Mechanistic Analysis of an Extracellular Signal-Regulated Kinase 2-Interacting Compound that Inhibits Mutant BRAF-Expressing Melanoma Cells by Inducing Oxidative Stress.

Constitutively active extracellular signal-regulated kinase (ERK) 1/2 signaling promotes cancer cell proliferation and survival. We previously described a class of compounds containing a 1,1-dioxido-2,5-dihydrothiophen-3-yl 4-benzenesulfonate scaffold that targeted ERK2 substrate docking sites and selectively inhibited ERK1/2-dependent functions, including activator protein-1-mediated transcription and growth of cancer cells containing active ERK1/2 due to mutations in Ras G-proteins or BRAF, Proto-oncogene B-RAF (Rapidly Acclerated Fibrosarcoma) kinase. The current study identified chemical features required for biologic activity and global effects on gene and protein levels in A375 melanoma cells containing mutant BRAF (V600E). Saturation transfer difference-NMR and mass spectrometry analyses revealed interactions between a lead compound (SF-3-030) and ERK2, including the formation of a covalent adduct on cysteine 252 that is located near the docking site for ERK/FXF (DEF) motif for substrate recruitment. Cells treated with SF-3-030 showed rapid changes in immediate early gene levels, including DEF motif-containing ERK1/2 substrates in the Fos family. Analysis of transcriptome and proteome changes showed that the SF-3-030 effects overlapped with ATP-competitive or catalytic site inhibitors of MAPK/ERK Kinase 1/2 (MEK1/2) or ERK1/2. Like other ERK1/2 pathway inhibitors, SF-3-030 induced reactive oxygen species (ROS) and genes associated with oxidative stress, including nuclear factor erythroid 2-related factor 2 (NRF2). Whereas the addition of the ROS inhibitor N-acetyl cysteine reversed SF-3-030-induced ROS and inhibition of A375 cell proliferation, the addition of NRF2 inhibitors has little effect on cell proliferation. These studies provide mechanistic information on a novel chemical scaffold that selectively regulates ERK1/2-targeted transcription factors and inhibits the proliferation of A375 melanoma cells through a ROS-dependent mechanism. SIGNIFICANCE STATEMENT: Constitutive activation of the extracellular signal-regulated kinase (ERK1/2) pathway drives the proliferation and survival of many cancer cell types. Given the diversity of cellular functions regulated by ERK1/2, the current studies have examined the mechanism of a novel chemical scaffold that targets ERK2 near a substrate binding site and inhibits select ERK functions. Using transcriptomic and proteomic analyses, we provide a mechanistic basis for how this class of compounds inhibits melanoma cells containing mutated BRAF and active ERK1/2.

Introduction

Deregulated protein kinase activity that promotes cell proliferation and survival is a hallmark of many cancers. Thus, targeted inhibition of protein kinases with small molecules has become an important therapeutic strategy to treat cancer (Wu et al., 2016). The extracellular signal–regulated kinase (ERK) 1 and 2 proteins are members of the mitogen-activated protein kinase family and major regulators of intracellular signaling events that provide cells with a proliferation and survival advantage. Constitutive activation of ERK1/2 occurs through mutations or elevated expression of upstream activators, including receptor tyrosine kinases, Ras G-proteins, and Raf kinases (Reuter et al., 2000; Shapiro, 2002; Bennasroune et al., 2004; Arora and Scholar, 2005; Mendelsohn and Baselga, 2006). Activating mutations in MEK1, a direct activator of the ERK1/2 proteins, may contribute to deregulated ERK1/2 activity (Emery et al., 2009). In addition, activating mutations in a conserved glutamate in a region of ERK2 associated with protein-protein interactions are associated with the progression of cutaneous T-cell lymphomas (Arvind et al., 2005; Mahalingam et al., 2008; da Silva Almeida et al., 2015). A number of small-molecular-weight anticancer drugs targeting the ATP binding or catalytic sites on receptor tyrosine kinases, BRAF, and MEK1/2 are being used in the clinic to block ERK1/2 signaling (Flaherty and McArthur, 2010; Zhong and Bowen, 2011; Wu et al., 2015). Despite promising initial responses, resistance to these kinase inhibitors is often observed, and subsequent therapeutic options are limited. Drug resistance occurs through several mechanisms, including acquired mutations in the kinase catalytic site targeted by the compounds, activation of alternative mechanisms to turn on proteins downstream of the targeted kinase, or increases in compensatory signaling pathways that provide a survival advantage independent of the kinase being targeted (Poulikakos and Rosen, 2011; Poulikakos and Solit, 2011; Tuma, 2011; Fedorenko et al., 2015). Recent efforts aimed at overcoming inherent or acquired resistance to existing kinase inhibitors are directly targeting ERK1/2 proteins (Morris et al., 2013; Maik-Rachline and Seger, 2016; Germann et al., 2017) and highlight the need to identify alternative approaches to inhibit constitutively active kinase signaling pathways that promote drug resistance and drive the survival of cancer cells.

Using computational modeling and biologic testing, we have previously identified small molecules that are designed to selectively inhibit ERK1/2 signaling by disrupting specific substrate interactions (Hancock et al., 2005; Chen et al., 2006; Boston et al., 2011; Samadani et al., 2015). This approach targeted two defined substrate docking sites on ERK2: the D-recruitment site (DRS) or F-recruitment site (FRS), which interact with substrates containing D-domains or DEF motifs, respectively (Burkhard et al., 2011; Lee et al., 2011; Sammons et al., 2019). Previous studies suggest that most ERK1/2 substrates contain D-domains, although some substrates, such as the ternary complex factor Elk-1, contain both a D-domain and a DEF motif (von Kriegsheim et al., 2009). Several transcription factors that promote cancer cell growth, including c-Myc and Fos family proteins, contain the DEF motif, which is characterized by a phenylalanine-any amino acid-phenylalanine (FXF) sequence that is 6–10 amino acids away from the serine or threonine phosphorylation site. Hydrophobic interactions between the FXF motif and the recruitment site on ERK1/2 facilitate kinase-substrate interactions necessary for efficient phosphoryl transfer. To selectively target DEF motif–containing transcription factors that promote tumor growth, we previously described the identification of a class of compounds containing a 1,1-dioxido-2,5-dihydrothiophen-3-yl 4-benzenesulfonate scaffold that inhibited DEF motif–containing substrates of the activator protein-1 (AP-1) family (Samadani et al., 2015). This class of compounds selectively inhibited the proliferation of melanoma cells harboring BRAF mutations and constitutively active ERK1/2. These compounds were also effective at inhibiting melanoma cell lines that are inherently resistant to clinically relevant inhibitors of mutated BRAF because of elevated mitogen-activated protein kinase kinase kinase-8 (MAP3K8) (Samadani et al., 2015).

To define the structure-activity relationship for the 1,1-dioxido-2,5-dihydrothiophen-3-yl 4-benzenesulfonate–based compounds, the current studies synthesized several analogs and identified a single double bond in the sulfur heterocycle that was essential for the compound’s biologic activity. The findings also demonstrate that this class of compounds forms a covalent interaction with ERK2 near the FRS and has differential effects on the expression levels of ERK1/2-regulated transcription factors containing DEF motifs. The most active compound, SF-3-030, was examined for global effects on gene and protein levels to evaluate ERK1/2-dependent and independent mechanisms of action. The findings support a mechanism by which SF-3-030 induced an oxidative stress response that mediated the inhibition of melanoma cells containing mutated and constitutively active BRAF.

Materials and Methods

Compound Synthesis.

All reactions were performed in oven-dried glassware under an inert (N2) atmosphere. Anhydrous solvents were used as supplied without further purification. Chemicals were purchased from Sigma-Aldrich (St. Louis, MO), Alfa Aesar (Ward Hill, MA), or Oakwood Chemicals (West Columbia, SC). Reactions were monitored by thin-layer chromatography, visualizing with KMnO4 stain and/or UV as appropriate. Silica gel flash column chromatography was performed with 60-Å, 230–400 mesh silica gel, and crude reaction mixtures were first adsorbed onto silica gel from CH2Cl2 at room temperature. 1H and 13C NMR spectra were recorded on a Varian 400-MHz NMR spectrometer (amaZon X; Bruker) at 25°C. Chemical shifts are reported in parts per million and are referenced to residual nondeuterated solvent peak (CHCl3: δH 7.26, δC 77.2; DMSO: δH 2.50, δC 39.5). Prior to biologic testing, final compounds were confirmed to be >95% pure by High-performance liquid chromatography. Detailed methods of the chemical synthesis are provided in the Supplemental Data.

Chemicals and Antibodies.

The MEK1/2 inhibitor AZD6244 (catalog number BV-2234-5), 6-(4-bromo-2-chloroanilino)-7-fluoro-N-(2-hydroxyethoxy)-3-methylbenzimidazole-5-carboxamide), was purchased from Axxora (Farmingdale, NY). The ERK1/2 inhibitor SCH772984 (catalog number S7101), ((R)-1-(2-oxo-2-(4-(4-(pyrimidin-2-yl)phenyl)piperazin-1-yl)ethyl)-N-(3-(pyridin-4-yl)-1H-indazol-5-yl)pyrrolidine-3-carboxamide), was purchased from Selleckchem (Houston, TX). Antioxidants N-acetyl l-cysteine (NAC) (catalog number A7250), sodium pyruvate (catalog number P2256), and mannitol (catalog number M4125) were purchased from Sigma-Aldrich. NRF2 inhibitors ML-385 (catalog number SML 1833) and brusatol (catalog number SML 1868) were purchased from Sigma-Aldrich. Antibodies against total c-Fos (catalog number 2250), FosB/B2 (catalog number 2251), Fos-related antigen 1 (Fra-1) (catalog number 5281), c-Jun (catalog number 9165), c-Myc (catalog number 5605), Elk-1 (catalog number 9182), β-actin (catalog number 4970), and cleaved poly (ADP-ribose) polymerase (PARP) (catalog number 9541) were purchased from Cell Signaling Technology (Beverly, MA). The phospho-specific ERK1/2 (pThr183/pTyr185) (catalog number M9692) and α-tubulin (catalog number T6074) antibodies were from Sigma-Aldrich. Phosphorylation-specific antibodies against histone H3 (pSer10) (catalog number 9701), Elk-1 (pS383) (catalog number 9186), and MEK1/2 (pSer217/pSer221) (catalog number 9121) were purchased from Cell Signaling Technology. Antibodies against ERK2 (catalog number sc-154), MEK1/2 (catalog number sc-81504), heme oxygenase-1 (HMOX1) (catalog number sc-136960), and NRF2 (catalog number sc-722) were from Santa Cruz Biotechnology. A second NRF2 antibody (catalog number ab 137550) and the NAD(P)H dehydrogenase [quinone] 1 (NQO1) antibody (catalog number ab34173) were purchased from Abcam (Cambridge, MA). The forkhead box protein D3 (FoxD3) (catalog number 631701) antibody was purchased from BioLegend (San Diego CA). The oxidative stress–induced growth inhibitor 1 (OSGIN1) (catalog number 15248-1-AP), glutamate-cysteine ligase modifier (GCLM) (catalog number 14241-1-AP), sulfiredoxin 1 (SRXN1) (catalog number 14273-1-AP), glutamate-cysteine ligase catalytic (GCLC) (catalog number 12601-1-AP), aldo-keto reductase family 1 member B10 (catalog number 18252-1-AP), glucose-6-phosphate dehydrogenase (catalog number 66373-1-Ig), and thioredoxin reductase 1 antibodies (catalog number 11117-1-AP) were purchased from ProteinTech (Rosemont, IL). The cystine/glutamate transporter antibody (catalog number NB300-317SS) was purchased from Novus Biologicals (Centennial, CO).

Cell Lines.

Cell lines used included HeLa cervical carcinoma cells, Jurkat T-cell leukemia cells, and NRas mutated promyelocytic leukemia HL-60 cells. Melanoma cell lines tested included A375 and SK-Mel-28 (homozygous BRAF V600E mutation), SK-MEL-5 (heterozygous BRAF V600E mutation), SK-Mel-2 (BRAF wild type, NRas Q61R mutation), and RPMI-7951 cells (heterozygous BRaf mutation), which are inherently resistant to BRAF inhibitors because of elevated MAP3K8 (Johannessen et al., 2010). Cells were purchased from American Type Culture Collection (Manassas, VA) and grown in Dulbecco’s modified Eagle’s medium (DMEM)or Eagle's minimal essential medium (EMEM) plus 10% FBS. All media were supplemented with penicillin and streptomycin. The cell lines have been authenticated by evaluating short tandem repeat DNA profiles and matching with the American Type Culture Collection data base. Cell lines are routinely tested for mycoplasma contamination using the MycoAlert detection kit (Lonza, Walkersville, MD).

Cell Viability/Apoptosis Assays.

Cells were seeded at 5000 cells per well in 96-well plates, cultured overnight, and treated for 48 hours with the indicated dose of compounds. IC50 values were determined using nine data points and 3-fold dilutions of 0.001–10 μM for AZD6244 and SCH772984 or 0.1–100 μM for SF-3-030. Cell viability was measured according to the manufacturer’s instructions using the fluorescent CellTiter Blue Assay (Promega, Madison, WI) or the CellTiter-Glo Luminescent Cell Viability Assay (Promega). To determine the minimal exposure time needed to induce an apoptotic response, cells were exposed to test compounds for various times (0–24 hours), the cells were washed, and then fresh medium was added without test compound for a total of 24 hours. Relative apoptosis was measured by immunoblotting for the cleaved form of PARP.

Reactive Oxygen Species Measurements.

Reactive oxygen species (ROS) were measured using the cell-permeable fluorogenic reagent CellROX Deep Red according to the manufacturer’s instructions (Thermo Fisher Scientific, Waltham, MA). Briefly, A375 melanoma cells were seeded at 104 cells per well in a 96-well plate. After 24 hours, the cells were coincubated with SF-3-030 ± various ROS inhibitors as well as 5 µM of CellROX Deep Red reagent for 1 hour at 37°C. Subsequently, cells were washed three times with Dulbecco’s modified Eagle’s medium containing no Phenol Red. Live cell fluorescence was imaged using a Nikon Eclipse Ti confocal microscope using a 10× objective magnification. Total fluorescent intensity was obtained in triplicate wells from four frames of view per well (∼500–1000 cells per frame) with a biologic duplicate performed for each treatment condition. The mean ± S.D. values were determined using NIS-Elements analysis software.

RNA Sequencing Analysis.

A375 cells were treated for 1 hour with 10 μM AZD6244 or 25 μM SF-3-030, and total RNA was isolated (RNAeasy; Qiagen, Hilden, Germany) from treated and untreated cells in duplicate samples. Differential expression analysis was done by GENEWIZ (South Plainfield, NJ) using Illumina Genome Analyzer and HiSeq instruments in the FastQ format as previously described (Anders et al., 2013). Data were analyzed using Ingenuity Pathway Analysis software. Changes in expression were shown for genes with false discovery rates <0.05.

Immunoassays.

Immunoblot analysis of protein expression, phosphorylation, and cell proliferation were done as previously described (Samadani et al., 2015). Cells were washed with cold PBS (pH 7.2; Invitrogen, Carlsbad, CA), and protein lysates were collected with 2× SDS-PAGE sample buffer (4% SDS, 5.7 M β-mercaptoethanol, 0.2 M Tris pH 6.8, 20% glycerol, 5 mM EDTA). Proteins were separated by SDS-PAGE, analyzed by immunoblotting, and detected using enhanced chemiluminescence (Pierce ECL; Thermo Fisher Scientific). Immunoblots were quantified via densitometry using the Azure c300 Chemiluminescent Western Blot Imaging System (Azure Biosystems, Dublin, CA) for image capture, and data were quantified on the Image Studio Lite Quantification software (v5.2; Li-COR Biotechnology, Lincoln, NE). Immunoassays were also performed using Wes Simple Western (ProteinSimple, San Jose, CA). Electropherograms were quantified using Compass 225 for SW software (v3.1.7; ProteinSimple) applying a Gaussian peak fit distribution for determining area under the curve.

Covalent Modification Analysis by High-Resolution Liquid Chromatography-Tandem Mass Spectrometry.

His-tagged ERK2 was purified from Escherichia coli BL21 (DE3) cells transformed with a wild-type erk2 construct using the previously described method (Burkhard et al., 2011). For covalent modification analysis, in vitro kinase reactions containing 100 µg of purified ERK2, 1 mM ATP, 1× NEBuffer for Protein Kinases (New England Biolabs, Ipswich, MA), and 50 µM SF-3-030 were incubated for 2 hours at 25°C. After the reactions, ERK2 protein was desalted, reduced, alkylated, and trypsinolyzed on filter as described previously (Wiśniewski et al., 2009; Erde et al., 2014). Tryptic peptides were separated on a nanoACQUITY Ultra Performance Liquid Chromatography (UPLC) analytical column (BEH130 C18, 1.7, 75 μm × 200 mm; Waters Corporation, Milford, MA) over a 165-minute linear acetonitrile gradient (3%–40%) with 0.1% formic acid on a Waters nanoACQUITY UPLC system (Waters Corporation) and analyzed on a coupled Thermo Scientific Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Scientific, San Jose, CA), as described (Williamson et al., 2016). Full scans were acquired at a resolution of 120,000, and precursors were selected for fragmentation by higher-energy collisional dissociation (normalized collision energy at 32%) for a maximum 3-second cycle. Tandem mass spectra were searched against the ERK2 protein sequence using a SEQUEST HT algorithm (Eng et al., 2008) and an MS Amanda algorithm (Dorfer et al., 2014) with a maximum precursor mass error tolerance of 10 ppm. Possible substitution (SN2′ and SN2, +115.9932), Michael addition (+324.0126), and carbamidomethylation of cysteine were treated as dynamic modifications. Resulting hits were validated at a maximum false discovery rate of 0.01 using a semisupervised machine learning algorithm, Percolator (Käll et al., 2007). The probabilities of modification sites were computed using a ptmRS algorithm (Taus et al., 2011).

Saturation Transfer Difference-NMR Analysis.

Saturation transfer difference-NMR (STD-NMR) analysis of ligand binding to ERK2 was done as previously described for p38 MAPK (Shah et al., 2017). A 1 mM stock solution of SF-3-030 was made in 85% D2O:15% d6-DMSO (v/v). STD-NMR samples contained 150 mM NaCl, 50 mM phosphate (pH 7), 200 μM SF-3-030, and 5 μM ERK2 protein in D2O. Spectra of both compound and ligand bound protein were recorded on an Agilent DD2 500-MHz spectrometer equipped with a 5-mm inverse proton-fluorine-carbon-nitrogen probe head at 25°C. Further detailed methods of the NMR protocol used are provided in the Supplemental Data.

Differential Protein Expression by High-Resolution Liquid Chromatography-Tandem Mass Spectrometry.

A375 cells grown on 10-cm plates were treated for 4 and 12 hours with 0.1% DMSO vehicle, 25 μM SF-3-030, or 10 μM SCH772984. After one wash in cold PBS, the cells were collected by scraping twice with cold PBS and centrifuged at 3000 rpm for 2 minutes; the cell pellets were stored at −80°C. Cells were lysed in 4% sodium deoxycholate, reduced, alkylated, and trypsinolyzed on filter as described (Wiśniewski et al., 2009). Tryptic peptides were separated on a nanoACQUITY UPLC analytical column (CSH130 C18, 1.7 μm, 75 μm × 200 mm; Waters Corporation) over a 180-minute linear acetonitrile gradient (3%–43%) with 0.1% formic acid on a Waters nanoACQUITY UPLC system (Waters Corporation) and analyzed on a coupled Thermo Scientific Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific) as described (Williamson et al., 2016). Full scans were acquired at a resolution of 120,000, and precursors were selected for fragmentation by higher-energy collisional dissociation (normalized collision energy at 30%) for a maximum 3-second cycle. Tandem mass spectra were searched against a UniProt human reference proteome using a SEQUEST HT algorithm (Eng et al., 2008) with a maximum precursor mass error tolerance of 10 ppm. Resulting hits were validated at a maximum false discovery rate of 0.01 using a semisupervised machine learning algorithm, Percolator (Käll et al., 2007). Abundance ratios were measured by comparing the mass spectrometer 1 peak volumes of peptide ions, whose identities were confirmed by mass spectrometer 2 sequencing as described above. Label-free quantifications were performed using an aligned Accurate Mass and Retention Time cluster quantification algorithm (Qi et al., 2012). Pathway and gene ontology analysis were performed with Qiagen Ingenuity and Panther Gene ontology databases, as described (Krämer et al., 2014; Mi et al., 2017). Proteins showing at least a doubling in expression as compared with untreated cells, with an FDR-adjusted ANOVA P value of <0.05, were considered significantly changed and used for further analysis. FDR-corrected Fisher’s exact P values < 0.05 using a previously described procedure for multiple testing were used in the gene ontology analyses to identify biologic processes, molecular functions, and cellular components associated with observed protein changes (Benjamini and Hochberg, 1995). The Ingenuity pathway data base was used to predict canonical pathways and upstream regulators according to the proteins that were significantly different using an absolute activation z-score > 2 for at least one condition with a Fisher’s exact test P value < 0.05.

Lactate Dehydrogenase/Cytotoxicity Assay.

Cells were seeded 5000 cells per well in 96-well plates, cultured overnight, and treated for 48 hours with the indicated dose of compound SF-3-030. Cytotoxicity via LDH release was measured according to the manufacturer’s instructions using the CyQUANT LDH Cytotoxicity Assay Kit (Thermo Fisher Scientific).

Statistical Analysis.

Graphical statistical analysis was performed using a one-tailed ANOVA with a 95% confidence interval using GraphPad-Prism version 5.01. Three biologic replicates were chosen for data subjected to statistical analysis. The data represent descriptive statistics and were used to summarize key findings. Results were considered statistically significant if the P value is less than 0.05.

Results

The Double Bond in the Sulfur Heterocycle Is Required for Biologic Activity.

Our previous studies described a novel class of small molecules containing a 1,1-dioxido-2,5-dihydrothiophen-3-yl 4-benzenesulfonate scaffold that selectively inhibited BRAF mutated melanoma cells containing constitutively active ERK1/2 signaling (Samadani et al., 2015). A limited structure-activity relationship study was performed to determine key chemical features required for the compound’s biologic activity. Modifications of the parent compound, SF-3-026, focused on the benzene ring, the linker region connecting the two ring structures, and removal of the double bond in the sulfur heterocycle (Fig. 1). The analogs were tested at a single high dose for effects on viability of cancer cell lines containing BRAF mutations (A375, RPMI-7951, and SK-Mel-28), NRas mutation (HL60), or no known mutations in ERK1/2 pathway proteins (HeLa and Jurkat) (Fig. 1). RPMI-7951 cells are inherently resistant to BRAF targeted inhibitors because of upregulated expression of MAP3K8 (Johannessen et al., 2010). In agreement with our previous studies (Samadani et al., 2015), SF-3-026, SF-3-027, SF-3-029, and SF-3-030 showed selective inhibition of cancer cell lines with activating mutations in the ERK1/2 pathway (Fig. 1). The inhibitory effects were lost in compounds in which the double bond in the sulfur heterocycle was removed (Fig. 1). Select modifications to the benzene group or linker in the absence of the double bond in the sulfur heterocycle had no effect on cell viability, indicating the 1,1-dioxido-2,5-dihydrothiophen-3-yl moiety was essential for the compound’s biologic activity (Fig. 1).

Fig. 1.

Effects of SF-3-026 and analogs on proliferation of cancer cell lines. Data show percent proliferation compared with controls (100%) for A375, RPMI-7951, SK-Mel-28, HL-60, HeLa, or Jurkat cells after treatment with 100 μM of each test compound for 48 hours. The mean and S.D. from three independent experiments are shown.

Previous studies showed that catalytic site inhibitors of MEK1/2 and ERK1/2 are selective for cells with activating BRAF or Ras mutations (Yeh et al., 2007; Morris et al., 2013). Similarly, we determined the IC50 values for SF-3-030 and ATP-dependent/catalytic site inhibitors of MEK1/2 and ERK1/2 in four cancer cell lines and showed that all compounds favored inhibition of cells containing activating BRAF and Ras mutations (Supplemental Table 1). The reported doubling times for A375, HeLa, and Jurkat cells are ∼25–30 hours, compared with ∼50 hours for the RPMI-7951 cells, indicating that each compound’s potency is independent of cell growth rates. These findings provide additional support that SF-3-030 inhibitory effects on cell growth are correlated with ERK1/2 activity.

The effects of SF-2-110 (inactive control), SF-3-026, and SF-3-030 (Supplemental Fig. 1A) on cell viability were further evaluated in dose-response assays using four melanoma cell lines containing either BRAF or NRas mutations. As expected, SF-2-110 did not affect the viability of any of the cell lines up to 50 μM (Supplemental Fig. 1, B–E). In contrast, SF-3-026 and SF-3-030 inhibited all cell lines with IC50 values in the 5–10 μM range, with SF-3-030 showing the highest potency against SK-MEL-28 and SK-MEL-2 cells (Supplemental Fig. 1, B–E). These findings indicate that the sensitivity of cancer cells in vitro to the lead compounds is independent of the BRAF or NRas mutational status.

SF-3-030 Directly Interacts with ERK2 through Noncovalent and Covalent Mechanisms.

Given the higher inhibitory potency of SF-3-030, we evaluated whether this compound directly interacted with ERK2. Using STD-NMR, it was confirmed that SF-3-030 interactions with ERK2 occur primarily through a reversible interaction of the naphthalene group and the sulfur heterocycle (Supplemental Fig. 2). However, the requirement for the double bond in the sulfur heterocycle of SF-3-030 for biologic activity suggested that the compound’s mechanism of action involves the formation of covalent adducts with cysteine residues. Two possible SF-3-030 adducts could result from a Michael addition or a substitution reaction that would increase the molecular mass of a cysteine by 324 or 116 Da, respectively (Fig. 2A). To test this, ERK2 was incubated with SF-3-030, and covalent adducts were analyzed by high-resolution liquid chromatography-tandem mass spectrometry. Of the seven cysteine residues in ERK2, cysteine 252 (C252) was the predominant residue modified by a 116-Da mass, suggesting a substitution reaction and decomposition of SF-3-030 (Fig. 2; Supplemental Fig. 3). C252 is located near the DEF motif–recruitment site, which is consistent with its proposed mechanism of targeting DEF motif–containing substrates (Samadani et al., 2015). Other major MAP kinases, such as p38α and c-Jun N-terminal kinase 1, do not have a cysteine located at this site. We have also previously determined that SF-3-030 had no effect on p38 MAPK signaling (Samadani et al., 2015), suggesting that SF-3-030 is not acting as a random alkylating agent or affecting similar MAP kinases. These data suggest that the initial binding interactions of SF-3-030 with ERK2 involve noncovalent interactions that position the compound to form a covalent bond between C252 and the sulfur heterocycle group.

Fig. 2.

SF-3-030 interacts with and modifies ERK2 via cysteine adduct formation. (A) Putative mechanisms of SF-3-030 adduct formation with cysteine residues on ERK2. (B) Space-filling model of ERK2 with highlighted cysteine residues and SF-3-030 modification site (adapted from Protein Data Bank Identification (PDB ID): 4GT3). Cysteine modifications determined by high-resolution liquid chromatography-tandem mass spectrometry. Cysteine residues are colored in (green), as well as the TXY motif (magenta), the F-recruitment site (red), and the primary modification site C252 (red font).

SF-3-030 Selectively Regulates ERK-Dependent Immediate Early Gene Expression.

We next examined relative protein levels of ERK1/2-regulated immediate early genes (IEGs). Previously, we showed that a 1-hour treatment with SF-3-030 caused differential changes in AP-1 protein levels characterized by inhibition of Fra-1, FosB, and the alternative splice variant FosB2 but no effect on c-Fos levels (Samadani et al., 2015). The relative levels of these proteins along with other AP-1 members were examined in A375 cells treated for 0–24 hours with SF-3-030 or the ERK1/2 catalytic site inhibitor SCH772984 as a positive control. Similar to our previous studies (Samadani et al., 2015), a 1 hour of treatment with SF-3-030 or SCH772984 had little effect on c-Fos levels (Fig. 3A). However, after 2 hours of treatment, c-Fos increased only with SF-3-030 and persisted for up to 24 hours (Fig. 3A). Another major component of the AP-1 complex, c-Jun, increased with either SF-3-030 or SCH772984 treatments (Fig. 3A). It should be noted that the increased levels of c-Fos and c-Jun do not correlate with increased AP-1 activity, which we have shown to be inhibited in these and other cell lines treated with SF-3-030 (Samadani et al., 2015; Defnet et al., 2019).

Fig. 3.

Lead compounds differentially regulate IEG levels in A375 cells. (A) Immunoblots for c-Fos, c-Jun, FosB/B2, Fra-1, and c-Myc proteins in cells treated with 10 μM SCH77294 or 25 μM SF-3-030 for 0–24 hours. (B) Phosphorylated (S383) and total Elk-1 levels in cells treated with 10 μM SCH77294 or 25 μM SF-3-030 for 0–24 hours. (C) Cells treated for 4 hours with varying doses of SF-2-110, SF-3-026, or SF-3-030. Controls include untreated or cells treated with 5 μM AZD6244 or SCH772984. Immunoblots show relative c-Fos, c-Jun, FosB/B2, Fra-1, and c-Myc protein levels. Levels of active ERK1/2 (ppERK1/2), total ERK2, and α-tubulin are shown to demonstrate ERK1/2 pathway activity and equal protein loading in (A) and (C). (D) Relative quantification of c-Fos, Fra-1, and c-Myc protein levels after 4 hours of exposure with 25 µM of SF-2-110 or SF-3-030. Mean and S.D. are from three independent experiments, and graph was determined via densitometry as described in the . * and ** represent P < 0.05 and P < 0.01, respectively. (E) A375 cells were exposed to SF-3-030 for the indicated times and then incubated with fresh media without compound for a total incubation time of 24 hours. Lysates were immunoblotted for cleaved PARP, phosphorylated histone H3 (pH3), and total ERK2 for a loading control. Data are representative of three independent experiments. The numbers in each immunoblot represent the relative levels of protein, normalized to α-tubulin, as determined by densitometry. Molecular mass markers (kDa) are indicated on the right of each immunoblot.

Other Fos family members, including FosB/B2 and Fra-1 decreased in cells treated with SCH772984 (Fig. 3A). Cells treated with SF-3-030 also showed decreased Fra-1 levels, whereas FosB/B2 levels transiently decreased, followed by an increase after 4 hours as observed with c-Fos (Fig. 3A). The levels of c-Myc, a potential target for treating melanoma (Polsky and Cordon-Cardo, 2003; Korkut et al., 2015), showed a rapid inhibition by SF-3-030 when compared with SCH772984 (Fig. 3A). Interestingly, c-Myc levels increased back to basal levels after 24 hours of treatment with SF-3-030 (Fig. 3A). However, this dose of SF-3-030 is lethal to ∼90% of A375 cells after 48 hours of exposure (Supplemental Fig. 1B). A transient increase in the ERK1/2-mediated phosphorylation of Elk-1, a ternary complex factor involved in regulating c-fos and other IEGs (Shaw and Saxton, 2003), was observed in SF-3-030–treated cells and correlated with ERK1/2 phosphorylation (Fig. 3B). However, there was no evidence that MEK1/2, the upstream activator of ERK1/2, was activated, indicating SF-3-030 was acting at the level of ERK1/2 (Supplemental Fig. 4).

The dose-dependent effect of the active compounds SF-3-026 and SF-3-030 on IEG levels was examined after a 4-hour exposure. The inactive control, SF-2-110, had no effect on c-Fos, c-Jun, Fra-1, FosB/B2, or c-Myc levels (Fig. 3C). In contrast, SF-3-026 and SF-3-030 caused a dose-dependent decrease in Fra-1, FosB/B2, and c-Myc protein levels but an increase in c-Jun and c-Fos (Fig. 3C). The increased potency of SF-3-030 in affecting IEG protein levels compared with the parent compound SF-3-026 is consistent with our previous studies (Samadani et al., 2015). It is also noted that SF-3-030 at a dose near its IC50 (∼5 μM) for inhibition of cell proliferation was as potent at inhibiting Fra-1 and c-Myc protein levels as the 5 μM doses of AZD6244 or SCH77294 (Fig. 3C). Quantitative analysis of c-Fos, Fra-1, and c-Myc protein levels after exposure with 25 µM SF-2-110 or SF-3-030 is shown in Fig. 3D. Together, these findings demonstrate that the lead compounds can differentially affect ERK1/2-regulated transcription factors.

It was next determined whether the changes in IEG expression observed after 1 to 2 hours with SF-3-030 (Fig. 3A) coincided with the exposure time needed to induce an apoptotic response. In these experiments, A375 cells were exposed to SF-3-030 for various times, washed to remove excess compound, and then incubated with serum-supplemented growth medium for a total of 24 hours. Apoptosis was evident after 2 hours of exposure to SF-3-030, as measured by the appearance of cleaved PARP (Fig. 3E). This exposure time also coincided with the loss of histone H3 phosphorylation at Ser10, which is a marker of mitosis and is observed during the transcription of selective IEGs in response to ERK1/2 pathway activation (Sassone-Corsi et al., 1999). Loss of histone H3 phosphorylation after 2 hours of exposure to SF-3-030 suggests that mitotic progression is inhibited and that phosphorylation of S10 is dispensable for regulating the transcription of IEG, such as c-Fos and c-Jun, as previously reported (Drobic et al., 2010).

SF-3-030 Regulates ERK1/2-Dependent and Independent Transcription.

RNA sequencing analysis was used to measure early changes in gene expression patterns and identify on- and off-target effects of SF-3-030. Total RNA was isolated from A375 cells treated for 1 hour with an IC90 dose (25 µM) of SF-3-030 or the MEK inhibitor AZD6244 (10 µM) as a positive control for ERK1/2 pathway inhibition. The IC90 dose for SF-3-030 was used, as it was determined to have no general cytotoxicity in LDH assays [data not shown; (Defnet et al., 2019)]. Approximately 16,000 transcripts were identified in each condition, and statistically significant changes (≥1.5-fold) in transcript expression between treated and untreated cells were reported. We identified 38 or 68 transcripts in cells treated with SF-3-030 or AZD6244, respectively, which showed significant decreases with 12 genes common to both treatments (Supplemental Table 2). ERK1/2 pathway targets involved in cell growth and survival that decreased with both treatments included c-Myc, CTGF, SOX11, and GATA3. AZD6244 also inhibited transcription of c-Fos, early growth reponse-1 (Egr-1), and IER3 genes, which are elevated in BRAF V600E mutated melanoma cells and may promote ERK-mediated growth of these tumor types (Pratilas et al., 2009). The inhibition of c-Myc by SF-3-030, AZD6244, and SCH772984 at the RNA and protein levels (Fig. 3) supports a shared mechanism of targeting ERK-mediated signaling by these compounds.

The data also revealed that AZD6244 inhibited many negative regulators of the ERK1/2 pathway, including Sprouty (SPRY) and dual specificity phosphatase transcripts (Supplemental Table 2). SPRY1 was the only negative regulator of ERK1/2 signaling that was inhibited by both AZD6244 and SF-3-030 (Supplemental Table 2). Additional growth-related genes that decreased only with AZD6244 treatment included regulators of epidermal growth factor receptor signaling including c-Fos, TGFA, and HBEGF (Supplemental Table 2). In contrast, only SF-3-030 inhibited a unique set of growth- and angiogenesis-promoting genes, including CYR61 (also called IGF-binding protein 10), which may promote tumor progression during hypoxia (Kunz et al., 2003), and BCL6, which has been associated with poor overall survival of patients with melanoma (Alonso et al., 2004).

There were 14 or 24 transcripts that increased >1.5-fold after treatment with SF-3-030 or AZD6244, respectively, relative to untreated cells, and only one gene, CHMP4C, was common to both treatments (Supplemental Table 2). CHMP4C is involved in sorting and delivery of endosomal vesicles to the lysosome and in cell cycle checkpoint control during cytokinesis (Carlton et al., 2012). Ingenuity pathway analysis (IPA) indicated that several genes upregulated in response to SF-3-030 treatment, including HMOX1, c-Fos, c-Jun, and DnaJ homolog subfamily B member 1 (Supplemental Table 2), were consistent with activation of an NRF2-mediated oxidative stress response. However, given that the exposure to SF-3-030 was only 1 hour, it is likely that other transcription factors besides NRF2 are regulating these genes at this time point.

Other genes upregulated by SF-3-030 included MAP3K14 (also called NF-κB–inducing kinase), which activates NF-κB, and the Egr-1 transcription factor, which may have tumor suppressor functions through regulation of p53 (Shin et al., 2006). Transcripts that increased only in response to AZD6244 treatment included SOX2 and FOXD3, which are associated with survival of melanoma-initiating cells, metastasis, and drug resistance (Abel et al., 2013; Santini et al., 2014). The activation status of the top 50 upstream regulators predicted to be affected after treatment with SF-3-030 or AZD6244 was determined using IPA software. Using Z-scores > 2 or < −2 to indicate activation or inactivation, respectively, and P values of <0.03, AZD6244 predictably inhibited many regulators of ERK1/2 and other MAP kinase signaling pathways (Supplemental Table 3). In contrast, SF-3-030 was predicted to affect only six regulators, including activation of Egr-1, as indicated previously (Supplemental Table 3). Given some overlap with AZD6244-regulated genes, these data suggest that SF-3-030 treatment affects fewer cellular signaling events than AZD6244 and that early MAP kinase signaling events largely remain intact.

SF-3-030 Enhances Markers of Mitochondrial Dysfunction and Oxidative Stress.

To gain further insight into the mechanism of SF-3-030 inhibition of cell proliferation, we next used high-resolution liquid chromatography-tandem mass spectrometry to determine the changes in levels of soluble proteins in A375 cells treated for 4 or 12 hours with SCH772984 or SF-3-030. After 4 hours, 22 (SF-3-030) or 16 (SCH772984) proteins were identified that showed significant (P < 0.05) changes compared with untreated cells (Supplemental Table 4). A summary of proteins that showed statistically significant increases or decreases (≥1.5-fold) with SF-3-030 or SCH772984 as compared with untreated controls is shown in Table 1. At the 4-hour time point, only one protein, adenosylhomocysteinase (AHCY), which regulates the generation of the S-adenosylmethionine methyl donor, was inhibited with SF-3-030 or SCH772984 treatment. Proteins that increased after exposure to SF-3-030 included the zinc finger protein 774 (ZNF774), which has recently been shown to have tumor suppressor functions in hepatocellular carcinoma (Guan et al., 2020); transcription factor 4 (TCF4), which regulates cell differentiation; and HMOX1 (Table 1). SF-3-030 decreased cytochrome c-type heme lyase (HCCS) levels, suggesting defects in mitochondrial electron transport function (Table 1). SCH772984 increased levels of the transmembrane and TPR repeat–containing protein 1, which indicated disruption of calcium homeostasis in the endoplasmic reticulum (Sunryd et al., 2014), and decreased the tetratricopeptide repeat protein, whose function is currently unknown (Table 1).

TABLE 1

High-resolution liquid chromatography-tandem mass spectrometry analysis of proteins that increase or decrease ≥1.5-fold after treatment with SF-3-030 or SCH772984 for 4 h relative to untreated A375 cells (set at 1)

Protein changes common to both SF-3-030 and SCH772984 are italicized.

Proteins that increase with SF-3-030	Fold increase
ZNF774	16.28
MEMO1	10.55
TTC39C	5.18
MYO18A	4.18
TCF4	2.24
HMOX1	1.77
Proteins that decrease with SF-3-030	Fold decrease
HCCS	3.35
KIAA1211	2.24
ACTN2	1.66
AHCY	1.53

ACTN2, α-actinin-2; KIAA1211, Capping protein inhibiting regulator of actin dynamics; MYO18A, unconventional myosin-XVIIIa; SRGAP1, SLIT-ROBO Rho GTPase-activating protein 1; (TMTC1), transmembrane and TPR repeat-containing protein 1; TTC29, tetratricopeptide repeat protein 29; TTC39C, Tetratricopeptide repeat protein 39C

After 12 hours, 44 and 48 proteins showed significant changes (P < 0.05) with SF-3-030 and SCH772984 treatment, respectively, as compared with untreated cells (Supplemental Table 5). Proteins that increased or decreased ≥1.5-fold with SF-3-030 or SCH772984 treatments relative to untreated controls are shown in Table 2. Common to SF-3-030 or SCH772984 treatments was increased levels of TCF4 and synthesis of cytochrome c oxidase 2 (SCO2) and decreased levels of a protein involved in ribosome biogenesis (RSL24D1). SCO2 has been associated with p53-mediated apoptosis signaling pathways through the generation of ROS (Madan et al., 2013).

TABLE 2

High-resolution liquid chromatography-tandem mass spectrometry analysis of proteins that increase or decrease ≥1.5-fold after treatment with SF-3-030 or SCH772984 for 12 h relative to untreated A375 cells (set at 1)

Protein changes common to both SF-3-030 and SCH772984 are italicized.

A.
Proteins that increase with SF-3-030	Fold increase
OSGIN1	21.04
ZNF774	18.18
TCF4	10.95
PXDN	9.43
Protein MEMO1	8.37
TTC39C	7.31
MYO18A	4.50
SCO2	3.89
Jouberin (AHI1)	3.12
HMOX1	3.02
Protein moonraker (KIAA0753)	2.26
ASB6	2.07
REPS1	1.98
NOC2L	1.73
Sickle tail protein homolog (KIAA1217)	1.71
NCAPD3	1.49
Proteins that decrease with SF-3-030	Fold decrease
Myomegalin (PDE4DIP)	2.80
HCCS	2.49
AP1M1	2.24
CNGB1	2.11
FOXR2	1.76
GRIP2	1.72
SUN1	1.68
RSL24D1	1.66
KIAA1211	1.62
INTS2	1.51

ABHD5, Abhydrolase domain-containing protein 5; AHI1, Abelson helper integration site 1 protein homolog; AP1M1, AP-1 complex subunit μ-1; ASB6, ankyrin repeat and SOCS box protein 6; ATP11A, ATPase class VI type 11A; CACNB4, Calcium channel voltage-dependent subunit β 4; CNGB1, cyclic nucleotide-gated cation channel β-1; COL1A1, collagen α-1(I) chain; DCN, decorin; FOXR2, forkhead box protein R2; GFPT2, glutamine-fructose-6-phosphate aminotransferase 2; GRIP2, glutamate receptor-interacting protein 2; HBS1L, HBS1-like protein; HK2, hexokinase-2; INTS2, integrator complex subunit 2; KIAA1211, uncharacterized protein KIAA1211; MYO18A, unconventional myosin-XVIIIa; NCAPD3, condensin-2 complex subunit D3; NOC2L, nucleolar complex protein 2 homolog; PABPC3, polyadenylate-binding protein 3; PDE5A, phosphodiesterase 5A cGMP-specific; PDIA3, protein disulfide-isomerase A3; PKP2, plakophilin-2; PXDN, peroxidasin homolog; REPS1, RalBP1-associated Eps domain-containing protein 1; RFC4, replication factor C subunit 4; RNMT, mRNA cap guanine-N7 methyltransferase; SUN1, Sad1/unc-84 protein-like 1; TFAP2A, transcription factor AP-2-α; TTC39C, chromosome 18 open reading frame 17_isoform CRA_c; WDR59, WD repeat-containing protein 59. KIAA0753, moonraker; KIAA1217, Sickle tail protein homolog; PDE4DIP, Phosphodiesterase 4D-interacting protein.

Unique to SF-3-030 was a significant increase in the OSGIN1 protein (Table 2), which is regulated by oxidized lipids and NRF2 during oxidative stress to promote cytochrome c release from mitochondria (Li et al., 2007). Mediator of ErbB2-driven cell motility 1 (MEMO1), a protein involved in the sustained production of ROS (MacDonald et al., 2014), was also increased in cells exposed to SF-3-030 (Table 1; Table 2). As was observed after 4 hours, HCCS levels were also decreased after a 12-hour exposure to SF-3-030 (Table 1; Table 2), further supporting compromised mitochondria function. Together, these findings indicate SF-3-030 induced signaling events that reduce mitochondrial function and increase ROS. After 12 hours of exposure to SCH772984, a significant induction of FOXD3 was observed (Table 2), which is consistent with previous studies that show upregulation of this protein promotes resistance to Raf and MEK inhibitors (Abel et al., 2013). Immunoblot analysis confirmed the induction of FOXD3 expression in A375 cells treated with SCH772984 but not in SF-3-030–treated cells (Supplemental Fig. 5).

Based on the observed changes in protein levels, IPA pathway analysis suggested a potential role for SF-3-030 induction of NRF2-mediated oxidative stress response and heme degradation in agreement with mitochondrial dysfunction. These findings are consistent with previous studies that implicate the ERK1/2 pathway in regulating oxidative phosphorylation in melanoma cells and the induction of ROS after treatment with inhibitors of BRAF and MEK1/2 (Haq et al., 2013; Cesi et al., 2017; Yuan et al., 2020). Immunoblot analysis confirmed that SF-3-030 treatment rapidly induced NRF2 levels that were sustained for at least 24 hours and that NRF2 induction could be inhibited by cotreatment with the ROS inhibitor NAC (Fig. 4).

Fig. 4.

SF-3-030 induces NRF2 levels. (A) Immunoblot analysis of NRF2 in A375 cells treated with 10 μM SCH772984 or 25 μM SF-3-030 for 0–24 hours. (B) NRF2 levels in A375 cells treated with 25 μM SF-3-030 ±5 mM NAC for 0, 1, or 4 hours. Nonspecific bands that crossreact with the NRF2 primary antibody are indicated (ns) for (A) and (B). The graph in (C) shows the densitometry quantitation of NRF2 under the conditions described in (B). Mean and S.D. are from three independent experiments. * indicates statistical significance compared with SF-3-030 treatment only (P < 0.05). The numbers in each immunoblot represent the relative levels of protein, normalized to α-tubulin, as determined by densitometry. Molecular mass markers are indicated on the left of each immunoblot.

SF-3-030 Inhibition of A375 Cell Proliferation Is Dependent on ROS Induction but Independent of NRF2.

The transcriptome and proteome data provided evidence for the generation of ROS in cells treated with SF-3-030. We confirmed the increased ROS production in A375 cells after a 1-hour treatment with SF-3-030, which was partially inhibited by cotreatment with several ROS inhibitors (Fig. 5A). NAC was the only ROS inhibitor that restored A375 cell proliferation in the presence of SF-3-030 (Fig. 5B). Similarly, only NAC reversed SF-3-030 induction of c-Fos (Fig. 5C). Although these data indicate that NAC may mitigate ROS production by SF-3-030 to restore A375 cell proliferation, there was also the possibility that NAC directly forms adducts with SF-3-030, which would reduce the compound’s effective concentration and biologic activity. However, using thin-layer chromatography to separate compounds, we found no evidence that NAC directly interacted with SF-3-030 after a 1- or 24-hour incubation in both PBS and methanol solvent systems (data not shown).

Fig. 5.

SF-3-030 induction of ROS mediates inhibition of A375 cell proliferation. (A) ROS was measured by CellROX Deep Red reagent fluorescence in A375 cells treated with 25 μM SF-3-030 in the absence (white bars) or presence (black bars) of the following ROS inhibitors (ROSi): 10 mM sodium pyruvate (NaPyr), 100 mM mannitol (Mann), or 10 mM NAC for 60 minutes. Graphs represent three independent experiments, and each data point represents the mean ± S.D. from three wells with four fields of view per well and each field containing between 500 and 1000 cells. Statistical significance was determined within each experiment (* and ** represent P < 0.05 and P < 0.01, respectively). (B) A375 cell viability was measured after 48 hours in untreated or SF-3-030–treated cells in the absence (white bars) or presence (black bars) of the ROS inhibitors at the concentrations described in (A). The mean and S.D. for cell proliferation are from three independent experiments. ** indicates statistical significance (P < 0.01) compared with SF-3-030 treatment only. (C) Relative levels of c-Fos after 4 hours of treatment with SF-3-030 alone (white bars) or in combination with the indicated ROS inhibitors (black bars) at the concentrations indicated in (A). Relative c-Fos levels were determined by ProteinSimple immunoanalysis and normalized to total β-actin.

Induction of NRF2 has been implicated in protecting against oxidative stress induced by anticancer drugs and may contribute to drug resistance (Telkoparan-Akillilar et al., 2019). To assess whether NRF2 induction protected cells from SF-3-030 inhibition, we cotreated A375 cells with the NRF2 inhibitors ML-385 or brusatol (Ren et al., 2011; Olayanju et al., 2015; Singh et al., 2016). SF-3-030 and sulforaphane, as a positive control, induced NRF2 and target genes such as HMOX1, OSGIN1, NAD(P)H dehydrogenase [quinone] 1 (NQO1), GCLM, and SRXN1 (Fig. 6, A and B; Supplemental Fig. 6A). In contrast, protein levels for other NRF2-regulated genes, including aldo-keto reductase family 1 member B10, GCLC, glucose-6-phosphate dehydrogenase, cystine/glutamate transporter/solute carrier family 7 member 11, and thioredoxin reductase 1 were not induced by SF-3-030 (Supplemental Fig. 7).

Fig. 6.

NRF2 inhibitors do not affect SF-3-030 inhibition of A375 cell proliferation. (A) A375 cells were treated for 8 hours with 25 µM SF-3-030 in the absence or presence of ML-385 (50 µM) or brusatol (Bru; 30 nM). Lysates were immunoblotted for relative levels of NRF2, NQO1, HMOX1, and OSGIN1, as shown. (B) A375 lysates from cells treated for 8 hours with 25 µM SF-3-030 (SF) in the absence or presence of Bru (30 nM) were immunoblotted for GCLM and SRXN1. Positive control lysates from HeLa and A549 cells were used for GCLM and SRXN1, respectively. The numbers in each immunoblot represent the relative levels of protein, normalized to β-actin, as determined by densitometry. Molecular mass markers are indicated on the left of each immunoblot. (C) A375 cell viability with varying doses of ML-385 or (D) brusatol. (E) Cell viability with varying doses of SF-3-030 in the absence (white bars) or presence (black bars) of 50 µM ML-385. (F) Combination index with varying doses of SF-3-030 and brusatol. Untreated control cells are indicated with a striped column. Relative cell viability was measured after 48 hours, and data are representative of three independent experiments.

As compared with sulforaphane, SF-3-030 induced a more robust and sustained induction of HMOX1, indicating qualitative differences in NRF2 activators. Brusatol was a potent inhibitor of SF-3-030 or sulforaphane induction of NRF2 and its target genes. Although ML-385 inhibited HMOX1 induced by sulforaphane after 24 hours (Supplemental Fig. 6, A and B), it was not as effective as brusatol at inhibiting NRF2 signaling induced by SF-3-030 (Fig. 6A). ML-385 had no effect on cell proliferation up to 50 µM; however, brusatol inhibited A375 cells in a dose-dependent manner (Fig. 6, C and D). Cotreatment with brusatol or ML-385 had little effect on SF-3-030 or sulforaphane inhibition of A375 cell proliferation, indicating NRF2 induction was dispensable for the cell response to these compounds (Fig. 6, E and F; Supplemental Fig. 6, C and D). Dose-dependent inhibition of A375 cells by sulforaphane or brusatol alone supports the high sensitivity these cells have to fluctuations in ROS (Meierjohann, 2014). Together, these data indicate that SF-3-030–mediated inhibition of A375 melanoma cell proliferation is through a mechanism that is ROS-dependent and NRF2-independent.

Discussion

The first objective of the current studies was to elucidate the structure-activity relationship of a 1,1-dioxido-2,5-dihydrothiophen-3-yl 4-benzenesulfonate chemical scaffold that we previously identified to selectively inhibit cancer cells containing activated ERK1/2 signaling (Samadani et al., 2015). The dependence on a double bond in the sulfur heterocycle for biologic activity indicated the compound’s mechanism of action involved the formation of covalent adducts with cysteine residues on ERK1/2 and perhaps other proteins through Michael addition chemistry. SF-3-030 was confirmed via STD-NMR and mass spectrometry–based analyses to interact with ERK2 via noncovalent interactions and the formation of a covalent adduct on C252 near the FRS. Although the formation of covalent adducts raises the concern about specificity and off-target effects, we have found no evidence that SF-3-030 leads to changes in other MAP kinase signaling pathways or randomly interacts with cysteine using the ROS inhibitor NAC (data not shown). This indicates that the chemical structure of SF-3-030 confers target selectivity and prevents random interactions with sulfhydryl groups.

There is the potential that SF-3-030 forms a covalent adduct with Kelch-like ECH-associated protein 1 (KEAP1), the negative regulator of NRF2. KEAP1 has up to 21 cysteine residues that may be modified by chemical stressors to relieve KEAP1 inhibition and allow NRF2 activation (Dayalan Naidu and Dinkova-Kostova, 2020). Sulforaphane, which modifies KEAP1 on C151 to activate NRF2, showed qualitative differences in the expression of NRF2-regulated genes as compared with SF-3-030 (Fig. 6; Supplemental Fig. 6). It will be interesting to evaluate whether SF-3-030 also modifies KEAP1 cysteine residues and how they impact NRF2 functions. As indicated in the current studies, SF-3-030 induction of NRF2 did not appear to affect A375 cell proliferative responses to SF-3-030.

Despite past concerns about safety and off-target toxicity, advances in structure-function relationships and medicinal chemistry have generated renewed interest in the development of irreversible binding drugs for treating disease (Singh et al., 2011). There are several small-molecule anticancer drugs in the clinic that form covalent adducts on receptor and nonreceptor tyrosine kinases through Michael addition [reviewed in Zhao and Bourne (2018)]. Afatinib, osimertinib, and neratinib form irreversible adducts on cysteine residues in the catalytic sites of epidermal growth factor receptors for the treatment of non–small-cell lung cancers and breast cancers, whereas ibrutinib targets Bruton’s tyrosine kinase for the treatment of lymphoma/leukemia. Additionally, there is renewed interest in targeting ERK1/2 with small molecules via allosteric inhibition of substrate interactions sites, such as the DRS (Sammons et al., 2019). Recent studies have identified a small-molecule inhibitor that covalently binds to a conserved cysteine (C159) in the DRS of ERK2 (Kaoud et al., 2019). This compound selectively blocked the activation of ERK1/2, did not modify other members of the MAP kinase family, and inhibited the proliferation of melanoma cells with BRAF (V600E) mutations that were resistant to BRAF inhibitors (Kaoud et al., 2019). Future studies will be aimed at identifying whether cysteine residues on other proteins may be modified by SF-3-030 to establish their role in cell signaling and growth inhibition.

The second objective of the studies was to evaluate global effects of SF-3-030 on gene and protein changes to gain insight into this chemical scaffold’s mechanism of action. The transcriptome and proteome data show that the overall number of genes and proteins affected by the lead compound, SF-3-030, are less than or similar to the number of proteins targeted by known ERK1/2 or MEK1/2 ATP-competitive or catalytic site inhibitors. Similar to the ATP-competitive ERK1/2 pathway inhibitors, SF-3-030 appears to be more selective for cancer cells containing constitutively activating ERK1/2 pathway mutations. Although rapidly dividing cells are typically more sensitive to anticancer drugs, SF-3-030 effects on a select number of cells with activating ERK1/2 pathway mutations were shown to be independent of cell doubling times. Although further studies with additional cell lines will be needed to verify this observation, these data indicate that off-target effects on nontransformed cells may be limited.

Transcriptome and proteome analyses indicated that SF-3-030 induced rapid changes in the levels of ERK1/2-regulated transcription factors consistent with an oxidative stress response. Transcription factors such as c-Fos and c-Jun are elevated in response to oxidative or metabolic stress (Webster et al., 1994), and ROS-mediated induction of ERK1/2 can enhance Elk-1–mediated transcription and c-Fos expression (Müller et al., 1997). It was intriguing that SF-3-030 caused a robust induction of c-Fos but inhibited c-Myc protein levels (Fig. 3) and that these changes occurred at the level of transcription (Supplemental Table 2). Other examples in which protein levels increase because of SF-3-030 induction of transcription include c-Jun and HMOX1. In contrast, SF-3-030 also inhibited Fra-1 levels, but there was no evidence this occurred through effects on transcription (Fig. 3; Supplemental Table 2). It is possible that the 1-hour time point used to evaluate transcriptome changes by RNA sequencing did not capture changes in Fra-1 transcription. The expression of Fra-1 and other Fos family proteins is regulated by c-Fos and subsequent AP-1 complex activity (Milde-Langosch, 2005). However, c-Fos induction in the current studies does not correlate with increased AP-1 activity, which we previously showed to be inhibited by SF-3-030 (Samadani et al., 2015; Defnet et al., 2019). Although apoptosis can be induced by c-Fos induction (Preston et al., 1996), further studies will be needed to clarify the requirement for c-Fos expression in SF-3-030–mediated cell death in the absence of AP-1 activity.

ERK1/2 activation increases the expression of c-Myc (Kerkhoff et al., 1998), which can protect melanoma cells against oxidative stress (Benassi et al., 2006). In contrast, SF-3-030 mediated downregulation of c-Myc and its potential role in responding to elevated ROS by regulating the synthesis of antioxidants such as glutathione (Gao et al., 2009). After exogenous hydrogen peroxide–induced oxidative stress, ERK1/2 phosphorylation of c-Myc at serine 62 facilitates c-Myc recruitment to the GCLC promoter and the expression of GCLC, the rate-limiting enzyme involved in glutathione synthesis (Benassi et al., 2006). However, in the current studies, the dramatic inhibition of c-Myc protein levels by SF-3-030 after 8 hours (Fig. 3A) is not observed with GCLC (Supplemental Fig. 7). Nonetheless, targeted inhibition of c-Myc with ERK1/2 pathway inhibitors is a suggested approach to decrease the proliferation of melanoma cells and overcome drug resistance (Ciuffreda et al., 2009; Korkut et al., 2015).

The role of ROS in promoting or inhibiting cancer cells is subject to debate, and there is evidence to support that either increased or decreased ROS production can sensitize cancer cells to growth inhibition and cell death (Trachootham et al., 2009; Galadari et al., 2017). Although the use of antioxidant strategies may protect against some cancers (Trachootham et al., 2009), there are limited clinical data that support the beneficial effects of using antioxidant supplements to reduce the risk of developing cancer (Goodman et al., 2011). However, the mechanisms used by many anticancer drugs to kill cells involves increased oxidative stress (Trachootham et al., 2009). Melanocytes and melanoma cells are particularly sensitive to ROS, and the ROS dose will likely determine whether cells are able to adapt or will die (Meierjohann, 2014). Additional evidence supports the use of oxidative stress to inhibit melanoma cell metastasis (Piskounova et al., 2015). Although ROS activation may not be sufficient to inhibit all cancer cells, enhanced ROS production may also play a role in improving anticancer efficacy of combination therapies (Adams et al., 2013). Recent data suggest that therapeutic approaches that increased ROS may be beneficial to patients who have developed resistance to BRAF and MEK1/2 inhibitors (Wang et al., 2018).

The selectivity of SF-3-030 toward cancer cells with activating ERK1/2 pathway mutations supports the idea that exploiting ROS homeostasis may be therapeutically beneficial in some cancers. Although upregulation of ROS may provide a survival advantage in melanomas that are resistant to ERK1/2 pathway inhibitors, additional increases in ROS with histone deacetylase inhibitors promoted an apoptotic response in these cells (Wang et al., 2018). Further studies will be needed to determine whether SF-3-030 induces DNA instability and ROS production like histone deacetylase inhibitors. DNA damage may induce the ROS-responsive tumor suppressor OSGIN1 protein to enhance apoptosis in osteosarcoma cells through a mechanism involving p53 disruption of mitochondria functions and cytochrome c release (Hu et al., 2012). Although SF-3-030 induced the protein levels of OSGIN1 and a subset of NRF2-regulated genes (Fig. 6, A and B), the NRF2 inhibitor brusatol did not affect SF-3-030 inhibition of A375 cell proliferation (Fig. 6F). These findings raise the potential to develop ROS activators that are lethal to cancer cells but do not promote an NRF2 response that leads to chemoresistance and cancer cell survival (Okazaki et al., 2020). Taken together, the current studies provide mechanistic support to evaluate how ROS production by SF-3-030 and related compounds can be used as an approach to inhibit the proliferation of cancers cells with activating mutations in the ERK1/2 pathway.