI3K family kinases. These pull-down experiments revealed that immobilized LY294002 binds to many other proteins not related to PI3Ks, but the specificity and affinity of these interactions has remained unclear. In an effort to characterize fully the target profile of LY294002, we employed a quantitative chemoproteomics-based binding assay. NHS-activated sepharose was derivatized either with an amine version of the lipid kinase active LY294002 (N-LY294002)17 or with inactive analogue LY303511 immobilized via its secondary amino group (Figure 1a). The two types of beads were incubated with HL60 cell extract, and captured proteins were quantified by tandem mass spectrometry (MS/MS) analysis of the combined peptide pools after tryptic digestion and isobaric labeling. The sets of proteins captured by the PI3K-active and PI3K-inactive analogues were
hemical probes for protein activity are valuable tools for the study of protein function in cells and for the discovery and validation of therapeutic targets. It is crucial to characterize fully the selectivity of such probes to rule out the possibility that the observed phenotypes may be resulting from off-targets rather than the perceived target protein. Recently, minimal criteria for selectivity have been suggested, and initiatives have been launched to make validated probes available.1,2 However, many commonly used probe compounds will likely not meet these criteria. Frequently, selectivity is only determined for structurally related targets and, in industrial research, for panels of receptors and enzymes with known pharmacological properties.3 Unfortunately, a comprehensive assessment of probe selectivity is not straightforward, but quantitative chemical proteomics offers a relatively unbiased approach to discover the off-targets of small molecules.4−6 A commonly used small-molecule probe in cell-signaling research is the chromone derivative LY294002 discovered in 1994,7 which is widely used to study the role of phosphoinositide 3-kinases (PI3Kα, β, γ, and δ) in the signal transduction pathways involved in cell migration, proliferation, survival, and metabolism.8,9 On the basis of these data, PI3Ks were proposed to be attractive targets for cancer10 and other indications.8 A PubMed search for LY294002 yields more than 6800 entries, and despite advances in developing more potent and drug-like compounds,11 LY294002 continues to be used to probe PI3K signaling. However, a number of studies also reported PI3Kindependent effects, on the basis of the use of the PI3K-inactive © 2013 American Chemical Society
Received: October 14, 2013 Accepted: November 29, 2013 Published: November 29, 2013 495
dx.doi.org/10.1021/cb400789e | ACS Chem. Biol. 2014, 9, 495−502
ACS Chemical Biology
Articles
Figure 1. LY294002 and LY303511 are BET inhibitors. (a) Structures of LY303511, LY294002, an amine derivative of LY294002 for bead immobilization (N-LY294002), I-BET151 (GSK1210151), and I-BET121 (GSK923121). (b) Immobilized N-LY294002 and LY30351 bind a largely identical set of proteins. Symbols represent individual proteins mapped by quantitative MS/MS from pull-down experiments with beads derivatized with N-LY294002 or LY30351. Overall protein abundance in the experiment represented by peptide intensities in the MS1 spectra (log10 scale, see the text in the Supporting Information) is plotted on the y axis, and protein amount on N-LY294002 beads relative to the amount on LY303511 beads is plotted on the x axis (log2 scale). Significantly different proteins (p < 0.001, see the text in the Supporting Information) are marked as blue circles. Red circles represent BRD2, 3, and 4. (c) Comparison of target potencies of LY294002 and LY303511. N-LY294002-derivatized beads were incubated with cell extract in the presence of different concentrations of free LY294002 or LY303511. Competition of the free compound with the immobilized ligand for each captured protein was quantified by LC−MS/MS, and binding inhibition was expressed as the half-maximal inhibitory 496
dx.doi.org/10.1021/cb400789e | ACS Chem. Biol. 2014, 9, 495−502
ACS Chemical Biology
Articles
Figure 1. continued concentration (IC50). Symbols represent average pIC50 values of proteins quantified in n = 3 (LY294002) or n = 2 (LY303511) experiments. Proteins with significantly (p < 0.001, see the text in the Supporting Information) different pIC50 values are marked as blue circles. Red circles represent BRD2, 3, and 4. (d, e) Comparison of LY294002 and LY303511 potencies for BETs and PI3K isoforms after correction for target depletion. Shown are selected concentration−response curves from chemoproteomic experiments in panel c. Average IC50 and Kdapp values as a measure of binding inhibition are determined in n = 3 (LY294002) or n = 2 (LY303511) experiments. Correction factors were generated as described in Supporting Information Figure 1. n.d., not determined.
Figure 2. LY294002 and LY303511 bind selectively to the first bromodomain of BET proteins. (a) I-BET151 competes with immobilized NLY294002 for the binding to BET bromodomains but does not affect PI3K binding. BRD4 and PI3Kδ binding to N-LY294002 beads in the presence of various concentrations of I-BET151 (0.05−40 μM) was assessed by immunoblot (Supporting Information Figure 3a). Symbols represent average fold changes vs DMSO ± SEM from n = 3 experiments. (b) LY294002 competes only weakly with immobilized I-BET121 for the binding to BET proteins. Competition of BRD4 binding to I-BET121, N-LY294002, and H4K5acK8acK12ac by various concentrations of LY294002 (4−400 μM) was assessed by immunoblot analysis. (c) Point mutations in the first but not the second bromodomain abrogate the binding of BRD4 to beads with immobilized LY303511, N-LY294002, and H4K5acK8acK12ac but do not affect binding to beads with immobilized I-BET121. Indicated bead types were incubated with extracts of HEK293T cells expressing a FLAG-tagged wt, Y97A, Y390A, or Y97/390A BRD4s constructs. Binding of exogenous BRD4 was monitored by immunodetection of FLAG. Immunodetection of endogenous BRD4 served as a binding control. (d) LY294002 competes only with the binding of BRD4sY390A to I-BET121. Nuclear extracts from HEK293T cells expressing FLAG-tagged wt, Y97A, or Y390A BRD4s constructs were treated with various concentrations of LY294002 before incubation with I-BET121 beads. Binding inhibition by LY294002 of exogenous BRD4 was monitored by immunodetection of the FLAG tag. (e) X-ray crystallography of LY294002 in (1) PI3Kγ (PDB ID 1e7v) and (2) BRD4-BD1, (3) binding mode of LY303511 in BRD4-BD1, and (4) overlay of LY294002 in BRD4-BD1 (blue) with BRD4-BD2 (magenta) of PDB ID 2yem.
highly overlapping with some exceptions, most notably, PI3Ks, PRKDC, mTOR, and CK2, which were significantly (p
