Mass spectrometry-based discovery proteomics is an essential tool for the proximal readout of cellular drug action. Here, we apply a robust proteomic workflow to rapidly profile the proteomes of five lung cancer cell lines in response to more than 50 drugs. Integration of millions of quantitative protein–drug associations substantially improved the mechanism of action (MoA) deconvolution of single compounds. For example, MoA specificity increased after removal of proteins that frequently responded to drugs and the aggregation of proteome changes across cell lines resolved compound effects on proteostasis. We leveraged these findings to demonstrate efficient target identification of chemical protein degraders. Aggregating drug response across cell lines also revealed that one-quarter of compounds modulated the abundance of one of their known protein targets. Finally, the proteomic data led us to discover that inhibition of mitochondrial function is an off-target mechanism of the MAP2K1/2 inhibitor PD184352 and that the ALK inhibitor ceritinib modulates autophagy.
Subscribe to Journal
Get full journal access for 1 year
only $14.08 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
Lamb, J. et al. The connectivity map: using gene-expression signatures to connect small molecules, genes and disease. Science 313, 1929–1935 (2006).
Ye, C. et al. DRUG-seq for miniaturized high-throughput transcriptome profiling in drug discovery. Nat. Commun. 9, 4307 (2018).
Bush, E. C. et al. PLATE-Seq for genome-wide regulatory network analysis of high-throughput screens. Nat. Commun. 8, 105 (2017).
Subramanian, A. et al. A next generation connectivity map: L1000 platform and the first 1,000,000 profiles. Cell 171, 1437–1452 (2017).
Litichevskiy, L. et al. A library of phosphoproteomic and chromatin signatures for characterizing cellular responses to drug perturbations. Cell Syst. 6, 424–443 (2018).
Abelin, J. G. et al. Reduced-representation phosphosignatures measured by quantitative targeted MS capture cellular states and enable large-scale comparison of drug-induced phenotypes. Mol. Cell Proteom. 15, 1622–1641 (2016).
Chernobrovkin, A., Marin-Vicente, C., Visa, N. & Zubarev, R. A. Functional Identification of Target by Expression Proteomics (FITExP) reveals protein targets and highlights mechanisms of action of small molecule drugs. Sci. Rep. 5, 11176 (2015).
Lee, R. F. S. et al. Expression proteomics study to determine metallodrug targets and optimal drug combinations. Sci. Rep. 7, 1590 (2017).
Saei, A. A. et al. ProTargetMiner as a proteome signature library of anticancer molecules for functional discovery. Nat. Commun. 10, 5715 (2019).
Hughes, C. S. et al. Ultrasensitive proteome analysis using paramagnetic bead technology. Mol. Syst. Biol. 10, 757 (2014).
Hughes, C. S. et al. Single-pot, solid-phase-enhanced sample preparation for proteomics experiments. Nat. Protoc. 14, 68–85 (2019).
Meier, F., Geyer, P. E., Virreira Winter, S., Cox, J. & Mann, M. BoxCar acquisition method enables single-shot proteomics at a depth of 10,000 proteins in 100 minutes. Nat. Methods 15, 440–448 (2018).
Duncan, J. S. et al. Dynamic reprogramming of the kinome in response to targeted MEK inhibition in triple-negative breast cancer. Cell 149, 307–321 (2012).
Montagut, C. et al. Elevated CRAF as a potential mechanism of acquired resistance to BRAF inhibition in melanoma. Cancer Res. 68, 4853–4861 (2008).
Wilson, T. R. et al. Widespread potential for growth-factor-driven resistance to anticancer kinase inhibitors. Nature 487, 505–509 (2012).
Wu, C. P. et al. Human ATP-binding cassette transporters ABCB1 and ABCG2 confer resistance to histone deacetylase 6 inhibitor ricolinostat (ACY-1215) in cancer cell lines. Biochem. Pharm. 155, 316–325 (2018).
Wood, S. A., Park, J. E. & Brown, W. J. Brefeldin A causes a microtubule-mediated fusion of the trans-Golgi network and early endosomes. Cell 67, 591–600 (1991).
Niu, T. K., Pfeifer, A. C., Lippincott-Schwartz, J. & Jackson, C. L. Dynamics of GBF1, a brefeldin A-sensitive Arf1 exchange factor at the Golgi. Mol. Biol. Cell 16, 1213–1222 (2005).
Echeverria, P. C., Bernthaler, A., Dupuis, P., Mayer, B. & Picard, D. An interaction network predicted from public data as a discovery tool: application to the Hsp90 molecular chaperone machine. PLoS ONE 6, e26044 (2011).
Lai, A. C. & Crews, C. M. Induced protein degradation: an emerging drug discovery paradigm. Nat. Rev. Drug Discov. 16, 101–114 (2017).
Huang, H. T. et al. A chemoproteomic approach to query the degradable kinome using a multi-kinase degrader. Cell Chem. Biol. 25, 88–99 (2018).
Han, T. et al. Anticancer sulfonamides target splicing by inducing RBM39 degradation via recruitment to DCAF15. Science 356, eaal3755 (2017).
An, J. et al. pSILAC mass spectrometry reveals ZFP91 as IMiD-dependent substrate of the CRL4(CRBN) ubiquitin ligase. Nat. Commun. 8, 15398 (2017).
Folkes, A. J. et al. The identification of 2-(1H-indazol-4-yl)-6-(4-methanesulfonyl-piperazin-1-ylmethyl)-4-morpholin-4-yl-thieno[3,2-d]pyrimidine (GDC-0941) as a potent, selective, orally bioavailable inhibitor of class I PI3 kinase for the treatment of cancer. J. Med. Chem. 51, 5522–5532 (2008).
Raynaud, F. I. et al. Biological properties of potent inhibitors of class I phosphatidylinositide 3-kinases: from PI-103 through PI-540, PI-620 to the oral agent GDC-0941. Mol. Cancer Ther. 8, 1725–1738 (2009).
Dowdle, W. E. et al. Selective VPS34 inhibitor blocks autophagy and uncovers a role for NCOA4 in ferritin degradation and iron homeostasis in vivo. Nat. Cell Biol. 16, 1069–1079 (2014).
Ganley, I. G., Wong, P. M., Gammoh, N. & Jiang, X. Distinct autophagosomal–lysosomal fusion mechanism revealed by thapsigargin-induced autophagy arrest. Mol. Cell 42, 731–743 (2011).
Kim, J., Kundu, M., Viollet, B. & Guan, K. L. AMPK and mTOR regulate autophagy through direct phosphorylation of Ulk1. Nat. Cell Biol. 13, 132–141 (2011).
Klaeger, S. et al. The target landscape of clinical kinase drugs.Science 358, eaan4368 (2017).
Hanzelmann, S., Castelo, R. & Guinney, J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics 14, 7 (2013).
Cheng, Y. & Tian, H. Current development status of MEK inhibitors. Molecules 22, 1551 (2017).
Barrett, S. D. et al. The discovery of the benzhydroxamate MEK inhibitors CI-1040 and PD 0325901. Bioorg. Med. Chem. Lett. 18, 6501–6504 (2008).
Savitski, M. M. et al. Tracking cancer drugs in living cells by thermal profiling of the proteome. Science 346, 1255784 (2014).
Kronke, J. et al. Lenalidomide causes selective degradation of IKZF1 and IKZF3 in multiple myeloma cells. Science 343, 301–305 (2014).
Ruprecht, B. et al. Hydrophilic strong anion exchange (hSAX) chromatography enables deep fractionation of tissue proteomes. Methods Mol. Biol. 1550, 69–82 (2017).
Ritorto, M. S., Cook, K., Tyagi, K., Pedrioli, P. G. & Trost, M. Hydrophilic strong anion exchange (hSAX) chromatography for highly orthogonal peptide separation of complex proteomes. J. Proteome Res. 12, 2449–2457 (2013).
Gower, C. M. et al. Conversion of a single polypharmacological agent into selective bivalent inhibitors of intracellular kinase activity. ACS Chem. Biol. 11, 121–131 (2016).
Medard, G. et al. Optimized chemical proteomics assay for kinase inhibitor profiling. J. Proteome Res. 14, 1574–1586 (2015).
Cox, J. & Mann, M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 26, 1367–1372 (2008).
Cox, J. et al. Andromeda: a peptide search engine integrated into the MaxQuant environment. J. Proteome Res. 10, 1794–1805 (2011).
Perez-Riverol, Y. et al. The PRIDE database and related tools and resources in 2019: improving support for quantification data. Nucleic Acids Res. 47, D442–D450 (2019).
Tyanova, S. et al. The Perseus computational platform for comprehensive analysis of (prote)omics data. Nat. Methods 13, 731–740 (2016).
Szklarczyk, D. et al. The STRING database in 2017: quality-controlled protein–protein association networks, made broadly accessible. Nucleic Acids Res. 45, D362–D368 (2017).
Shannon, P. et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 13, 2498–2504 (2003).
Eid, S., Turk, S., Volkamer, A., Rippmann, F. & Fulle, S. KinMap: a web-based tool for interactive navigation through human kinome data. BMC Bioinformatics 18, 16 (2017).
Stathias, V. et al. LINCS Data Portal 2.0: next generation access point for perturbation-response signatures. Nucleic Acids Res 48, D431–D439 (2020).
We thank T. Huynh for help with compound plating and A. Vickers and S. Smith for assistance with imaging experiments.
All authors are or were employees of Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., Kenilworth, NJ, USA.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended Data Fig. 1 Proteome-wide mapping of drug-induced protein abundance changes in five cell lines.
a, The heatmap shows the pairwise Pearson correlations (color-coded from 0.95–1) of proteomes from one 96-well plate (the average pairwise Pearson R is 0.983; data is based on 96 biologically independent samples). b, To define MS1 mass-to-charge ratio (m/z) segments, the m/z ion intensity was recorded over a 120 min gradient, summed up and distributed into 1 Da bins (only z > 1 was considered). After calculating the percentage of intensity per 1 Da bin, the m/z range was distributed into segments such that the overall intensity in each segment is equal. The graph shows an example for the distribution of the ion density into three segments (S): S1 = m/z 350–525; S2 = m/z 525–695; S3 = m/z 695–1,200. c, Heatmap showing the number of regulated proteins per inhibitor and cell line (two-sided t-test with permutation-based FDR < 0.05; n = 12 (DMSO) and 3 (each inhibitor) biologically independent samples).
Extended Data Fig. 2 Improving single drug mechanism of action (MoA) deconvolution through pharmaco-proteomic context.
a, Global, drug-induced kinome reprograming visualized in a kinome tree. The bubbles represent individual kinases and their size reflects the number of inhibitor-cell line combinations their abundance was upregulated in (two-sided t-test with permutation-based FDR < 0.05; n = 9 (DMSO) and 3 (inhibitor) biologically independent samples). Among them are many kinases with known roles in lung cancer drug resistance (for example MET, AXL or RAF1). b, Expression of the drug efflux pump ABCB1 was significantly induced by LMK235 (iHDAC4/5) in 2030 cells. ABCB1 is known to bind and export structurally similar compounds (q value derived by permutation-based FDR correction of two-sided t-test derived P values; n = 11 (DMSO) and 3 (iHDAC4/5) biologically independent samples). c, Bar plot showing the number of proteins (two-sided t-test with permutation-based FDR < 0.05; n = 12 (DMSO) and 3 (inhibitor) biologically independent samples) which were exclusively regulated in a single cell line (bars facing down), or consistently across multiple cell lines (bars facing up; color code indicates the number of cell lines). Only the 10 inhibitors with the highest number of regulated proteins across the five cell lines are shown. d, The dot plots depict the log2(FC) of the same protein for different inhibitors (each inhibitor is a dot) in different cell lines (x-axis). The red dot shows Ribociclib (iCDK4/6) which is known to bind to CDK4. Summing up the log2(FC) for the same inhibitor across different cell lines (labeled with “sum”) increased signal-to-noise e, Examples for known drug targets which display outlier characteristics after aggregation of log2 fold changes across cell lines. The dots represent the summed log2(FC) of the same protein (labeled in the header) for different inhibitors. A red dot indicates an inhibitor which is known to bind to the labeled protein (5-FU = 5-fluorouracil).
a, Waterfall plot showing all cell line aggregated log2(FC) outliers (> 3 s.d.; 52 inhibitors; n = 12 (DMSO) and 3 (inhibitor) biologically independent samples) for the PIK3C3 inhibitor PIK-III (ranked by magnitude of protein log2(FC)). The protein SQSTM1 (also known as p62; marked in red), which is a known marker for autophagy modulation was strongly and consistently upregulated in response to PIK-III (iPIK3C3) treatment. b, Indicated cell lines were treated with inhibitors for 24 h (30 μM CQ, 3 μM iALK, 10 μM iCAMK2 or DMSO) and stained for endogenous SQSTM1 (p62, green fluorescence). Cell nuclei were stained using Hoechst (blue). Chloroquine (CQ) and PIK-III (iPIK3C3) served as positive controls. Dots represent extracted, background corrected and normalized intensity values for autophagosome puncta from four individual wells (center line represents the median). c, Chemoproteomic kinase selectivity profiling of iALK and iCAMK2 using pan-kinase beads in A549 cell lysate. EC50 values were derived from a dose–response curve based on eight doses (n = one replicate; four-parameter nonlinear regression fit). The dot plot shows the -log10 EC50 values for kinases competed by iCAMK2 (left) and iALK (right). Red dots represent protein kinase competed by both compounds, whereas grey dots represent unique protein kinases. iCAMK2 also competed six subunits of the PRKA (AMPKA) complex (marked in purple). d, Calu6 cells were treated with increasing doses of the indicated compounds for 16 h and stained for SQSTM1 (p62). Puncta representing autophagosomes were automatically extracted and quantified. PF-562271 is a known inhibitor of the kinases PTK2, PTK2B and FER which are also targets of iALK and iCAMK2. A full list of tested compounds and curve fitting parameters in all five cell lines can be found in Supplementary Data 4.
Extended Data Fig. 4 Using differential proteome changes of pharmacologically related compounds to identify off-target MoAs.
a, Volcano plots comparing iMEK I and iMEK II in A549 and Calu1 cells (q value derived by permutation-based FDR correction of two-sided t-test P values; n = 9 (DMSO for A549), 12 (DMSO for Calu6) and 3 (inhibitors) biologically independent samples; proteins with FDR < 0.1 are shown in black; mitochondrial ribosome subunits are labeled). The mitochondrial translation machinery was strongly downregulated by iMEK II but not iMEK I in A549 and Calu6 cells. b, Aggregation analysis for iMEK I (> 3 s.d.; 53 inhibitors; 5 cell lines; outlier for < 3 compounds). Data is based on n = 12 (DMSO) and 3 (inhibitor) biologically independent samples. Two subunits of the mitochondrial ribosomes are highlighted in red.
Supplementary Figs. 1–3 and Note.
Table 1: Compound names including target/MoA annotation. Table 2: Matrix containing CTG-derived compound EC50 values per inhibitor and cell line and corresponding canonical doses used for proteome profiling. Table 3: Position of compounds on plates per cell line. This annotation can be used to map inhibitors to the individual LFQ values of the datasets in Tables 3–7. Tables 4–8: Protein quantification data and quantification/regulation statistics for individual cell lines A549, Calu1, Calu6, 2030 and 2122. Table 9: Valid values per sample for each cell line. Table 10: Matrix showing the number of regulated proteins per inhibitor and cell line (FDR < 0.05).
Table 1: Count of significant regulation for each protein per cell line and across the entire dataset (two-sided t-test with permutation-based FDR < 0.05; n = 12 (DMSO) and 3 (inhibitor) biologically independent samples). This singles out frequently responding proteins which are less likely to be compound specific. Table 2: Brefeldin A-regulated proteins in 2122 cells (two-sided t-test with permutation-based FDR < 0.05; n = 10 (DMSO) and 3 (brefeldin A) biologically independent samples). Frequent responders are indicated in red (including Reactome pathway enrichment based on 226 frequent responders; statistical framework based on String database v11.0). Specific proteins are colored green (including Reactome pathway enrichment based on 228 proteins remaining after subtraction of frequent responders; statistical framework based on String database v11.0). Table 3: Global aggregation analysis (AA). Individual log2(FC) for a given protein are summed up and compared to the summed log2(FC) of that protein across the remaining 52 inhibitors (n = 12 (DMSO) and 3 (inhibitor) biologically independent samples). Only sum log2(FC) values that deviate three standard deviations from the mean of the sum log2(FC) of the remaining 52 inhibitors are included. The user can filter based on the number of cell lines the protein was found in, the log2(FC) mean across cell lines and on the number of compounds that also show outlier behavior for the respective protein. Table 4: ‘Leave one out’ analysis to quantify the robustness of aggregation analysis. The table shows the R2 value (based on Pearson R correlation) of remaining aggregated log2(FC) after one cell line has been removed (n = 12 (DMSO) and 3 (inhibitor) biologically independent samples). Table 5: Filtered aggregation analysis (AA) for tanespimycin (iHSP90). Tables 6–10: Protein quantification data and quantification/regulation statistics for TL12–186 and TL13–27 in individual cell lines A549, Calu1, Calu6, 2030 and 2122 (two-sided t-test with permutation-based FDR control; n = 4 biologically independent samples). Table 11: Aggregation analysis (AA) for TL12–186 without TL13–27. Table 12: Aggregation analysis (AA) for TL12–186 with TL13–27. Table 13: Aggregation analysis (AA) for indisulam. Tables 14–18: Protein quantification data and quantification/regulation statistics for indisulam in individual cell lines A549, Calu1, Calu6, 2030 and 2122 (two-sided t-test with permutation-based FDR control; n = 4 biologically independent samples).
Table 1: L-1000 dataset annotation (20 overlapping compounds in A549; n = 1 replicate). Table 2: Combined L-1000 dataset in A549. Table 3: Correlation of z-scored abundance changes (536 overlapping genes) for single inhibitors between L-1000 and proteome profiling. Table 4: Combined proteome dataset of overlapping compounds in A549 (>50% valid values). Table 5: Azacitidine mRNA dataset and overlap with other datasets.
Table 1: Dose-resolved kinome selectivity profiling of iCAMK2 using pan-kinase beads and A549 cells (curve fit is based on eight doses; n = 1 replicate). Table 2: Dose-resolved kinome selectivity profiling of iALK using pan-kinase beads and A549 cells (curve fit is based on eight doses; n = 1 replicate). Table 3: Kinome targets of iALK and iCAMK2 including curve fitting parameters (curve fit is based on eight doses; n = 1 replicate). Table 4: Summary table for imaging-based quantification of SQSTM1 (p62) and MAP1LC3B (LC3) puncta in five cell lines A549, Calu1, Calu6, 2030 and 2122. All data have been acquired in a dose-dependent fashion and the table contains the individual curve fitting parameters. Tables 5–14: Extended curve characteristics for p62 and LC3 in individual cell lines.
Table 1: Global Reactome pathway enrichment across cell lines and inhibitors. The underlying protein quantification data are based on n = 12 (DMSO) and 3 (inhibitor) biologically independent samples. For pathway enrichment, GVSE was performed. The final list of proteins used as input for pathway enrichment was n(A549) = 6,738, n(Calu1) = 7,036, n(Calu6) = 7,163, n(2030) = 6,807, n(2122) = 6,092. P values were calculated using two-sided moderated t-statistics and corrected for multiple comparisons using a false discovery rate. Table 2: Dose-resolved kinome selectivity profiling of MEK inhibitors PD184352 and PD0325901 using pan-kinase beads and A549 cells (curve fit is based on eight doses; n = 1 replicate).
About this article
Cite this article
Ruprecht, B., Di Bernardo, J., Wang, Z. et al. A mass spectrometry-based proteome map of drug action in lung cancer cell lines. Nat Chem Biol 16, 1111–1119 (2020). https://doi.org/10.1038/s41589-020-0572-3