Mutations in the small GTPase protein KRAS are one of the leading drivers of cancers including lung, pancreatic, and colorectal, as well as a group of developmental disorders termed “Rasopathies”. Recent breakthroughs in the development of mutant-specific KRAS inhibitors include the FDA approved drug Lumakras (Sotorasib, AMG510) for KRAS G12C-mutated non-small cell lung cancer (NSCLC), and MRTX1133, a promising clinical candidate for the treatment of KRAS G12D-mutated cancers. However, there are currently no FDA approved inhibitors that target KRAS mutations occurring at non-codon 12 positions. Herein, we focused on the KRAS mutant A146T, found in colorectal cancers, that exhibits a “fast-cycling” nucleotide mechanism as a driver for oncogenic activation. We developed a novel high throughput time-resolved fluorescence resonance energy transfer (TR-FRET) assay that leverages the reduced nucleotide affinity of KRAS A146T. As designed, the assay is capable of detecting small molecules that act to allosterically modulate GDP affinity or directly compete with the bound nucleotide. A pilot screen was completed to demonstrate robust statistics and reproducibility followed by a primary screen using a diversity library totaling over 83,000 compounds. Compounds yielding >50% inhibition of TR-FRET signal were selected as hits for testing in dose-response format. The most promising hit, UNC10104889, was further investigated through a structure activity relationship (SAR)-by-catalog approach in an attempt to improve potency and circumvent solubility liabilities. Overall, we present the TR-FRET platform as a robust assay to screen fast-cycling KRAS mutants enabling future discovery efforts for novel chemical probes and drug candidates.
The small GTPase KRAS acts as a molecular signaling switch, cycling between a GDP-bound “off,” and a GTP-bound “on” state. The core G-domain of KRAS includes five consensus motifs that mediate nucleotide binding, with the P-loop (residues 10 – 17), NKXD (residues 113 – 120) and SAK (residues 145 – 147) motifs providing important interactions for nucleotide affinity and specificity. The switch-I (residues 25 – 40) and switch-II (residues 57 – 74) regions are conformationally responsive to the γ-phosphate of GTP [
]. In addition, KRAS contains a C-terminal “hyper-variable” tail which undergoes lipidation and other modifications important for membrane localization and cellular activity [
- Vetter I.R.
- Wittinghofer A.
The guanine nucleotide-binding switch in three dimensions.
Science. 2001; 294: 1299-1304
]. In normal cells, KRAS predominantly exists in the inactive GDP-bound state. Upon extracellular signaling, KRAS is stimulated by guanine nucleotide exchange factors (GEFs), which catalyze the release of bound GDP, allowing for GTP, which exists at a higher cellular concentration, to bind. Once GTP-bound, activated KRAS engages with multiple downstream effector proteins such as BRAF, PI3 kinase, and RalGDS. GTPase activating proteins (GAPs) return KRAS to its inactive, GDP-bound state by accelerating the slow, intrinsic GTP hydrolysis reaction [
- Haidar M.
- Jacquemin P
Past and future strategies to inhibit membrane localization of the KRAS oncogene.
Int J Mol Sci. 2021; 22
]. When properly tuned, KRAS signaling helps drive cell cycle progression, growth and metabolism, and motility [
- Simanshu D.K.
- Nissley D.V.
- McCormick F.
RAS proteins and their regulators in human disease.
Cell. 2017; 170: 17-33
- Mukhopadhyay S.
- Vander Heiden M.G.
- McCormick F.
The metabolic landscape of RAS-driven cancers from biology to therapy.
Nat Cancer. 2021; 2: 271-283
KRAS plays a role in multiple cellular signaling pathways, and predictably, KRAS mutation leads to a variety of cancers. The RAS (HRAS, NRAS, and KRAS) family comprise the most frequently mutated oncogene family, occurring in ∼25% of all cancers [
]. Approximately 75% of these mutations are found in the KRAS isoform, where they are implicated in cancers such as pancreatic, lung, and colon [
- Huang L.
- Guo Z.
- Wang F.
- Fu L.
KRAS mutation: from undruggable to druggable in cancer.
Signal Transduct Target Ther. 2021; 6: 386
- Prior I.A.
- Hood F.E.
- Hartley J.L
The frequency of ras mutations in cancer.
Cancer Res. 2020; 80: 2969-2974
]. Activating mutations in KRAS are typically found in one of five residue hotspots: G12, G13, Q61, K117, and A146 [
- Siegel R.L.
- Miller K.D.
- Fuchs H.E.
- Jemal A.
Cancer statistics, 2022.
CA Cancer J Clin. 2022; 72: 7-33
- Hobbs G.A.
- Der C.J.
RAS mutations are not created equal.
Cancer Discov. 2019; 9: 696-698
]. These oncogenic mutations cause KRAS to become populated in the GTP-bound state through two prevalent mechanisms: 1) mutations that prevent the ability of GAPs to catalyze GTP hydrolysis in KRAS, 2) or those that accelerate intrinsic nucleotide exchange (the so-called “fast-cyclers”) [
- Hobbs G.A.
- Der C.J.
- Rossman K.L
RAS isoforms and mutations in cancer at a glance.
J Cell Sci. 2016; 129: 1287-1292
- Johnson C.
- Burkhart D.L.
- Haigis K.M.
Classification of KRAS-activating mutations and the implications for therapeutic intervention.
Cancer Discov. 2022; 12: 913-923
Even though KRAS has been a well validated oncology target for 50 years, drugs inhibiting KRAS have only recently been developed. Challenges to KRAS drug discovery efforts include its picomolar affinity for guanine nucleotides [
], the high intracellular concentration of these guanine nucleotides [
- John J.
- et al.
Kinetics of interaction of nucleotides with nucleotide-free H-ras p21.
Biochemistry. 1990; 29: 6058-6065
], and the lack of a deep drug-like pocket on its surface. Due to the challenges presented above, it wasn't until 2013 that using small molecules to directly inhibit KRAS appeared to be possible. This was shown when a disulfide fragment screen produced compounds that were able to covalently bind the mutant cysteine on KRAS G12C [
- Traut T.W.
Physiological concentrations of purines and pyrimidines.
Mol Cell Biochem. 1994; 140: 1-22
- Lim S.M.
- et al.
Therapeutic targeting of oncogenic K-Ras by a covalent catalytic site inhibitor.
Angew Chem Int Ed. 2014; 53: 199-204
]. These molecules contain an electrophilic warhead that forms a covalent bond with the nucleophilic thiol on the substituted cysteine residue. The demonstration that KRAS G12C could be covalently targeted led to a rapid improvement of new small molecule inhibitors. These small molecules were found to occupy a previously unknown pocket, termed the switch-II pocket (SIIP), and lock KRAS in the inactive GDP state. Eventually, these discoveries led to the development of AMG510 (Sotorasib), which received FDA approval in 2021 for the treatment of KRAS G12C driven non-small cell lung cancer (NSCLC) [
- Ostrem J.M.
- Peters U.
- Sos M.L.
- Wells J.A.
- Shokat K.M.
K-Ras(G12C) inhibitors allosterically control GTP affinity and effector interactions.
Nature. 2013; 503: 548-551
- Lanman B.A.
- et al.
Discovery of a covalent inhibitor of KRASG12C (AMG 510) for the treatment of solid tumors.
J Med Chem. 2020; 63: 52-65
- Punekar S.R.
- Velcheti V.
- Neel B.G.
- Wong K.-K.
The current state of the art and future trends in RAS-targeted cancer therapies.
Nat Rev Clin Oncol. 2022; 19: 637-655
The KRAS G12C inhibitors are an exciting advance in the field, however, these molecules are not as potent or compatible against KRAS mutants lacking a targetable cysteine, many of which are more commonly found in cancer than G12C. Most recently, MRTX1133, a highly potent (picomolar KD) KRAS G12D inhibitor [
] has been developed. MRTX1133 also binds the SIIP of KRAS G12D with ∼1000-fold selectivity over wild type KRAS and has been shown to reduce tumor volume in vivo. However, even with the recent G12C and G12D inhibitor successes there has been less focus on developing drugs that target KRAS mutations at non-G12 sites.
- Wang X.
- et al.
Identification of MRTX1133, a noncovalent, potent, and selective KRASG12D inhibitor.
J Med Chem. 2022; 65: 3123-3133
The fast-cycling mutants of KRAS, such as G13D, K117N and A146T, represent a relatively unexplored opportunity for KRAS drug discovery. Due to their reduced nucleotide affinity and rapid cycling of nucleotides, they undoubtedly sample broad conformational space as compared to KRAS mutants with wild-type GDP/GTP affinity [
]. This raises the possibility of discovering novel small molecule probes that may stabilize these mutants in inactive conformations. The recent crystal structure of KRAS A146T-GDP supports this hypothesis, where the mutation results in a dramatic repositioning and structuring of switch-I, expulsion of the magnesium ion and a two-residue translation of β−strand 3 relative to its position in KRAS WT [
- Pantsar T.
The current understanding of KRAS protein structure and dynamics.
Comput Struct Biotechnol J. 2020; 18: 189-198
], thus providing a mechanism for its reduced nucleotide affinity. In addition to presenting novel, druggable pockets on KRAS, the reduced nucleotide affinity of KRAS A146T may render it susceptible to inhibition by small molecule GDP-competitors. To this end, we have developed a novel screening assay for KRAS A146T to support the discovery of first-in-class small molecule modulators.
- Poulin E.J.
- et al.
Tissue-specific oncogenic activity of KRASA146T.
Cancer Discov. 2019; 9: 738-755
Herein we describe the development of a time-resolved fluorescence resonance energy transfer (TR-FRET) assay designed to identify small molecule modulators of nucleotide binding to the fast-cycler KRAS A146T [
]. Use of the TR-FRET assay has several advantages including: 1) enabling rapid completion of a small molecule high-throughput screening (HTS) campaign in a time-resolved format, 2) maintaining a robust and reproducible signal-to-noise readout, and 3) reducing protein and ligand concentrations thus allowing for a low tight-binding limit and reducing reagent consumption. The TR-FRET assay format was validated with control ligands, including GDP, and a pilot screen run in duplicate. Completion of the primary screen with >83,000 compounds identified several putative small molecule hits that produced dose-response inhibition in the low micromolar range. To further explore the top hit compound, we conducted a small SAR study that yielded only a modest increase in potency. During extensive validation studies with orthogonal and biophysical assays, we noted poor solubility issues and confirmed that solubility improvement did not correlate with potency in the TR-FRET assay. We conclude that the hit series from the primary screening completed to date is not tractable for further medical chemistry hit-to-lead efforts due to the solubility liabilities. However, efforts reported here can enable future and more extensive screening campaigns for discovery of novel fast-cycling KRAS inhibitors.
- Janakiraman M.
- et al.
Genomic and biological characterization of exon 4 KRAS mutations in human cancer.
Cancer Res. 2010; 70: 5901-5911
Materials and methods
The LANCE Eu-W1024 streptavidin conjugate was purchased from PerkinElmer. Tris-base, Tween-20, GDP, 1,4-Diothio-DL-threitol (DTT), magnesium chloride hexahydrate (MgCl2) were purchased from Sigma-Aldrich. EDA-CY5-GDP was purchased from Jena Bioscience. Assay and dilution plates were obtained from Grenier Bio-One (cat. 784904 and 781280 respectively). The LOPAC1280 library was purchased from Sigma-Aldrich. The 57K DIVERSet-CL library was purchased from ChemBridge. The Nucleotide Compound Library was purchased from MedChemExpress. The 100k diversity and 5k kinase libraries are stored and managed within the Center for Integrative Chemical Biology and Drug Discovery (CICBDD) at UNC-Chapel Hill and were originally purchased from Enamine. Retested compounds were purchased from Molport, Life Chemicals, and MedChemExpress.
Protein expression, purification, and biotinylation
A bacterial expression construct for Avi-tagged KRAS was assembled by fusing the avi-tag coding sequence (GLNDIPEAQKIEWHE) in frame with the 3’ end of KRAS4b 1-185 A146T/C185S in pProEx HTb (Invitrogen), in which the His6-tag was removed. Avi-KRAS A146T/C185S was expressed in Rosetta2 BL21(DE3) E. coli cells. Cell cultures were grown in TB/AMP/CHL at 37 °C to an OD600 of 0.6 and then induced with 0.5 mM IPTG overnight with continuous shaking at 22 °C. Bacterial cell pellets were resuspended in S1 buffer (10 mM sodium phosphate pH 6.5, 10 mM NaCl, 2 mM MgCl2, 2 mM DTT, 10 µM GDP and 2% glycerol) then lysed using an Emulsiflex C5 homogenizer (Avestin). Lysates were clarified by centrifugation at 40,000 x g at 4°C for 30 min and loaded onto a HiPrep SP HP 16/10 (Cytiva) cation exchange column equilibrated in S1 buffer. Avi-tagged KRAS was eluted with a 0 – 100% S2 (S1 + 500 mM NaCl) linear gradient and KRAS-containing fractions were concentrated and loaded onto a HiLoad 16/600 Superdex 200 pg (Cytiva) size exclusion column equilibrated in S200 buffer (20mM Tris pH 8.0, 200 mM NaCl, 2 mM MgCl2, 2 mM DTT and 5% glycerol). The protein was then concentrated to 3 mg/ml using an Amicon stirred-cell (Millipore), snap frozen in liquid nitrogen, and stored at -80⁰C. Avi-tagged protein was biotinylated in vitro using the BirA500 biotin-protein ligase kit (Avidity). Briefly, BiomixA, BiomixB, avi-tagged KRAS A146T at 40 µM final concentration, and 4 µg His-tagged BirA biotin-protein ligase were mixed in an Eppendorf tube at 1 mL volume. The reaction was incubated for 1 hour at 30°C. His-tagged BirA protein was removed after incubation by His-Tag isolation Dynabeads (ThermoFisher). A Nap-5 spin column (Cytiva) was used to desalt and buffer exchange the biotinylated-KRAS protein which was snap frozen in liquid nitrogen and stored at -80°C.
Guanine nucleotide exchange assays
KRAS WT and KRAS A146T/C185S were expressed and purified as above, then loaded with BODIPY FL-GDP (Jena). For this, 0.2 mM KRAS was incubated in a solution containing 1 mM BODIPY FL-GDP, 20 mM HEPES pH 7.6, 10 mM EDTA, 20 mM ammonium sulfate and 1 mM DTT for 1 hr at 4°C. The reaction was stopped with the addition of 20 mM MgCl2 and excess nucleotide was removed on a 5 ml HiTrap Desalting column (Cytiva) equilibrated in 10 mM HEPES pH 7.6, 100 mM NaCl, 2 mM MgCl2, 2 mM DTT and 10% glycerol. Guanine nucleotide exchange reactions were performed on an Agilent Cary Eclipse spectrometer in assay mixtures containing 20 mM HEPES pH 7.4, 100 mM NaCl, 2 mM MgCl2 and 1 mM DTT. Reaction mixtures were allowed to equilibrate to 20°C with continuous stirring before the addition of 0.2 µM KRAS-bGDP. Nucleotide exchange was initiated by the addition of 100 µM GTP (Sigma) while fluorescence intensity was monitored (λex = 488 nm, λem = 512 nm). Reactions were performed in duplicate. Data were plotted in GraphPad Prism 9.4.1 and fit as single exponential decays to yield the rate (kobs) of nucleotide exchange
General TR-FRET assay conditions
A stock solution of 10X KRAS TR-FRET buffer (200 mM Tris pH 7.5, 1500 mM NaCl, 100 mM MgCl2, and 0.05% Tween 20) was prepared, 0.2 µm filtered, stored at room temperature, and was used throughout. Assays were completed using freshly made 1X KRAS TR-FRET buffer containing 20 mM Tris pH 7.5, 150 mM NaCl, 10 mM MgCl2, 0.005% Tween 20, and 1 mM DTT. 384-well microplates (Greiner, #784904) were used for assay development and screening with a total assay volume of 10 µL. 384-well polypropylene plates (Greiner, #781280) were used for compound serial dilutions. For compounds with stock solutions in water, serial dilutions were made using KRAS TR-FRET buffer. For compounds stored in DMSO, serial dilutions were made using DMSO. Following addition of all assay components (5 nM biotin-KRAS A146T, 1 nM Eu-labeled streptavidin, and 3 nM Cy5-GDP), plates were sealed with clear covers, gently mixed on a tabletop shaker for 1 minute, centrifuged at 1000 x g for 2 minutes, and allowed to equilibrate in a dark space for one hour before reading. Measurements were taken on an EnVision® 2103 Multilabel Plate Reader (Perkin Elmer) using an excitation filter at 320 nm and emission filters at 615 nm and 665 nm. 615 nm and 650 nm emission signals were measured simultaneously using a dual mirror at D400/D630. TR-FRET output signal was expressed as emission ratios of acceptor/donor (665/615 nm) counts. Percent inhibition was calculated on a scale of 0% (i.e., activity with DMSO vehicle only) to 100% (1 µM GDP) using full column of controls on each plate. The interquartile mean of control wells was used to calculate Z-prime (Z’). For dose-response curves, data were fit with a four-parameter nonlinear regression analysis using GraphPad Prism 7.0 or ScreenAble software to obtain IC50 values.
TR-FRET assay optimization
A 10 point, two-fold serial dilution was prepared separately for both Eu-labeled streptavidin and Cy5-GDP, TR-FRET donor and acceptor labels respectively, in a 384, deep-well polypropylene plate using KRAS TR-FRET buffer. Solutions were made to 10x the final concentration and typically started at 100 nM for Eu-labeled streptavidin and 500 nM Cy5-GDP. Upon addition to the plate, the highest final concentrations were typically 10 nM and 50 nM, respectively. The tenth well of each dilution contained only buffer for background signal information. A 1.25X solution of biotinylated-KRAS A146T was prepared in the KRAS TR-FRET buffer for a final concentration of 5 nM in the well. 1 µL of Eu-labeled streptavidin was added to columns across the assay plate followed by addition of 1 µL of Cy5-GDP to the rows of the assay plate. Lastly, 8 µL of the protein solution was added to all the wells.
Z′, DMSO tolerance
Full columns of high signal (no competitor, DMSO only) and low signal (100% inhibition with competitor compound; 1 µM GDP) were produced to calculate the Z’. To complete a DMSO tolerance test, the high and low signals were obtained using concentrations of 0, 1, 2, 3, 4, and 5% DMSO. A 10 µM solution of GDP was prepared using KRAS TR-FRET buffer. Using an automated multi-channel pipette, 1 μL of the competitor was added to five columns of the plate to be read for the low signal. Likewise, 1 μL of KRAS TR-FRET buffer was transferred to five columns for the high signal measurement. 10X solutions of DMSO were prepared using KRAS TR-FRET buffer to 0, 10, 20, 30, 40, and 50% DMSO. 1 μL of the DMSO solutions were transferred to both high and low signal columns measurement appropriately. A mixture of biotin-KRAS A146T, Eu-labeled streptavidin, and Cy5-GDP was made into a single tube. The solution was gently mixed and then aliquoted appropriately to a new column in a deep-well plate. Using an automated multichannel pipette, 8 µL of the mixture was added to the assay plate.
A 16 point, three-fold serial dilution of compound (GDP) was prepared using KRAS TR-FRET buffer in a 384, deep-well polypropylene plate. Using an automated multichannel pipette, 1 µL of the compound dilution was added to three columns of the assay plate. A mixture of biotin-KRAS A146T, Eu-labeled streptavidin, and Cy5-GDP was made into a single tube. The solution was gently mixed and then aliquoted to a new column in the deep-well plate. Using an automated multi-channel pipette, 9 µL of mixture was transferred from the deep-well plate to the assay plate.
Assay format and run
For testing compounds in higher throughput, 384-well assay ready plates were prepared in standard plate format: column 1 was used for high signal control (DMSO only), column 2 was used for a low-to-high control (0 to 100% inhibition with competitor compound), column 23 was used for a mid-point control (50% inhibition with competitor compound), column 24 was used for low signal (100% inhibition with competitor compound), and columns 3–22 were used for 20 µM single-dose test compounds. Chemical library sets maintained in the UNC Center for Integrative Chemical Biology and Drug Discovery (CICBDD) were used. First, controls were added to a mother plate where column 1 contained DMSO for a 1% final concentration, column 2 contained a 16 point, three-fold serial dilution of compound (GDP) prepared using KRAS TR-FRET buffer starting at 1 mM, column 23 contained 3 µM GDP in KRAS TR-FRET buffer, and column 24 contained 100 µM GDP in KRAS TR-FRET buffer. Test compounds were dispensed across the mother plate at 100X (2 mM) concentration in columns 3–22 using a TECAN Freedom EVO liquid handling workstation. Using a TTP Labtech Mosquito® HTS liquid handling instrument, assay ready plates were stamped with 100 nL of DMSO into columns 1, 100 nL of mother plate compounds into columns 3–22, and 100 nL of control compounds into columns 2, 23, and 24. Assay ready plates were stored at -20°C until ready for use. 5 nM biotin-KRAS A146T, 1 nM Eu-labeled streptavidin, and 3 nM Cy5-GDP were combined and gently mixed by pipetting and rocking. 10 µL of this mixture was then added to each well of a thawed assay ready plate at room temperature using a Multidrop Combi (ThermoFisher). Percent inhibition was calculated on a scale of 0% (i.e., activity with DMSO vehicle only) to 100% (1 µM GDP) from the full column controls on each plate.
To test hit compounds selected for follow-up in dose-response curves, assay ready plates were formatted and prepared similar to the single-dose test plates described above. Columns 3–12 and 13–22 were used for 10-point serial dilutions of test compounds. For mother plate preparations, test compounds were serially diluted in DMSO 3X across the plate at 100X concentration using a TECAN Freedom EVO liquid handling workstation. The top concentration was 10 mM. Using a TTP Labtech Mosquito® HTS liquid handling instrument, assay ready plates were stamped with 200 nL of compound solutions from the mother plate. Biotin-KRAS A146T, Eu-labeled streptavidin, and Cy5-GDP were added together and then mixed gently by pipetting and rocking. 10 µL was added to each well of an assay ready plate using a Multidrop Combi (ThermoFisher).
10 mM DMSO samples of hit compounds were analyzed for content and solubility using chemiluminescence nitrogen detection (CLND) as part of the kinetic solubility protocol by Analiza Inc. (Cleveland, OH). Compounds were diluted to 200 µM into 1X PBS, pH 7.4 and 2% DMSO. Samples were then incubated at 23° C for 24 hours and then vacuum filtered where the filtrate then underwent nitrogen detection analysis and concentration was reported in µg/mL.
Fluorescent polarization (FP) assay
Briefly, a 2D titration was performed and ∼10-fold assay window was seen using 250 nM KRAS A146T and 20 nM Cy5-GDP (data not shown). Assay ready plates were formatted and prepared as described in the dose-response validation protocol. KRAS A146T and Cy5-GDP were prepared in KRAS TR-FRET buffer and 10 µL was added to each well of an assay ready plate using a Multidrop Combi. Following a 30-minute incubation, fluorescence polarization was measured using an EnVision plate reader, set up with 620 nM excitation and 688 nM S and P emission filters with a D658 FP–D688 dichroic mirror. The S and P values were processed with the standard FP calculation formula (mP = 1000 * [S – G * P] / [S + G * P], where G is the G-factor and is approximately 1).
Development of a GDP-displacing TR-FRET assay for KRAS A146T
TR-FRET is a well-established platform for high-throughput screening (HTS), hit validation, and structure-activity relationship (SAR) studies [
]. We utilized this technology to develop an HTS-compatible TR-FRET assay to screen for small-molecule compounds that may potentially interfere with GDP/GTP exchange of the fast-cycling KRAS mutant, A146T [
- Rectenwald J.M.
- et al.
A general TR-FRET assay platform for high-throughput screening and characterizing inhibitors of Methyl-Lysine reader proteins.
SLAS Discov. 2019; 24: 693-700
]. KRAS A146T exchanges GDP for GTP ∼100-fold faster than KRAS WT, rendering KRAS A146T active independently of RAS GEF stimulation (Fig. 1A). Due to the fast-cycling nature of the KRAS A146T mutant compared to KRAS WT, we reasoned that we could develop a TR-FRET-based nucleotide displacement assay. To establish a general TR-FRET method for screening KRAS A146T, we used europium (Eu)-streptavidin as the fluorescence donor and Cy5-GDP as the acceptor. Energy transfer between the donor and acceptor probes occurs upon formation of the Eu-streptavidin/KRAS/Cy5-GDP complex (Fig. 1B). The major advantage of TR-FRET over other fluorescence-based assays is time-resolved detection afforded by the use of the lanthanide europium donor, with a hundreds-of-millisecond emission time. Taking a measurement of the emission with a delayed time window reduces signal interference that may be caused by background fluorescence from the assay plate, buffers, and fluorescent compounds which typically decay in the low-nanosecond timescale. The ratio-metric calculation of acceptor emission to donor emission also reduces inter-well variations of signal [
- Poulin E.J.
- et al.
Tissue-specific oncogenic activity of KRASA146T.
Cancer Discov. 2019; 9: 738-755
- Glickman J.F.
- et al.
Assay development for protein kinase enzymes.
in: Sittampalam G.S. Assay guidance manual. Eli Lilly & Company and the National Center for Advancing Translational Sciences, 2004
To determine optimal conditions for TR-FRET signal, titrations of donor and acceptor reagents were carried out in a 10 × 10 (2D-titration) format using a low nM concentration of biotinylated KRAS A146T (5 nM protein experiment shown in SI. 1A). We typically aim to use the lowest target protein and detection reagents, while not sacrificing assay quality, to reduce the tight binding limit for future dose-response studies and to reduce reagent consumption and associated cost. Using optimized donor and acceptor TR-FRET reagent concentrations, 1 nM and 3 nM respectively, a titration of biotin-KRAS A146T was performed to identify the concentration that provides a robust signal and acceptable signal to background; we selected 5 nM protein for the assay and additional assay development and validation experiments (Fig. 2A). A robust TR-FRET assay typically has a ratiometric signal ratio ≥ 3-fold over background. After final assay optimization, our signal to background was ∼5-fold. To ensure the assay reached equilibrium, measurements were taken over multiple time points which showed the TR-FRET signal stabilized at approximately one hour. Final assay measurements were then taken following an incubation time between one and one-and-a-half hours (SI. 1B). Finally, a direct titration of the tracer ligand, Cy5-GDP, in the TR-FRET assay was measured and yielded an EC50 of 18.7 nM (Fig. 2B).
A Z’ test was performed using full columns of unlabeled GDP as the low-end control, and 1% DMSO for the high-end control. The calculated interquartile means of these measurements were used to determine a Z’ of 0.86, indicating that the assay was robust and reproducible [
] (SI. 1C). We next completed a DMSO tolerance test to determine the maximum solvent limit and potential for assay interference. Z’ scores were estimated with DMSO concentrations ranging from 0-5% as a high-end control. Z’ scores ≥0.85 were obtained for each DMSO concentration indicating the solvent had a negligible effect on assay performance (SI. 1C).
- Zhang J.H.
- Chung T.D.
- Oldenburg K.R
A simple statistical parameter for use in evaluation and validation of high throughput screening assays.
J Biomol Screen. 1999; 4: 67-73
Finally, we performed competition TR-FRET experiments to validate the assay for HTS screening and establish its utility in finding KRAS A146T competitor small molecules. Unlabeled GDP was used in a dose-response format, where GDP was titrated in the presence of 3 nM Cy5-GDP to validate tracer ligand displacement. By developing the assay using low concentrations of both KRAS-A146T and TR-FRET reagents, a low tight-binding limit is achieved. This further allows for direct comparison of the IC50 value to the Kd of the competitor molecule [
]. An IC50 value of 26.9 nM was obtained for GDP (Fig. 2C), in close agreement with the value obtained for direct titration with Cy5-GDP (EC50 = 18.7 nM). These data show that GDP can act as a competitor molecule and that Cy5-GDP can be displaced by direct competition, further supporting use of this TR-FRET assay in a compound screening campaign.
- Cheng Y.
- Prusoff W.H.
Relationship between the inhibition constant (K1) and the concentration of inhibitor which causes 50 per cent inhibition (I50) of an enzymatic reaction.
Biochem Pharmacol. 1973; 22: 3099-3108
Primary library screening
Following optimization of the assay, a pilot (a.k.a. robustness) screen was completed using the LOPAC-1280 (Library of Pharmacologically Active Compounds) compound set. The library was screened in duplicate using 20 µM test compound concentration to compare reproducibility and verify robustness (Z’ > 0.9) (Fig. 3A). The screen of 1280 compounds yielded 17 hits with >50% inhibition and showed excellent reproducibility with a correlation coefficient (R) of 0.92 (Fig. 3B-C). As this library's purpose was primarily for HTS assay validation, hits from this library were inspected chemically and it was decided not to pursue them further due to low likelihood for medicinal chemistry tractability.
We next explored a focused nucleotide compound set (MedChemExpress Nucleotide Compound Library) of 251 small molecules. This set served two purposes; first as guanine nucleotides are present in this library it allows further validation that the assay can identify KRAS A146T competitor nucleotide ligands (such as GDP), and second, we sought to identify if any previously unexplored nucleotide analogs are able to compete with Cy5-GDP and serve as potential starting points for novel small molecule KRAS A146T chemical probes. Screening compounds at a dose of 20 µM with this library identified 37 hits with > 50% inhibition, including both GDP and GTP (Fig. 4). Many of the other hits were closely related GDP/GTP analogs and other non-chemically tractable moieties. These results validate the utility of our KRAS TR-FRET assay to identify small molecule binders of KRAS A146T.
We then initiated a larger primary screening campaign to potentially discover novel hits of KRAS A146T suitable for a hit-to-lead campaign. The primary screen consisted of three small molecule screening sets including: Chembridge 57K DIVERSet-CL, the UNC 5K Kinase Library, and a partial set of the UNC 100K Diversity Library (hereafter referred to as the CICBDD Diversity Library). These libraries were selected since they contain non-overlapping chemical diversity and contain drug-like and lead-like compounds. We included the 5K Kinase Library, a collection of compounds computationally selected and predicted to contain a classic ‘hinge-binding’ motif known to be important for ATP-competitive kinase inhibitors, simply because it has historically produced hits for non-kinase targets and compound purity across the set is typically >90%. In total, over 83,000 compounds were screened, and the assay performance was consistent and robust with Z’ greater than 0.8 in all screening plates. The screen yielded 157 hits with ≥ 50% inhibition (Fig. 4). However, 63 of the putative hit compounds were determined to quench Eu emission by >25% and were excluded from further investigation. Following further filtering to remove promiscuous and pan assay interference compounds (PAINS) molecules [
], such as those that auto-fluoresce or chelate metal ions, 63 compounds were selected for further validation by dose-response using the TR-FRET assay (SI. 2).
- Baell J.B.
- Holloway G.A.
New substructure filters for removal of pan assay interference compounds (PAINS) from screening libraries and for their exclusion in bioassays.
J Med Chem. 2010; 53: 2719-2740
Seven compounds from the nucleotide set and 63 compounds from the primary screening set were repurchased from chemical vendors and plated in 10-point serial dilutions starting at 200 µM final concentration. Of the seven nucleotide compounds tested, only two had an IC50 below 100 µM (SI. 2). Of the 63 compounds tested from the primary HTS libraries, twelve validated in dose-response assays with IC50 values <20 µM (SI. 2). The remaining compounds were observed to quench Eu at the higher dose ranges or had an IC50 >100 µM. These included the 41 hit compounds from the ChemBridge 57K DIVERSet-CL library. Further investigation into the ChemBridge library using chemiluminescent nitrogen detection (CLND) and LC-MS analysis revealed issues with compound identity and/or concentration and purity (data not shown), thus explaining the low validation rate for this set. Based on the TR-FRET dose-response data the molecule UNC10104889, from the UNC 5K Kinase set, was chosen for secondary assay validation and further follow up efforts (Fig. 5A-B). UNC10104889 had an IC50 of 13.2 µM, yielded an acceptable Hill slope (∼1.6), did not quench Eu emission at the highest concentration tested (200 µM), and was not flagged as a potential PAINs compound. UNC10104889 was also an attractive potential hit since it did not show promiscuity with a variety of assay data for both non-kinase and kinase targets (internal data in the UNC CICBDD database). Additionally, we identified a number of commercially available analogs for a small SAR study.
In attempt to show SAR and potentially find more potent and soluble compounds, we pursued UNC10104889 as the top hit candidate and conducted a follow-up screen of commercially available compounds based on structural similarity (i.e., “SAR-by-catalog”). Twenty-five purchased compounds were tested by dose-response in the TR-FRET assay. A subset of these analogs was tested for aqueous solubility and showed a wide range of values, from 5 µM to 200 µM. However, there was no correlation between compound affinity and solubility (SI. 3). Thirteen compounds with an IC50 below 50 µM were identified via SAR-by-catalog, with several showing similar potencies and slightly improved aqueous solubility compared to UNC10104889. Based on the dose-response data, a basic SAR off the core group A, where slight modifications were made on the phenyl moiety, appeared encouraging (Position R1; Table 1).
Table 1Purchased analogs of UNC10104889 with core group A (orange box), and truncated core group A (green box), and TR-FRET assay IC50 values.
Unfortunately, attempts to further validate UNC10104889 and related analogs by biophysical techniques have thus far proven difficult. Unfavorable solubility has prevented successful efforts to measure binding affinities by standard methods such as isothermal titration calorimetry (ITC) and surface plasmon resonance (SPR). An orthogonal fluorescent polarization (FP) assay was developed to determine if these hit compounds could be confirmed in a non-TR-FRET-based orthogonal assay. While unlabeled GDP was able to directly compete and displace Cy5-GDP from KRAS A146T in the FP assay, evaluation of UNC10104889 and its most soluble analog, UNC10309850B, revealed no Cy5-GDP displacement from KRAS A146T (SI. 4), indicating neither compound acts by direct competition with GDP. UNC10309850 was the most soluble analogue observed in the SAR study and originally had an IC50 of 8.4 µM. However, upon resynthesis, purification, and LC-MS/NMR quality control (via MedChemExpress SI. 6, 7), referred to here as UNC10309850B, the potency decreased to 69.2 µM (SI. 5). In total, these validation attempts and re-synthesis results raise the flag that this UNC10104889 hit series from the primary screen are not true hits. We cannot rule out the possibility that the activity observed for the UNC10104889 series analogs may result from TR-FRET assay related, or non-specific binding, artifacts such as colloidal aggregation [
]. Given the solubility liabilities, lack of robust SAR, and lack of activity in orthogonal and biophysical assays, we decided not to pursue this compound series further.
- Feng B.Y.
- et al.
A high-throughput screen for aggregation-based inhibition in a large compound library.
J Med Chem. 2007; 50: 2385-2390
From intensive research efforts focused on oncogenic KRAS and drug discovery in the past decade, several new drugs have made it into the clinic and market. Yet these efforts have been focused on more prevalent KRAS mutants such as G12D and the covalently modifiable G12C. Our work here is focused on the less prevalent A146T mutation with its unique mechanism of driving protein signaling by rapidly exchanging guanine nucleotides, appropriately dubbed a “fast-cycling” mutant. Herein, we show the development of a novel high-throughput TR-FRET assay to screen fast-cycling KRAS mutants using A146T as the proof-of-concept mutant due to the protein's reduced affinity for nucleotides.
We validated the assay by successfully identifying known endogenous ligands (GDP and GTP) from the screening campaign. Additionally, we presented the utility of the TR-FRET assay by completing, to our knowledge, the first HTS campaign against A146T, totaling over 83,000 compounds. Follow-up of the primary screen using a dose-response TR-FRET assay identified several initially promising small molecule inhibitors with an IC50 in the low micromolar range. A top hit molecule, UNC10104889, was selected and a small SAR-by-catalog screen further identified additional molecules with a modest increase in potency. Unfortunately, attempts to validate these hits by biophysical techniques were unsuccessful. The low solubility of many analogs in this series prevented the use of ITC and poor/no direct binding was observed by SPR (data not shown). Further evaluation of UNC10104889 with an orthogonal fluorescent polarization (FP) assay revealed no Cy5-GDP displacement from KRAS A146T, indicating UNC10104889 and its series may be TR-FRET-related assay artifacts. Due to the lack of convincing validation studies and the solubility liabilities, we decided not to progress this series further.
The assay development and HTS campaign described here demonstrate a novel screening approach that can be used to find new potential hit compounds leading to chemical probes and drug candidates targeting KRAS A146T and potentially other fast-cycling KRAS mutants that are activated independent of SOS mediated nucleotide exchange. Even though we do not report a bona-fide hit compound, our hit triaging process demonstrates the trials and tribulations of traditional screening campaigns for validating hits from a primary screen. Future directions with our robust and reproducible KRAS A146T TR-FRET assay will include screening additional diversity libraries and computational virtual screening approaches in search of a validated inhibitor. In addition, since we have demonstrated precedence for assay development and screening for KRAS A146T, the other fast-cycling KRAS mutants can likely be investigated using a similar approach toward discovery of novel chemical probes and drug discovery candidates.
The authors disclosed receipt of the following financial supports for the research, authorship, and/or publication of this article: funding for this work was supported by NIH 1R01CA223495, the UNC Lineberger Comprehensive Cancer Center Innovation Award, and the UNC Lineberger Comprehensive Cancer Center's University Cancer Research Fund.
CRediT authorship contribution statement
Jacob E. Larson: Conceptualization, Investigation, Visualization, Writing – review & editing. P. Brian Hardy: Resources. Noah K. Schomburg: Resources. Xiaodong Wang: Data curation. Dmitri Kireev: Data curation. Kent L. Rossman: Conceptualization, Investigation, Resources, Visualization, Writing – review & editing, Supervision, Funding acquisition. Kenneth H. Pearce: Conceptualization, Writing – review & editing, Supervision, Funding acquisition.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
We thank the members of the UNC CICBDD for suggestions and helpful discussions, particularly Dr. Tina M. Leisner for review of the manuscript.
Appendix. Supplementary materials
- The guanine nucleotide-binding switch in three dimensions.Science. 2001; 294: 1299-1304
- Past and future strategies to inhibit membrane localization of the KRAS oncogene.Int J Mol Sci. 2021; 22
- RAS proteins and their regulators in human disease.Cell. 2017; 170: 17-33
- The metabolic landscape of RAS-driven cancers from biology to therapy.Nat Cancer. 2021; 2: 271-283
- KRAS mutation: from undruggable to druggable in cancer.Signal Transduct Target Ther. 2021; 6: 386
- The frequency of ras mutations in cancer.Cancer Res. 2020; 80: 2969-2974
- Cancer statistics, 2022.CA Cancer J Clin. 2022; 72: 7-33
- RAS mutations are not created equal.Cancer Discov. 2019; 9: 696-698
- RAS isoforms and mutations in cancer at a glance.J Cell Sci. 2016; 129: 1287-1292
- Classification of KRAS-activating mutations and the implications for therapeutic intervention.Cancer Discov. 2022; 12: 913-923
- Kinetics of interaction of nucleotides with nucleotide-free H-ras p21.Biochemistry. 1990; 29: 6058-6065
- Physiological concentrations of purines and pyrimidines.Mol Cell Biochem. 1994; 140: 1-22
- Therapeutic targeting of oncogenic K-Ras by a covalent catalytic site inhibitor.Angew Chem Int Ed. 2014; 53: 199-204
- K-Ras(G12C) inhibitors allosterically control GTP affinity and effector interactions.Nature. 2013; 503: 548-551
- Discovery of a covalent inhibitor of KRASG12C (AMG 510) for the treatment of solid tumors.J Med Chem. 2020; 63: 52-65
- The current state of the art and future trends in RAS-targeted cancer therapies.Nat Rev Clin Oncol. 2022; 19: 637-655
- Identification of MRTX1133, a noncovalent, potent, and selective KRASG12D inhibitor.J Med Chem. 2022; 65: 3123-3133
- The current understanding of KRAS protein structure and dynamics.Comput Struct Biotechnol J. 2020; 18: 189-198
- Tissue-specific oncogenic activity of KRASA146T.Cancer Discov. 2019; 9: 738-755
- Genomic and biological characterization of exon 4 KRAS mutations in human cancer.Cancer Res. 2010; 70: 5901-5911
- A general TR-FRET assay platform for high-throughput screening and characterizing inhibitors of Methyl-Lysine reader proteins.SLAS Discov. 2019; 24: 693-700
- Assay development for protein kinase enzymes.in: Sittampalam G.S. Assay guidance manual. Eli Lilly & Company and the National Center for Advancing Translational Sciences, 2004
- A simple statistical parameter for use in evaluation and validation of high throughput screening assays.J Biomol Screen. 1999; 4: 67-73
- Relationship between the inhibition constant (K1) and the concentration of inhibitor which causes 50 per cent inhibition (I50) of an enzymatic reaction.Biochem Pharmacol. 1973; 22: 3099-3108
- New substructure filters for removal of pan assay interference compounds (PAINS) from screening libraries and for their exclusion in bioassays.J Med Chem. 2010; 53: 2719-2740
- A high-throughput screen for aggregation-based inhibition in a large compound library.J Med Chem. 2007; 50: 2385-2390
Published online: December 20, 2022
Accepted: December 8, 2022
Received in revised form: November 18, 2022
Received: October 19, 2022
© 2022 The Authors. Published by Elsevier Inc. on behalf of Society for Laboratory Automation and Screening.
User licenseCreative Commons Attribution (CC BY 4.0) |
How you can reuse
Elsevier's open access license policy
Creative Commons Attribution (CC BY 4.0)
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article
- Reuse portions or extracts from the article in other works
- Sell or re-use for commercial purposes
Elsevier's open access license policy