Phenotypic assays have become an established approach to drug discovery. Greater disease relevance is often achieved through cellular models with increased complexity and more detailed readouts, such as gene expression or advanced imaging. However, the intricate nature and cost of these assays impose limitations on their screening capacity, often restricting screens to well-characterized small compound sets such as chemogenomics libraries. Here, we outline a cheminformatics approach to identify a small set of compounds with likely novel mechanisms of action (MoAs), expanding the MoA search space for throughput limited phenotypic assays. Our approach is based on mining existing large-scale, phenotypic high-throughput screening (HTS) data. It enables the identification of chemotypes that exhibit selectivity across multiple cell-based assays, which are characterized by persistent and broad structure activity relationships (SAR). We validate the effectiveness of our approach in broad cellular profiling assays (Cell Painting, DRUG-seq, and Promotor Signature Profiling) and chemical proteomics experiments. These experiments revealed that the compounds behave similarly to known chemogenetic libraries, but with a notable bias toward novel protein targets. To foster collaboration and advance research in this area, we have curated a public set of such compounds based on the PubChem BioAssay dataset and made it available for use by the scientific community.

Robust and reliable proteome measurements provide mechanistic insights in biomedical research. SOMAmer (Slow Off-rate Modified Aptamer) reagents are modified, DNA-based, affinity reagents that measure defined target proteins with reproducibility and accuracy similar to monoclonal antibodies. Applying SOMAmer reagent technology, we developed SomaScan, a clinical proteome profiling platform with capability to measure 7,523 proteoforms for 6,594 human proteins by UniprotID in small sample volumes (e.g., 55μl plasma or serum). We evaluated the platform by profiling the proteome of a panel of well characterized Cell Line Encyclopedia (CCLE) cancer models. Unsupervised machine learning analyses demonstrate the SomaScan assay distinguishing cell lines on the basis of their proteome signatures, and identifying both tissue-specific and oncogenic pathways. The proteome measured by SomaScan correlates with published CCLE transcriptome at a level comparable to other published transcript to proteome studies. Taken together, we demonstrate that the SomaScan platform is a technically reproducible system suitable for biomedical and clinical applications that reliably illuminates underlying biomolecular mechanisms.### Competing Interest StatementFunding and Conflict of Interest: This research was resourced as part of a Novartis-Somalogic multiyear collaboration and all authors were employees of Novartis or Somalogic at the time of their contribution to this work.

p-ERK MSD results in RAS mutant cells ectopically expressed ARAF variants

Proliferation IC50 values for various cell lines and cell line derivatives

Supplementary Data from Exquisite Sensitivity to Dual BRG1/BRM ATPase Inhibitors Reveals Broad SWI/SNF Dependencies in Acute Myeloid Leukemia

Identifying high quality chemical starting points is a critical and challenging step in drug discovery, which typically involves screening large compound libraries or repurposing of compounds with known mechanisms of actions (MoAs). Here we introduce a novel cheminformatics approach that mines existing large-scale, phenotypic high throughput screening (HTS) data. Our method aims to identify bioactive compounds with distinct and specific MoAs, serving as a valuable complement to existing focused library collections. This approach identifies chemotypes with selectivity across multiple cell-based assays and characterized by persistent and broad structure activity relationships (SAR). We prospectively demonstrate the validity of the approach in broad cellular profiling assays (cell painting, DRUG-seq, Promotor Signature Profiling) and chemical proteomics experiments where the compounds behave similarly to known chemogenetic libraries, but with a bias towards novel protein targets and required no synthetic effort to improve compound properties. A public set of such compounds is provided based on the PubChem BioAssay dataset for use by the scientific community.

In vivo anti-tumor effects of LXH254 and Trametinib

Supplementary Figure 2. Western blot analysis of CRISPR-CAS9 cell lines and exogenously expressed ARAF variants used in experiments shown in Figures 3-7.

Growth inhibition IC50 data of LXH254 in CCLE panel

Proliferation and p-ERK IC50 results in RAS-mutant cells expressing only one RAF paralog

Results of Kinativ studies using LXH254 in HCT 116 cells

Table S9. Differential gene analysis of LXH254 sensitive and insensitive cells lines

Supplementary Figure 5. Model for LXH254 interactions with RAF monomers and dimers (A). (B) Upper panel, simulated structure of ARAF (green) overlaid on the structure of BRAF (blue). LXH254 interacting with ARAF is shown in yellow and with BRAF in purple. (B) Lower panel, enhanced view of hydrogen bond interactions between LXH254 (purple) and the backbone carbonyl of F595 of BRAF (Blue) and LXH254 (yellow) and backbone carbonyl of ARAF (green).

Supplementary Figure 3. (A) Shown are western blots using antibodies directed against T202/Y204 phosphorylated ERK1/2 following incubation with a range of concentrations of LXH254 over a 4 - 72 hour period. Which individual RAF protein is not expressed in each HCT 116 variant is indicated above the relevant panels. (B) Shown are western blots using Abs directed against the indicated proteins in either whole cell lysates (WCL) or material immuno-precipitated from cell lysates using Abs directed against ARAF (ARAF IP) or BRAF (BRAF IP). Lysates were generated from either parental MIA PaCa-2 cells or a MIA PaCa-2 variant lacking expression of CRAF. (C) Recombinant MEK1 can be phosphorylated by immuno-precipitates from both LXH254 treated and untreated MIA PaCa-2 and MEL-JUSO cells using an anti-body directed against ARAF contains MEK1 kinase activity. Proteins detected in either whole cell lysates (WCL) or ARAF-Ab immuno-precipitates (IP:ARAF) are indicated on the right. (D) IP-kinase activity on recombinant MEK1 in immuno-precipitates from HEK293 cells expressing FLAG-tagged ARAF using Abs directed against FLAG requires ATP and occurs on both wild type (WT) and kinase-dead (K97M) MEK1. (E) Shown are western blots using antibodies directed against T202/Y204 phosphorylated ERK1/2 following incubation with a range of concentrations of LXH254 over a 4-72 hour period. The individual RAF protein expressed in each HCT 116 variant is indicated above the relevant panels. (F) Shown are western blots of phosphorylated MEK1/2 (S218/S221) and ERK1/2 (T202/Y204) following a 4 hr. incubation with a range of LXH254 concentrations in the RAS/RAF wild type cell line PC-9.

Supplementary Figure 1. Shown in (A) are the IC50 values for the inhibition of proliferation for BRAF ( ), KRAS ( ), and NRAS ( ) mutant as well as "WT" ( ) cell lines. Below the graph Fischer's exact test using 1, 2, and 2.5ï­M sensitivity cutoffs are provided. (B) Comparison of cell line sensitivities for LXH254 and RAF709. LXH254 values are provided in this manuscript (Sup. Table 3) and RAF709 values obtained from (25). (C) mRNA expression (RNASeq, https://portals.broadinstitute.org/ccle/data) for cell lines presented in (A) with the 2.0ï­M LXH254 sensitivity threshold delineated. (D) ARAF mRNA expression in KRAS mutant and wild type cells from TCGA Pan-cancer Atlas Project (47), using the cBIO cancer genomics portal (https://www.cbioportal.org/, (1,2))

Supplementary Figure 4. Growth of individual tumors shown in Fig. 7A-C are given for HCT 116 (A), MIA PaCa-2 (B) and MEL-JUSO (C) cells.

Growth inhibition IC50 values (nM) of several RAF inhibitors in a panel of cell lines

Abstract Various subunits of mammalian SWI/SNF chromatin remodeling complexes display loss-of-function mutations characteristic of tumor suppressors in different cancers, but an additional role for SWI/SNF supporting cell survival in distinct cancer contexts is emerging. In particular, genetic dependence on the catalytic subunit BRG1/SMARCA4 has been observed in acute myelogenous leukemia (AML), yet the feasibility of direct therapeutic targeting of SWI/SNF catalytic activity in leukemia remains unknown. Here, we evaluated the activity of dual BRG1/BRM ATPase inhibitors across a genetically diverse panel of cancer cell lines and observed that hematopoietic cancer cell lines were among the most sensitive compared with other lineages. This result was striking in comparison with data from pooled short hairpin RNA screens, which showed that only a subset of leukemia cell lines display sensitivity to BRG1 knockdown. We demonstrate that combined genetic knockdown of BRG1 and BRM is required to recapitulate the effects of dual inhibitors, suggesting that SWI/SNF dependency in human leukemia extends beyond a predominantly BRG1-driven mechanism. Through gene expression and chromatin accessibility studies, we show that the dual inhibitors act at genomic loci associated with oncogenic transcription factors, and observe a downregulation of leukemic pathway genes, including MYC, a well-established target of BRG1 activity in AML. Overall, small-molecule inhibition of BRG1/BRM induced common transcriptional responses across leukemia models resulting in a spectrum of cellular phenotypes. Implications: Our studies reveal the breadth of SWI/SNF dependency in leukemia and support targeting SWI/SNF catalytic function as a potential therapeutic strategy in AML.

Abstract Purpose: Targeting RAF for antitumor therapy in RAS-mutant tumors holds promise. Herein, we describe in detail novel properties of the type II RAF inhibitor, LXH254. Experimental Design: LXH254 was profiled in biochemical, in vitro, and in vivo assays, including examining the activities of the drug in a large panel of cancer-derived cell lines and a comprehensive set of in vivo models. In addition, activity of LXH254 was assessed in cells where different sets of RAF paralogs were ablated, or that expressed kinase-impaired and dimer-deficient variants of ARAF. Results: We describe an unexpected paralog selectivity of LXH254, which is able to potently inhibit BRAF and CRAF, but has less activity against ARAF. LXH254 was active in models harboring BRAF alterations, including atypical BRAF alterations coexpressed with mutant K/NRAS, and NRAS mutants, but had only modest activity in KRAS mutants. In RAS-mutant lines, loss of ARAF, but not BRAF or CRAF, sensitized cells to LXH254. ARAF-mediated resistance to LXH254 required both kinase function and dimerization. Higher concentrations of LXH254 were required to inhibit signaling in RAS-mutant cells expressing only ARAF relative to BRAF or CRAF. Moreover, specifically in cells expressing only ARAF, LXH254 caused paradoxical activation of MAPK signaling in a manner similar to dabrafenib. Finally, in vivo, LXH254 drove complete regressions of isogenic variants of RAS-mutant cells lacking ARAF expression, while parental lines were only modestly sensitive. Conclusions: LXH254 is a novel RAF inhibitor, which is able to inhibit dimerized BRAF and CRAF, as well as monomeric BRAF, while largely sparing ARAF.

Abstract Cellular senescence is a stress-induced state of stable growth arrest characterized by high expression of cell cycle inhibitors; a dramatic change in cell morphology, including an increase in lysosomal content; and secretion of large numbers of proteins involved in immune signaling and extracellular matrix remodeling. The physiological importance of cellular senescence has been attributed to prevention of carcinogenesis, aging, development, and tissue repair, and tumor cells can undergo senescence in response to therapeutic agents. In this work, we sought to validate the senescence-inducing activity of two known inducers (doxorubicin and TGFβ1) and a CDK4/6 inhibitor (ribociclib) and to identify proteins that can kill senescent tumor cells (senolytic targets) if knocked out and to identify downstream components of the tumor cell senescence pathway. Huh7 hepatocellular carcinoma cells and SK-MEL-28 melanoma cells were induced to senescence by treatment with three different agents: ribociclib, low doses of doxorubicin, or TGFβ1. The induction of senescence was confirmed by observing growth arrest, an increase in SA-β-gal staining, dramatic cell morphology changes, loss of c-Myc protein, increased expression of p15, and increased expression of senescence-associated secretory phenotype (SASP) proteins. Induction of SASP components was measured by RNAseq and SOMAscan. All three agents induced a senescent state, with blockage at different stages of the cell cycle observed. Induction of known immune factors, including IL-8 and IL-11, were identified in senescent cells (Huh7). A whole-genome CRISPR screen identified proteins required to enter senescence and those that were incompatible with the senescent state if knocked out. Expected hits were observed (eg, TGFBR1/TGFBR2 for TGFβ1, RB for ribociclib, and TOP2A for doxorubicin) for guide DNAs (gDNAs) that blocked entry into the senescent state. No gDNA candidates for common downstream senescence pathway components were observed, suggesting that these components are essential genes or that they do not exist. gDNA-induced knockouts that were incompatible with the senescent state dropped out of the screen and represent potential senolytic targets. The screen identified BCL2L1 as the only common senolytic hit across multiple senescence-inducing reagents, confirming published reports suggesting it is a senolytic target. These data show that ribociclib, doxorubicin, and TGFβ1 induced senescence in cancer cell lines. Whole-genome CRISPR screens identified senescence pathway components for each of these agents, as well as a common senolytic target. Citation Format: Pasupuleti Rao, Jennifer Tullai, Peter Aspesi, Felipa Mapa, Nadire Cochran, Frederic Sigoillot, Guglielmo Roma, Scott Gleim, Jaison Jacob, Jason Marchese, Jonathan Solomon. Characterization of cancer cell lines made senescent by exposure to ribociclib, doxorubicin, or TGFβ1, and identification of genes required for entry into senescence and senescent cell survival [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2025.