Out of the dark, a new kinase target in ovarian cancer
Ovarian cancer is frequently fatal; it is difficult to detect and challenging to treat. Uncovering genetic signatures has done little to improve clinical outcomes. Using cell lines and patient tumors, Kurimchak et al. analyzed the functional kinome of high-grade serous ovarian carcinoma (HGSOC) and identified a relatively uncharacterized kinase called MRCKA, among potentially others, as being important for HGSOC cell survival and tumor growth. Decreasing MRCKA abundance or pharmacologically inhibiting its kinase activity killed HGSOC cells but, as with so many growth inhibitors, led to activation of other survival-promoting kinases in remaining cells. Cotargeting one of these (PAK1) or cotreating cells with platinum-based chemotherapy had greater cytotoxicity. The findings identify multiple potential therapeutic strategies to treat HGSOC and prevent drug resistance in patients.
Abstract
High-grade serous ovarian carcinoma (HGSOC) is the most lethal gynecological cancer with few effective, targeted therapies. HGSOC tumors exhibit genomic instability with frequent alterations in the protein kinome; however, only a small fraction of the kinome has been therapeutically targeted in HGSOC. Using multiplexed inhibitor beads and mass spectrometry, we mapped the kinome landscape of HGSOC tumors from patients and patient-derived xenograft models. The data revealed a prevalent signature consisting of established HGSOC driver kinases, as well as several kinases previously unexplored in HGSOC. Loss-of-function analysis of these kinases in HGSOC cells indicated MRCKA (also known as CDC42BPA) as a putative therapeutic target. Characterization of the effects of MRCKA knockdown in established HGSOC cell lines demonstrated that MRCKA was integral to signaling that regulated the cell cycle checkpoint, focal adhesion, and actin remodeling, as well as cell migration, proliferation, and survival. Moreover, inhibition of MRCKA using the small-molecule BDP9066 decreased cell proliferation and spheroid formation and induced apoptosis in HGSOC cells, suggesting that MRCKA may be a promising therapeutic target for the treatment of HGSOC.
INTRODUCTION
High-grade serous ovarian carcinoma (HGSOC) is one of the most common and lethal forms of ovarian carcinoma, and current treatments have only modestly affected survival (1). Frequent alterations in DNA repair proteins such as BRCA1/2 sensitize HGSOC tumors to DNA-damaging agents, such as carboplatin, and to drugs that target homologous recombination such as poly(adenosine diphosphate–ribose) polymerase (PARP) inhibitors (2). Although frontline cytotoxic chemotherapy is initially effective for many patients, relapse and emergence of drug resistance are common with no subsequent therapies available (3). HGSOCs are characterized by near-universal mutation and/or loss of the TP53 tumor suppressor gene and, consequently, exhibit genome instability and aberrant cell signaling (1, 4). Signaling abnormalities in HGSOCs, particularly those involving aberrant protein kinases, represent potential therapeutic avenues for treatment (5–7).
Therapies targeting recognized, well-understood, cancer-associated protein kinases such as EGFR, ERBB2, and SRC have shown limited clinical benefit in HGSOC, prompting the search for new anticancer kinase targets (8, 9). In the present study, we interrogated the HGSOC kinome for previously unidentified kinase vulnerabilities using multiplexed inhibitor beads and mass spectrometry (MIB-MS) in combination with loss-of-function assays. MIB-MS, along with other kinase enrichment techniques, such as Kinobeads and KiNativ, provides a quantitative proteomics method to measure kinase levels for a substantial proportion of the kinome, including many largely unexplored kinases for which there are few existing references, reagents, or inhibitors (10–12). Functional interrogation of MIB-MS kinome signatures using small interfering RNA (siRNA) knockdown strategies in cancer cells assigns growth and survival functions to unexplored kinases (11).
RESULTS
Mapping the kinome landscape of HGSOC tumors using MIB-MS uncovers a shared kinome signature
Purification of endogenous kinases from patient tumors was carried out using MIB columns composed of four immobilized adenosine triphosphate (ATP)–competitive pan-kinase inhibitors (Purvalanol B, VI16832, PP58, and CTx-0249885) (Fig. 1A) (13). To quantify the MIB-bound kinome of HGSOC primary and PDX tumors, we used a super-stable isotope labeling by/with amino acids in cell culture (s-SILAC) strategy (14). High concordance of kinase levels of sections from a single HGSOC tumor demonstrates the robust reproducibility of combining MIB-MS and s-SILAC to kinome profile tumors (Fig. 1B and data file S1). Using this workflow, we quantified the MIB-enriched kinomes of 25 stage III and IV primary HGSOC patient tumors and 10 HGSOC patient-derived xenograft (PDX) models derived from patient tumor and ascites specimens (data file S2).
(A) Schematic of experimental approach. MIB-MS was used to quantify kinase abundance in HGSOC patient tumors to map the proteomic landscape of the kinome and identify prevalent kinases previously unexplored in HGSOC. m/z, mass/charge ratio. (B) Measurement of kinase abundance in a patient HGSOC tumor sectioned into four pieces, each individually kinome-profiled by MIB-MS, and kinase abundance was determined by s-SILAC quantitation. Pearson’s correlation thereof was determined using Perseus. (C) Proportion of the kinome quantitated by MIB-MS in HGSOC (“HGS”) tumors that represent established (red) and unexplored (gray) targets in HGSOC. Data are from one independent assay per sample in 25 tumor tissues and 10 PDX tumor tissues. The kinome tree was generated in KinMap and was reproduced courtesy of Cell Signaling Technology. (D and E) PCA, including PC1 versus PC2 (D) and PC1 versus PC3 (E), of MIB-MS determined kinome profiles of HGSOC primary and PDX tumors. Data are from one independent assay per sample in 25 tumor tissues and 10 PDX tumor tissues. (F) Volcano plot comparison of HGSOC primary and PDX tumor MIB-MS kinome profiles. Statistical differences in kinase log2 s-SILAC ratios comparing HGSOC primary versus PDX tumors were determined by paired t test Benjamini-Hochberg adjusted P values at an FDR of <0.05. Data are from one independent assay per sample in 25 tumor tissues and 10 PDX tumor tissues. (G) Predominant kinome signature identified by MIB-MS among HGSOC patient and PDX tumors, representing kinases detected at similar abundances among tumors. Statistical differences in kinase log2 s-SILAC ratios comparing HGSOC tumors from the main cluster relative to others were determined by paired t test Benjamini-Hochberg adjusted P values at an FDR of <0.05. Heat map depicts kinase log2 s-SILAC ratios, highlighting established HGSOC drivers, including those with drugs approved or in clinical trials, as well as those previously unexplored in HGSOC. Data are from one independent assay per sample in 25 tumor tissues and 10 PDX tumor tissues. (H) Kinases previously unexplored in HGSOC enriched in the MIB-MS kinome signature from (G). Data are from one independent assay per sample in 25 tumor tissues and 10 PDX tumor tissues. Kinome plot was produced using KinMap. The kinome tree was reproduced courtesy of Cell Signaling Technology. (I) Bar graph depicting kinases that were increased ≥2-fold relative to the s-SILAC reference in HGSOC tumors. The number of HGSOC tumors with kinases exhibiting log2 s-SILAC values (≥1.0) was determined, and the bar graph was generated in Prism software. Related data (additional PCA, correlation, and hierarchical clustering of MIB-MS profiles of primary and PDX HGSOC tumors) can be found in fig. S1.
In total, we measured MIB-binding s-SILAC values for 324 kinases, with 206 kinases quantitated in 70% of the tumors; most of these have not previously been explored as therapeutic targets in HGSOC (Fig. 1C and data file S3). Principal component analysis (PCA) of kinase s-SILAC ratios revealed several of the HGSOC primary and PDX tumors grouped together exhibiting similar MIB-MS kinome profiles (Fig. 1, D and E, and fig. S1, A and B). Volcano plot comparison of HGSOC primary and PDX tumor MIB-MS profiles showed some differences in kinase levels among tumor types, including elevated levels of myosin light chain kinase (MYLK), protein kinase D1 (PRKD1), and protein kinase G1 (PRKG1) in primary tumors, and increased p21-activated kinase 4 (PAK4) and hepatocyte growth factor receptor (MET) levels in PDX tumors (Fig. 1F). Several of the kinases elevated in primary tumors had stromal-related functions, consistent with recent transcriptomic studies showing that most of the genes up-regulated in primary versus matched HGSOC PDX tissues were stromal-specific genes (15). Unsupervised hierarchical clustering of kinase s-SILAC ratios identified a main cluster of HGSOC tumors (n = 27) that exhibited a shared MIB-MS kinome signature—kinases that displayed a reproducible difference in expression relative to the s-SILAC reference across HGSOC tumors (Fig. 1G, fig. S1C, and data file S4). The MIB-MS tumor kinome signature was composed of established HGSOC driver kinases, as well as several kinases that had no established function in or previous association with HGSOC (Fig. 1H). Moreover, a number of these unexplored kinases were elevated ≥2-fold in HGSOC tumors relative to the s-SILAC cancer cell line reference, signifying that these kinases were highly abundant in some HGSOC tumors (Fig. 1I and fig. S1D).
Loss-of-function screen targeting the MIB-MS tumor kinome signature in HGSOC cells identifies MRCKA (CDC42BPA)
Next, we selected several kinases that were previously unexplored in HGSOC from the MIB-MS tumor signature and, using targeted RNA interference (RNAi), defined their functions in growth and survival in the HGSOC cell line OVCAR4 (Fig. 2A). A number of known kinase drivers of HGSOC from the tumor kinome signature, as well as others, were used as controls. Knockdown of myotonic dystrophy kinase–related CDC42-binding kinase alpha (MRCKA; also known as CDC42BPA) inhibited cell growth to the greatest extent among unexplored kinases, similarly to established HGSOC kinases, checkpoint kinase 1 (CHEK1), protein tyrosine kinase 2 (PTK2 or FAK1), PAK4, or aurora kinase A (AURKA). Knockdown of eukaryotic translation initiation factor 2 alpha kinase 2 (EIF2AK2) also inhibited viability of OVCAR4 cells and established HGSOC targets. Furthermore, knockdown of MRCKA reduced cell viability to varying degrees in three additional HGSOC cell lines, whereas EIF2AK2 depletion showed no effect on these additional cell lines (Fig. 2B). Knockdown efficiency of MRCKA or EIF2AK2, and positive control CHEK1 in HGSOC cells, was confirmed by Western blot (Fig. 2C). Of particular interest, knockdown of MRCKA induced apoptosis in OVCAR4 cells based on PARP cleavage (Fig. 2D), signifying that MRCKA may be essential for HGSOC cell survival, nominating MRCKA as a plausible therapeutic kinase target in HGSOC. Knockdown of MRCKA in OVCAR4 or KURAMOCHI with siRNAs targeting distinct regions of the kinase gene (Fig. 2E) and siGENOME siRNA pools (containing individual siRNAs #4 and #8) confirmed on-target MRCKA ablation, growth inhibition, and PARP cleavage (Fig. 2, F to I).
(A) Cell-Titer Glo assay for cell viability of established HGSOC cell line transfected with siRNAs targeting the indicated kinases or with control siRNA (NT2) and cultured for 120 hours. Data were analyzed as Z scores, presented as means of three independent assays. (B) Cell viability in three additional established HGSOC cell lines to those presented in (A), cultured and assessed after knockdown of select kinases as described in (A). Data were analyzed as % cell viability, presented as means ± SD of three independent assays. *P ≤ 0.05 by Student’s t test. (C) Knockdown efficiency of MRCKA, EIF2AK2, and CHEK1 in HGSOC cell lines presented in (B). Cells were transfected with siRNAs targeting MRCKA, EIF2AK2, and CHEK1/2 or with control siRNAs and cultured for 72 hours. MRCKA and cleaved PARP protein abundance was assessed by immunoblot every 24 hours. Blots are representative of three independent experiments. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (D) Apoptosis assessed by immunoblotting for cleaved-PARP abundance in OVCAR4 cells. Cells were transfected with siRNAs targeting MRCKA or with control siRNAs and cultured for 72 hours. MRCKA and cleaved PARP protein abundance was assessed by immunoblot every 24 hours. Blots are representative of three independent experiments. (E) Targeted spectrum in CDC42BPA-coding gene of two distinct siRNAs used in study. CDC42BPA siRNAs 1 and 2 are present in the siGENOME pools. bp, base pair. (F to I) Effect of MRCKA knockdown with two distinct siRNAs on cell viability by Cell-Titer Glo assay and apoptosis by cleaved-PARP immunoblotting in OVCAR4 (F and H) and KURAMOCHI (G and I) cells transfected with CDC42BPA-targeted siRNAs for 48 hours. Data are means ± SD of three independent experiments. *P ≤ 0.05 by Student’s t test.
MRCKA is frequently altered in HGSOC patient tumors and promotes proliferation, migration, and survival of HGSOC cells
MRCKA is an established CDC42 downstream effector kinase that has recently emerged as a therapeutic target in breast, skin, and lung cancers due to its role in actin cytoskeleton remodeling, cell migration, invasion, and metastasis (16, 17). To characterize MRCKA expression across larger numbers of patient HGSOC tumors, we queried publicly available HGSOC genomics [The Cancer Genome Atlas (TCGA)] and proteomics databases [Clinical Proteomic Tumor Analysis Consortium (CPTAC)] (1, 4). Frequent genomic alteration of CDC42BPA (44%, 259 of 594 HGSOC tumors) was observed in HGSOC tumors, with gene amplification (8%, 47 of 594) and altered gene expression (up-regulation in 26%, 157 of 594; and down-regulation in 11%, 64 of 594), but no homozygous deletions were detected (Fig. 3A). Only a single HGSOC tumor had a mutation in CDC42BPA (missense, L1719F). In addition, global proteomics evaluation of HGSOC tumors by CPTAC revealed that MRCKA was up-regulated at the protein level (16%, 27 of 174) and down-regulated (14%, 25 of 174), demonstrating correlation between CDC42BPA mRNA and protein levels (r = 0.53; fig. S2A). Levels of CDC42BPA mRNA expression were not statistically associated with survival, stage, or molecular subtype (fig. S2, B to D); however, tumors with high levels of CDC42BPA did show notable pathway enrichment for CDC42 signaling, as well as focal adhesion, integrin, RHOA, RAC1, and inositol phosphate signaling (Fig. 3B and fig. S2E). Conversely, low CDC42BPA-expressing tumors showed negative enrichment for these pathways but were instead enriched for DNA repair signaling.
(A) Analysis of CDC42BPA copy number analysis (CNA), mRNA, and protein expression changes in HGSOC tumors from TCGA and CPTAC studies (1, 18). mRNA levels were determined by U133 microarray, and protein level was determined by MS. Change in MRCKA abundance among tumors was determined at Z score >1 or <−1. Molecular subtype of HGSOC and disease stage are presented in relation to CDC42BPA mRNA levels. (B) Gene set enrichment analysis of HGSOC tumors expressing high or low CDC42BPA mRNA levels. Detailed description of bioinformatics methods is described in Materials and Methods. (C and D) IHC analysis of MRCKA in HGSOC TMAs. Immunoreactivity of MRCKA protein was evaluated by a pathologist, and IHC score was given based on the intensity of protein stain. 0: negative; 1+: mild (low)/weak; 2+: moderate/intermediate; 3+: strong (high)/intensive. MRCKA IHC images were captured at 100 μm. Data are from duplicate analysis of 105 tumor tissues. (E) MRCKA protein abundance among HGSOC cell lines determined by immunoblot. Blots are representative of three independent experiments. (F) Cell-Titer Glo assay for cell viability of HGSOC cells transfected with siRNAs targeting MRCKA or control siRNAs and cultured for 120 hours. Data were analyzed as % cell viability of siRNA control-treated cells, presented as means ± SD of three independent assays. (G) Immunoblot analysis for cell cycle markers in OVCAR4 cells transfected with MRCKA or control siRNAs for 24, 48, or 72 hours. Blots are representative of two independent experiments. (H) FACS analysis of OVCAR4 cells transfected with MRCKA or control siRNAs collected at the indicated time points. The cells were analyzed with a FACScan (BD) flow cytometer, and the data were analyzed using FlowJo (BD). Data were analyzed as % phase of cell cycle, presented as means ± SD of three independent assays. *P ≤ 0.05 by Student’s t test. (I) Apoptosis assessed by immunoblotting for cleaved-PARP abundance in HGSOC cells transfected with siRNAs targeting MRCKA or control siRNA and cultured for 72 hours. PARP cleavage and MRCKA protein levels were determined by Western blot. Blots are representative of three independent experiments. (J) Cell-Titer Glo assay for cell viability of FTSECs transfected with siRNAs targeting MRCKA or control siRNAs and cultured for 120 hours. Data were analyzed as % cell viability of siRNA control-treated cells, presented as means ± SD of three independent assays. (K) Apoptosis assessed by immunoblotting for cleaved-PARP abundance in FTSECs transfected with siRNAs targeting MRCKA or control siRNA and cultured for 72 hours. Blots are representative of three independent experiments. (L) Immunoblot analysis of MLC2 and MYPT1 phosphorylation in COV362 and KURAMOCHI cells transfected with MRCKA or control siRNAs for 72 hours. Blots are representative of three independent experiments. (M) Analysis of cell migration in COV362 cells transfected with MRCKA or control siRNAs and cultured for 72 hours. Migration was monitored over a 24-hour period using the xCELLigence Real-Time Cell Analyzer (RTCA). Data were analyzed as cell index, presented as means ± SD of three independent assays. *P ≤ 0.05 by Student’s t test. Related data (impact of MRCKA expression on overall survival of HGSOC, gene set enrichment analysis, IHC of MRCKA in fallopian tube and ovarian surface epithelium, and immunoblot analysis of MRCKA protein abundance in HGSOC tumors and HGSOC cell lines) can be found in fig. S2.
Widespread expression of MRCKA among HGSOCs was confirmed by immunohistochemical (IHC) analysis of tissue microarrays (TMAs) containing 105 HGSOC patient tumors (Fig. 3C and data file S5). Nearly all HGSOC tumor cores stained positive for MRCKA with 30.5% of tumors exhibiting strong MRCKA staining (3+), 52.4% showing moderate staining (2+), and 15.2% showing weak/low MRCKA protein staining (1+), while only 1.9% of tumors showed no MRCKA staining (Fig. 3D). MRCKA protein was localized to the membrane and cytoplasmic regions of the tumor with predominant staining in the cytoplasm. MRCKA abundance was also detected in normal fallopian tube epithelium and, to a lesser extent, in ovarian surface epithelium (fig. S2, F to I).
To establish growth and survival functions of MRCKA in HGSOC, we performed loss-of-function siRNA studies targeting CDC42BPA (or MRCKA) in a panel of 10 ovarian cancer cell lines likely to be HGSOC based on genomic similarities with HGSOC tumors (18). Protein expression of MRCKA was detected across HGSOC cell lines, with most cells expressing MRCKA at similar protein levels, with the exception of OVCAR4, OVCAR3, and JHOS-2, which had lower levels (Fig. 3E and fig. S2J). Knockdown of MRCKA reduced cell viability across most of the HGSOC cell lines, with SNU119, JHOS-4, OVCAR3, OVCAR4, OVCAR8, and KURMAOCHI cells exhibiting >50% growth inhibition (Fig. 3F). Knockdown of MRCKA in OVCAR4 cells reduced MYC and cyclin D1 protein abundance and increased cyclin E1 and p21 protein abundance within 48 hours, suggesting a cell cycle arrest (Fig. 3G). Fluorescence-activated cell sorting (FACS) analysis of MRCKA knockdown in OVCAR4 cells confirmed an increase in the percentage of cells in G1-S phase relative to cells transfected with control siRNA (Fig. 3H).
MRCKA knockdown resulted in increased amounts of cleaved PARP in KURAMOCHI, OVSAHO, SNU119, OVCAR4, OVCAR3, JHOS-4, and OVCAR8 cells, but not in COV362, CAOV3, or JHOS-2 cells, demonstrating that in 7 of 10 ovarian cancer cell lines, MRCKA depletion induces apoptosis (Fig. 3I). Emerging findings suggest that HGSOC arises from fallopian tube secretory epithelial cells (FTSECs) (19). To determine the effect of MRCKA disruption on putative HGSOC progenitor cells, siRNA-mediated depletion of MRCKA was evaluated in two independent immortalized human FTSEC lines (20) and did not block cell viability >50% or induce apoptosis, suggesting that MRCKA has an emergent dependency in some HGSOC cell lines (Fig. 3, J and K).
MRCKA has been shown to regulate actin remodeling, migration, and invasion through phosphorylation of MYPT1 and/or myosin light chain (MLC) in breast and lung cancer cells (21, 22). Consistent with these findings, MRCKA knockdown resulted in reduced phosphorylation of MLC2 (at Thr18/Ser19) and/or MYPT1 (at Thr696) in HGSOC cells (Fig. 3L). Next, we assayed cell migration in response to MRCKA knockdown in COV362 cells, which do not undergo apoptosis after MRCKA depletion in the time frame migration was assessed. Depletion of MRCKA inhibited migration of COV362 cells (Fig. 3M), supporting MRCKA as an important mediating kinase in HGSOC cell migration. Levels of phosphorylated MLC2 (at Thr18/Ser19) or MYPT1 (at Thr696) in HGSOC cell lines were not predictive of sensitivity toward MRCKA knockdown in HGSOC cells, suggesting that MRCKA may have additional functions in HGSOC cells apart from its recognized role in actin remodeling (fig. S2, K to M).
Knockdown of MRCKA substantially rewires kinase signaling in HGSOC cells
To understand the molecular consequences of MRCKA inhibition in HGSOC cells, we carried out a detailed proteomics analysis after MRCKA knockdown in OVCAR4 cells using MIB-MS kinome profiling and global phosphoproteomics (TiO2) (Fig. 4A) (23). At 48 hours after knockdown, both induction and repression of kinase MIB binding were observed, demonstrating that MRCKA depletion dynamically reprograms kinase signaling in HGSOC cells (Fig. 4B, fig. S3A, and data file S6). MRCKA protein depletion after siRNA-mediated gene knockdown resulted in reduced MIB binding of established kinase drivers known to promote HGSOC growth and survival [including CHEK1, PTK2 (FAK1), and AURKA] and induced MIB binding of epidermal growth factor receptor (EGFR) and protein kinase B2 (AKT2). Similar to MIB-MS profiling, Kinase Substrate Enrichment Analysis (KSEA) (24) of control and MRCKA knockdown phosphoproteomics datasets predicted inhibition of cell cycle kinases in MRCKA-depleted cells (Fig. 4C; fig. S3, B to D; and data file S7). Activation of several kinases was predicted by KSEA after MRCKA knockdown, including PAK1.
(A) Application of MIB-MS and global phosphoproteomics to explore the consequence of MRCKA knockdown on HGSOC signaling. (B) Volcano plot depicts kinases with induced or repressed MIB binding after MRCKA knockdown. OVCAR4 cells were transfected with siRNAs targeting MRCKA or control siRNAs for 48 hours and subjected to MIB-MS analysis. Statistical differences in kinase log2 s-SILAC ratios comparing MRCKA relative to control siRNA were determined by analysis of variance (ANOVA) Benjamini-Hochberg (BH) adjusted P values at an FDR of <0.05. Kinase log2 s-SILAC ratios were determined by comparing ratio of ratios (siMRCKA/s-SILAC relative siControl/s-SILAC). MIB-MS profiling was performed in biological duplicate. (C) Volcano plot depicts kinases predicted to be activated or inhibited in response to MRCKA knockdown using Kinase Substrate Enrichment Analysis (KSEA). Phosphoproteomics datasets were queried using PhosphoSitePlus at a P value cutoff of 0.05, a NetworkKIN score cutoff of 2, and a set substrate count cutoff of 10. Statistical differences in log2 s-SILAC ratios of phosphosites comparing siMRCKA/s-SILAC relative to siControl/s-SILAC were determined by paired t test, Benjamini-Hochberg adjusted P values at an FDR of <0.05. Phosphoproteomics profiling was performed in biological duplicate. (D) Immunoblot analysis of kinase remodeling in OVCAR4 cells transfected with MRCKA or control siRNAs and cultured for 72 hours. Protein abundance and phosphorylation levels were assessed 24, 48, or 72 hours after MRCKA knockdown. Blots are representative of three independent experiments. (E) Densitometric analysis of immunoblots presented in (D). Values indicate the optical density of total protein levels from an immunoblot normalized to total protein content (loading control, GAPDH). Quantitation of immunoblot bands was performed in ImageJ using three independent biological replicates. *P ≤ 0.05 by Student’s t test. Related data (hierarchical clustering of MIB-MS and phosphoproteomics profiles, a KSEA bar graph, and additional immunoblot and densitometry analysis) can be found in fig. S3.
Immunoblot analysis confirmed a time-dependent reduction in protein abundance of FAK1, CHEK1, and AURKA after MRCKA knockdown, as well as an increase in total EGFR protein levels (Fig. 4, D and E). The decrease in phosphorylation of FAK1 (at Tyr397) and increase in phosphorylation of EGFR (at Tyr1068) in response to MRCKA depletion were due to corresponding changes in total FAK1 or EGFR protein levels, respectively (fig. S3E). A visible increase in phosphorylation of PAK1 (at Ser144) was detected 24, 48, and 72 hours after MRCKA knockdown by immunoblot with minimal change in total PAK1 protein levels (Fig. 4D).
Together, the data suggest that MRCKA knockdown remodeled the kinome of HGSOC cells in a manner that affected several established HGSOC signaling pathways, including the repression of FAK1 and CHEK1 with an induction of PAK1 and others. However, as OVCAR4 cells exhibited dependency on MRCKA for viability, subsequent validation studies will be required to parse out kinase changes directly due to MRCKA knockdown versus those associated with cellular stress or apoptosis.
Genetic depletion of MRCKA impairs focal adhesion signaling and sensitizes HGSOC cells to carboplatin and PAK inhibitors
Reduced abundance of FAK1 protein and phosphorylation of FAK1 (at Tyr397) was observed by immunoblot in all HGSOC cells after MRCKA ablation (Fig. 5, A and B). This result was confirmed using two distinct MRCKA siRNAs (fig. S4A). Decreased levels of FAK1 phosphorylation in response to MRCKA knockdown were due to changes in total FAK1 protein levels for most of the HGSOC cell lines, with the exception of OVCAR8 and JHOS-4 cells, where MRCKA knockdown reduced phosphorylation of FAK1 (at Tyr397) in addition to FAK1 protein levels (fig. S4B). A minimal, insignificant reduction in PTK2 mRNA expression was observed in OVCAR4 cells after MRCKA knockdown (fig. S4C), indicating that MRCKA protein loss likely affected FAK1 at the posttranslational, protein level. In addition, we observed a time-dependent decrease in FAK1-mediated phosphorylation of its substrate paxillin at (Tyr118) after MRCKA knockdown (Fig. 5C). Furthermore, knockdown of MRCKA reduced filamentous actin fibers and the formation of focal contacts in HGSOC cells, supporting an integral role for MRCKA in facilitating focal adhesion signaling in HGSOC cells (Fig. 5, D and E, and fig. S4D). Clusters of viable tumor cells (spheroids) in patients with HGSOC represent a major component of chemoresistant and recurrent disease (25). Knockdown of MRCKA in HGSOC cells grown in three-dimensional (3D) nonadherent cultures reduced spheroid formation, supporting MRCKA as a potential therapeutic target for treatment of residual disease (Fig. 5, F and G).
(A) Immunoblot analysis of FAK1 activating phosphorylation and total FAK1 protein abundance in HGSOC cells transfected with control or MRCKA siRNAs for 72 hours. Blots are representative of three independent experiments. (B) Densitometric analysis of immunoblots presented in (A). Values indicate the optical density of total protein levels from an immunoblot normalized to total protein content (loading control, GADPH) expressed as a percent change (si-MRCKA/si-control). Quantitation of immunoblot bands was performed in ImageJ using three independent biological replicates. (C) Immunoblot analysis of focal adhesion signaling markers in COV362 cells transfected with MRCKA or control siRNAs for 24, 48, or 72 hours. Blots are representative of two independent experiments. (D) Confocal fluorescence microscopy of focal adhesion and actin cytoskeleton in COV362 cells transfected with MRCKA or control siRNAs for 72 hours. DAPI: nuclear staining by DAPI; vinculin: focal contacts revealed by anti-vinculin antibody; actin: F-actin detected by TRITC-conjugated phalloidin; merge: merged stain of DAPI, phalloidin, and vinculin. Images are representative of three independent experiments. Scale bars, 50 μm. (E) Quantitation of focal adhesion contacts in COV362 and KURAMOCHI cells transfected with MRCKA or control siRNAs for 72 hours. Data were analyzed as number of focal contacts, presented as means of three independent assays. *P ≤ 0.05 by Student’s t test. (F and G) Assessment of spheroid formation in COV362 and KURAMOCHI cells transfected with MRCKA or control siRNAs for 72 hours. Representative microscopy images of KURAMOCHI or COV362 spheroids from three independent biological replicates. Scale bars, 1000 μm. (G) Quantitation of spheroid density determined by ImageJ. *P ≤ 0.05 by Student’s t test. (H and I) Cell-Titer Glo assay for cell viability of KURAMOCHI (H) or COV362 (I) cells transfected with siRNAs targeting MRCKA or control siRNAs, cultured for 72 hours (KURAMOCHI) or 120 hours (COV362) and treated with increasing doses of carboplatin or DMSO. Data were analyzed as percentage of DMSO control, presented as means of three independent assays. GI50 were determined using PRISM. (J) Apoptosis assessed by immunoblotting for cleaved-PARP abundance in HGSOC cells transfected with siRNAs targeting MRCKA or with control siRNAs, cultured for 72 hours, and treated with carboplatin (25 μM) or DMSO. Blots are representative of two independent experiments. (K) Immunoblot analysis of phosphorylated and total levels of PAK1 in HGSOC cells transfected with MRCKA or control siRNAs for 72 hours. Blots are representative of three independent experiments. *P ≤ 0.05 by Student’s t test. (L) Cell-Titer Glo assay for cell viability of OVCAR4 or COV362 cells transfected with siRNAs targeting MRCKA or control siRNAs, cultured for 120 hours, and treated with PF-03758309 (3.5 nM OVCAR4 or 50 nM COV362) or DMSO. Data were analyzed as percentage of DMSO control, presented as means of three independent assays. Related data (immunoblot analysis of FAK1 protein levels using two distinct siRNAs, FAK1 mRNA levels after MRCKA knockdown, confocal fluorescence microscopy of focal adhesion and actin cytoskeleton in KURAMOCHI cells, and Bliss synergy analysis of carboplatin and MRCKA siRNA drug synergy) can be found in fig. S4.
Oncogenic FAK signaling has been shown to promote chemotherapy resistance in HGSOC, and inhibition of FAK1 sensitizes HGSOC cells to DNA-damaging agents such as platinum (26). On the basis of our observations that MRCKA knockdown reduced FAK1 activity in HGSOC cells, we hypothesized that MRCKA depletion may enhance sensitivity toward chemotherapy agents. Cotreatment of KURAMOCHI or COV362 cells with MRCKA siRNAs and carboplatin was synergistic at blocking cell viability (Fig. 5, H and I, and fig. S4, E to F). In addition, the combination of MRCKA siRNAs and carboplatin induced a greater amount of PARP cleavage in HGSOC cell lines than did knockdown or drug alone, supporting MRCKA as a candidate therapeutic target to be combined with chemotherapy agents (Fig. 5J).
Kinome reprogramming or rewiring in response to kinase inhibition represents an established resistance mechanism in cancer cells (11). Consistent with this idea, MIB-MS profiling and phosphoproteomics analysis of MRCKA inhibition in OVCAR4 cells had revealed that MRCKA depletion resulted in enhanced activity of several kinases, including PAK1/2 (described above). MRCKA knockdown increased phosphorylation of PAK1 (at Ser144) in OVCAR4 and COV362 but not in OVCAR3, KURAMOCHI, or OVCAR8 cells (Fig. 5K), suggesting that MRCKA inhibition may induce a compensatory response involving Rac1-PAK1/2 signaling in some HGSOC cells. Because both MRCKA and PAK1 compete for CDC42 binding in cells (27), knockdown of MRCKA may enable more binding of PAK1 to CDC42, thereby resulting in PAK1 activation. Cotreatment of OVCAR4 or COV362 cells with MRCKA siRNAs and pan-PAK inhibitor PF-03758309 (28) inhibited growth beyond MRCKA knockdown alone, suggesting that cotargeting MRCKA and PAKs may be more effective at inhibiting viability than targeting MRCKA alone in some HGSOC cells (Fig. 5L).
BDP9066 represents a highly selective MRCKA and MRCKB inhibitor for the treatment of HGSOC cells
Our RNAi-based studies revealed a role for MRCKA in HGSOC cell growth, migration, and survival through promoting a variety of oncogenic signaling, highlighting MRCKA as a promising kinase drug target in HGSOC. Recently, selective kinase inhibitors, BDP5290 and BDP9066, targeting MRCKA and related kinase myotonic dystrophy kinase–related CDC42-binding kinase beta (MRCKB or CDC42BPB), respectively, were developed and shown to block migration and invasion of cancer cells (16, 21). Kinase profiling of selectivity of BDP5290 revealed little activity against MRCKA or MRCKB, but some presumably off-target effects on several kinases, including AURKB (fig. S5A and data file S8). Conversely, in vitro kinase assays confirmed BDP9066 as a potent inhibitor of MRCKA and MRCKB, with IC50 (half-maximal inhibitory concentration) values of 65 and 72 nM, respectively (Fig. 6, A and B). Moreover, MIB-MS kinome profiling of OVSAHO cells treated with 2 μM BDP9066 for 4 hours revealed selective inhibition of MRCKA and MRCKB among the MIB-bound kinome (Fig. 6C and data file S9). In comparison, treatment of OVSAHO cells with A-674563, a pan-kinase inhibitor that targets MRCKA (29), reduced MIB binding of MRCKA and MRCKB, as well as several other kinases (fig. S5B and data file S9). Treatment of HGSOC cells with increasing doses of BDP9066 for 4 hours reduced phosphorylation of MRCKA substrate MLC2 at Thr18/Ser19, confirming inhibition of MRCKA kinase activity (Fig. 6D). BDP9066 treatment reduced filamentous actin staining in HGSOC cells and reduced migration of COV362 and KURAMOCHI cells (Fig. 6, E and F, and fig. S5, C and D), demonstrating that small-molecule targeting of MRCKA using BDP9066 inhibits actin remodeling in HGSOC cells.
(A and B) In vitro radioisotope-based 33P assays testing the effect of BDP9066 on MRCKA and MRCKB activity. BDP9066 was tested in a 10-dose IC50 mode with threefold serial dilution starting at 10 μM. Reactions were carried out at 1 μM ATP. Curve fits were performed where MRCKA or MRCKB activities at the highest concentration of BDP9066 were less than 65%. Hill slope and EC50 values were determined in molar (M). (C) MIB-MS kinome profile of OVSAHO cells treated with BDP9066 for 4 hours. SILAC-labeled OVSAHO cells were treated with 2 μM BDP9066 or DMSO for 4 hours, and lysates were incubated with CTx-0249885 beads. Line graph depicts average SILAC-determined log2 fold changes in kinase MIB binding as a ratio of BDP9066/DMSO. Biological duplicates of SILAC heavy-BDP9066/light-DMSO or light-BDP9066/heavy-DMSO were performed. (D) Immunoblot analysis of MLC2 phosphorylation in HGSOC cells treated with increasing concentrations of BDP9066 (0, 0.5, 1, and 3 μM) for 4 hours. Blots are representative of three independent experiments. (E) Confocal fluorescence microscopy of actin cytoskeleton in COV362 treated with DMSO or 1 μM BDP9066 for 48 hours. F-actin was detected by TRITC-conjugated phalloidin, and nuclear staining was detected by DAPI. Images are representative of three independent experiments. Scale bars, 50 μm. (F) Analysis of cell migration in COV362 cells treated with DMSO or 1 μM BDP9066. Migration was monitored over a 24-hour period using the xCELLigence RTCA. Data were analyzed as cell index, presented as means ± SD of three independent assays. *P ≤ 0.05 by Student’s t test. Related data (in vitro kinase assays testing BDP5290 selectivity, a comparison of BDP9066 and A-674563 MIB-MS profiles, and the consequence of BDP9066 treatment on actin cytoskeleton and migration of KURAMOCHI cells) can be found in fig. S5.
MRCKA inhibition induces apoptosis and blocks spheroid formation, establishing MRCKA as a therapeutic target in HGSOC
Next, we explored the impact of small-molecule inhibition of MRCKA kinase activity on cell growth and survival. BDP9066 treatment inhibited cell viability to varying degrees across the HGSOC cell lines, with KURAMOCHI, OVCAR8, SNU119, and OVCAR3 among the most affected by MRCKA inhibition, exhibiting GI50 (growth inhibition of 50%) values <0.5 μM (Fig. 7A). Moreover, BDP9066 reduced colony formation of HGSOC cells with nearly complete inhibition of colonies at a dose of 0.312 μM in SNU119, OVCAR8, OVCAR3, KURAMOCHI, and OVCAR4 (Fig. 7B). These cell lines were among the most sensitive to MRCKA knockdown, demonstrating the overlap between small-molecule inhibition and genetic depletion of MRCKA. BDP9066 treatment increased levels of cleaved PARP in SNU119, OVCAR8, OVCAR3, and KURAMOCHI cells (Fig. 7C), indicating that small-molecule inhibition of MRCKA kinase activity can trigger apoptosis in HGSOC cells. COV362 cells showed no increase in cleaved PARP levels, consistent with MRCKA knockdown findings. Because MRCKB is also targeted by BDP9066, we depleted MRCKB using siRNAs in HGSOC cells to determine whether MRCKB inhibition contributed to the observed response. Knockdown of MRCKB marginally inhibited cell viability (~40% growth inhibition) of OVCAR4, JHOS-4, and OVCAR3, but not other HGSOC cell lines (fig. S6, A and B), suggesting that MRCKA inhibition is likely the major contributing factor to the observed anti-growth effects of BDP9066.
(A) Cell-Titer Glo assay for cell viability of HGSOC cell lines treated with increasing concentrations of BDP9066 or DMSO and cultured for 120 hours. Data were analyzed as % cell viability of DMSO control, presented as means of three independent assays. GI50 values for BDP9066 were generated in Prism. (B) Long-term 14-day colony formation assay of HGSOC cells treated with BDP9066 or DMSO. Colony formation was assessed by crystal violet staining. Representative images of three biological replicates are shown. (C) Apoptosis assessed by immunoblotting for cleaved-PARP abundance in HGSOC cells treated with DMSO or 1 μM BDP9066 for 72 hours. Blots are representative of three independent experiments. (D) Immunoblot analysis of FAK1 phosphorylation and total FAK1 protein abundance in HGSOC cells treated with escalating doses of BDP9066 (0, 0.5, 1, and 3 μM) for 48 hours. Blots are representative of three independent experiments. (E) Immunoblot analysis of phosphorylation of FAK1 and paxillin in HGSOC cells treated with 2 μM BDP9066 for 48 hours. Blots are representative of three independent experiments. (F) Densitometric analysis of immunoblots presented in (E). Values indicate the optical density of phosphorylated or total protein levels from an immunoblot normalized to total protein content expressed as a percent change (BDP9066/DMSO). Phosphorylated proteins were normalized to loading control and then normalized to total abundance of the respective protein. Quantitation of immunoblot bands was performed in ImageJ using three independent biological replicates. *P ≤ 0.05 by Student’s t test. (G) Confocal fluorescence microscopy of focal adhesion and actin cytoskeleton in KURAMOCHI cells treated with DMSO or 1 μM BDP9066 for 24 hours. DAPI: nuclear staining by DAPI; vinculin: focal contacts revealed by anti-vinculin antibody; actin: F-actin detected by TRITC-conjugated phalloidin; merge: merged stain of DAPI, phalloidin, and vinculin. Images are representative of three independent experiments. Scale bars, 50 μm. (H and I) Assessment of spheroid formation in HGSOC cells treated with DMSO or 1 μM BDP9066 for 72 hours. Representative images of HGSOC cell spheroids (H) from three independent biological replicates are shown. Scale bars, 1000 μm. (I) Quantitation of spheroid density determined by ImageJ. *P ≤ 0.05 by Student’s t test. (J) Model of the mechanism, depicting the integral role of MRCKA in promoting proliferation, migration, and survival of HGSOC cells, thus highlighting MRCKA as a new kinase inhibitor target for the treatment of HGSOC. Related data (knockdown of MRCKB’s impact on HGSOC cell viability, drug synergy studies involving BDP9066 and carboplatin or PAK1 inhibitors, as well as quantitation of number of focal contacts in KURAMOCHI cells in response to BDP9066) can be found in fig. S6.
In our loss-of-function studies (described above), we showed that MRCKA knockdown sensitized HGSOC cells to carboplatin or PAK inhibitors. In agreement, cotreatment of HGSOC cells with BDP9066 and carboplatin improved growth inhibition relative to single agents in HGSOC cell lines trending toward drug synergy (fig. S6, C to F). Similar to MRCKA knockdown, increased activating phosphorylation of PAK1 (at Ser144) was observed in response to BDP9066 treatment in COV362 cells, and combination therapies involving BDP9066 and PAK1 inhibitor FRAX1036 (30) further inhibited growth compared to single-agent drugs, with a Bliss synergy score of 6.4 (fig. S6, G and H). Thus, PAK1 induction may represent a potential compensatory resistance mechanism to BDP9066 therapy in some HGSOC cells, requiring combination therapies that cotarget MRCKA and PAK1.
Treatment of HGSOC cells with escalating doses of BDP9066 reduced activating phosphorylation of FAK1 (at Tyr397) but did not lower total protein abundance of FAK1 that was observed with MRCKA depletion, signifying that MRCKA may have both kinase-dependent and kinase-independent functions in focal adhesion signaling (Fig. 7D). Treatment of KURAMOCHI cells with BDP9066 decreased phosphorylation of FAK1 at (Tyr397) and reduced both total paxillin levels and phosphorylation of paxillin at (Tyr118) (Fig. 7, E and F). Moreover, BDP9066 treatment reduced the number of focal contacts relative to control-treated cells, demonstrating that small-molecule inhibition of MRCKA kinase activity blocks focal adhesion signaling similarly to MRCKA knockdown (Fig. 7G and fig. S6I). Furthermore, treatment of HGSOC cell lines grown in 3D cultures with BDP9066 blocked spheroid formation (Fig. 7, H and I), supporting BDP9066 as a promising drug for the treatment of HGSOC, including metastatic ascites. BDP9066 treatment had no impact on spheroid formation of OVSAHO cells, indicating that some HGSOC cells are intrinsically resistant to small-molecule inhibition of MRCKA.
DISCUSSION
Protein kinases are arguably one of the most important regulatory enzymes in cells, and drugs targeting these enzymes have led to great improvements in treating cancer. However, the vast majority of kinase publications focus on a comparatively small group of well-understood kinases, yet synthetic lethality screens repeatedly identify various understudied kinases as playing a vital role in cancer cell proliferation and survival (31). This is particularly evident in ovarian cancer, where only a small percentage of the kinome has been therapeutically explored. Here, we profiled the kinome in HGSOC tumors using MIB-MS and identified MRCKA, a CDC42 effector kinase not previously linked to HGSOC to be highly abundant in HGSOC patient and PDX tumors. Loss-of-function studies targeting MRCKA in HGSOC cell lines revealed an essential role for MRCKA in HGSOC cell proliferation, survival, and migration, establishing MRCKA as a plausible drug target in this disease (Fig. 7J).
In addition to MRCKA’s established role in actin remodeling (32), here, we provide evidence that MRCKA protein is essential for FAK1 activity and protein stability, because knockdown of MRCKA reduced both activating FAK1 phosphorylation and total protein levels independent of RNA changes. Moreover, focal adhesions were reduced in HGSOC cells after MRCKA inhibition blocking spheroid formation, supporting a previously unknown role for MRCKA in focal adhesion signaling. Future experiments assessing focal adhesion formation after forced expression of FAK1 or constitutively active FAK1 mutants in MRCKA-depleted cells will be required to establish a causal link of MRCKA with FAK1 in focal adhesion signaling. We have extensively characterized the loss of FAK1 protein in response to MRCKA knockdown, but the mechanism by which MRCKA protein depletion influences FAK1 protein levels in HGSOC cells has not yet been defined. FAK1 protein stability has been shown to be controlled by E3 ligase activity regulating focal adhesion signaling in cells (33). Evaluation of FAK1 ubiquitination and E3 ligase activity after MRCKA depletion could provide insight into the mechanism linking MRCKA activity and FAK1 protein stability. One could envisage MRCKA directly or indirectly regulating a critical E3 ligase that controls FAK1 protein levels, where blockade of MRCKA induces E3 ligase activity promoting FAK1 degradation.
Intriguingly, we observed inhibition of MRCKA kinase activity using BDP9066 reduced phosphorylation of FAK1 but did not reduce FAK1 protein abundance as observed with MRCKA knockdown. These findings suggest that MRCKA may have kinase-independent functions that promote FAK1 protein stability and focal adhesion signaling in HGSOC cells. Rescue experiments expressing kinase-dead MRCKA in cells depleted of MRCKA protein could help parse kinase-dependent versus kinase-independent functions of MRCKA in HGSOC cells. Additional proteomics studies investigating whether MRCKA influences FAK1 activity and/or stability directly through phosphorylation or indirectly through protein interaction will be required. MRCKA is a large protein that has several binding domains (34), where MRCKA may function as a scaffold facilitating FAK1 activity in focal adhesion signaling. Recently, proteolysis targeting chimeras (PROTACs) have been used to selectively degrade proteins providing a therapeutic strategy to block both the kinase-dependent and kinase-independent functions of a specific kinase (35). This was recently applied to FAK1, which showed that degradation of FAK1 was superior to kinase inhibition alone, owing to the blockade of both FAK1 kinase signaling and scaffolding capabilities (36). Development of PROTACs that degrade MRCKA may provide a promising technique to gain insight into MRCKA’s kinase-dependent and kinase-independent functions, as well as offer a therapeutic strategy to block both MRCKA kinase and nonkinase functions in HGSOC.
Tumors often bypass targeted kinase inhibition by activating alternative kinase-mediated survival pathways, a process termed kinome reprogramming (37). Proteomic characterization of MRCKA-depleted HGSOC cells showed activation of several kinases including PAK1, suggesting a potential kinome reprogramming mechanism linked to MRCKA protein loss in HGSOC cells. Previous studies have shown that both MRCKA and PAK1 compete for CDC42 binding in cells (27); therefore, knockdown of MRCKA protein may facilitate binding of PAK1 to CDC42, resulting in PAK1 activation. Because PAK1 is an established oncogenic kinase in ovarian cancer (38), combination therapies blocking PAK1 feedback activation may be required to achieve durable responses when targeting MRCKA in some HGSOC cells. Small-molecule inhibitors of CDC42/RAC1 have emerged that can block cell migration and metastasis in cancer models (39). Targeting upstream of MRCKA using a dual CDC42/RAC1 inhibitor such as MBQ-167 (39) may represent an effective strategy to shut down both MRCKA and PAK signaling in HGSOC cells.
Our evaluation of current MRCK inhibitors revealed that BDP5290 showed minimal activity against MRCK kinases, whereas BDP9066 was a highly selective and potent inhibitor of MRCKA and MRCKB in HGSOC cells. Of particular interest, MIB-MS analysis of HGSOC cells treated with 2 μM BDP9066 showed selective reduction in MRCKA and MRCKB MIB binding (of 216 kinases) but not the closely related kinase ROCK1, which has similar protein structure overlap (21). The MRCK family of kinases consists of three members: MRCKA, MRCKB, and myotonic dystrophy kinase–related CDC42-binding kinase gamma (CDC42BPG or MRCKG), with the latter exhibiting a more tissue-specific expression pattern (34). We did not detect MRCKG in our proteomics analysis of BDP9066 selectivity in OVSAHO cells; however, further studies will be required to determine whether MRCKG is expressed and/or inhibited by BDP9066 in other HGSOC cells. Knockdown of MRCKB reduced the proliferation of some HGSOC cell lines, suggesting that inhibition of MRCKB may contribute to the overall efficacy of BDP9066. Future studies exploring the overlapping or distinct functions of these related MRCK kinases in HGSOC signaling will be of particular interest.
Malignant ascites, a buildup of fluid containing cancer cells in the abdomen, often occurs in patients with late-stage HGSOC with residual chemoresistant disease, causing substantial morbidity and death (25). Tumor cells within ascites of patients with ovarian cancer often grow as spheroids or aggregates of tumor cells, which are typically difficult to treat with conventional therapies (40). Our findings showed that BDP9066 treatment impaired focal adhesion signaling, resulting in reduced spheroid formation of HGSOC cells grown in 3D cultures, suggesting that MRCKA inhibition may have therapeutic potential in late-stage patients with residual disease. Of particular interest, our proteogenomics analysis of HGSOC patient tumors revealed that MRCKA was overexpressed in ~25% of tumors, and these tumors were enriched for CDC42 and FAK1 signaling. Preclinical studies in HGSOC tumor xenograft models, including those overexpressing MRCKA, will be required to test the efficacy of BDP9066 therapy to inhibit tumor growth and cause tumor regression, as well to define potential toxicities associated with targeted MRCKA inhibition.
MATERIALS AND METHODS
Cell lines
Cell lines were verified by IDEXX laboratories and were verified mycoplasma negative (7 January 2019) using the Hoechst DNA stain method. OVCAR4, KURAMOCHI, OVSAHO, and OVCAR8 cell lines were maintained in RPMI 1640 supplemented with 10% fetal bovine serum (FBS), penicillin-streptomycin (100 U/ml), and 2 mM GlutaMAX. OVCAR3 cells were maintained in RPMI 1640 supplemented with 10% FBS, penicillin-streptomycin (100 U/ml), 2 mM GlutaMAX, and insulin (5 μg/ml). SNU-119 cells were maintained in RPMI 1640 supplemented with 10% FBS, penicillin-streptomycin (100 U/ml), 2 mM GlutaMAX, and 25 mM HEPES. COV362 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS, penicillin-streptomycin (100 U/ml), and 2 mM GlutaMAX. CAOV3 cells were maintained in DMEM supplemented with 10% FBS, penicillin-streptomycin (100 U/ml), 2 mM GlutaMAX, and 100 mM sodium pyruvate. JHOS-2 and JHOS-4 cells were maintained in DMEM/F12 supplemented with 10% FBS, penicillin-streptomycin (100 U/ml), 2 mM GlutaMAX, and nonessential amino acids. FTSECs were maintained in DMEM supplemented with 4% FBS, penicillin-streptomycin (100 U/ml), 2 mM GlutaMAX, and 1× Insulin-Transferrin-Selenium (Gibco). Cell lines used for the super-SILAC reference sample (MOLT4, UACC257, ACHN, Colo205, and PC-3) were grown for seven doublings in arginine- and lysine-depleted media supplemented with heavy isotope–labeled [13C6,15N4]arginine (Arg10) (84 mg/liter) and [13C6]lysine (Lys8) (48 mg/liter) (Sigma-Aldrich) and unlabeled leucine (50 mg/liter) (Thermo Fisher Scientific) as described previously (41). OVSAHO SILAC cells were grown for seven doublings in arginine- and lysine-depleted RPMI 1640 supplemented with either unlabeled l-arginine (84 mg/liter), l-lysine (48 mg/liter), and leucine (50 mg/liter) (Thermo Fisher Scientific) or equimolar amounts of heavy isotope–labeled [13C6,15N4]arginine (Arg10) and [13C6]lysine (Lys6) (Sigma-Aldrich). All cells were kept at 37°C in a 5% CO2 incubator.
Patient samples
All patient samples used in this study were obtained from the Fox Chase Cancer Center (FCCC) Biosample Repository Facility (BRF). The FCCC BRF maintains a long-standing Institutional Review Board–approved protocol for collection, banking, and distribution of deidentified biospecimens and associated clinical data. Consent and authorization for the use of deidentified specimens and associated clinical data for unrestricted research were obtained from all BRF participants before specimen collection.
Frozen tumor samples
Ovarian tumors from 28 chemo-naïve patients with stage III or IV HGSOC were obtained from the FCCC BRF for this Institutional Review Board–approved research (16-9031 and 14-809). Tumor tissue was snap-frozen at the time of collection and stored at −80°C in the FCCC BRF. Tumor histology and cellularity were confirmed by the FCCC BRF pathologist at the time of banking. Tumors isolated from HGSOC PDX models were snap-frozen and stored at −80°C until sample processing for MIB-MS analysis. A description of patient tumors and PDX tumor tissue details can be found in data file S2.
Compounds
A-674563, carboplatin, FRAX1036, and PF-03758309 were purchased from Selleckchem. BDP5290 and BDP9066 were purchased from Glixx Laboratories and ProbeChem, respectively (data file S10). For the compounds used in MIB synthesis, purvalanol B was purchased from Abcam. PP58 (42) and VI16832 (43) were custom-synthesized according to previously described methods by The Center for Combinatorial Chemistry and Drug Discovery, Jilin University, P.R. China. CTx-0294885 (44) was purchased from MedKoo Biosciences Inc. (406457). Conjugation of inhibitors to beads was performed by carbodiimide coupling to ECH Sepharose 4B (CTx-0294885, VI16832, and PP58) or EAH Sepharose 4B (purvalanol B) (GE Healthcare).
In vitro kinase profiling
In vitro kinase assays for BDP5290 and BDP9066 were performed by Reaction Biology Corp. (Malvern, PA). BDP5290 compound was tested in a single-dose duplicate mode at a concentration of 10 μM against 371 kinases. BDP9066 was tested against MRCKA/CDC42BPA or MRCKB/CDC42BPB. BDP9066 was tested in 10-dose mode with threefold serial dilution starting at 10 μM. Reactions were carried out at 1 μM ATP control. Staurosporine was tested in 10-dose IC50 mode with fourfold serial dilution starting at 20 or 100 μM. Alternative control compounds were tested in 10-dose IC50 mode with three- or fourfold serial dilution starting at 10, 20, or 50 μM. Data pages are presented in data file S8, include raw data % enzyme activity [relative to dimethyl sulfoxide (DMSO) controls] and curve fits. List of kinase sources used in kinase profiling can be found at www.reactionbiology.com/webapps/site/KinaseDetail.aspx?page=Kinases&id=1&filter=512.
Immunoblotting
Samples were harvested in MIB lysis buffer, subjected to SDS–polyacrylamide gel electrophoresis chromatography, and transferred to polyvinylidene difluoride membranes before Western blotting with primary antibodies. For pMYPT and pMLC2 blots, cells were harvested in a buffer containing 1% (w/v) SDS and 50 mM tris (pH 7.5) supplemented with protease and phosphatase inhibitors buffer and were passed through QIAshredder columns (Qiagen, 79654) (21). For the list of primary antibodies used, see data file S10. Secondary horseradish peroxidase anti-rabbit and horseradish peroxidase anti-mouse were obtained from Thermo Fisher Scientific. SuperSignal West Pico and Femto Chemiluminescent Substrates (Thermo Scientific) were used to visualize blots. Western blot images were quantified using the Analyze>Gels function in ImageJ open source software (National Institutes of Health). Optical density values were normalized by sum (fig. S2, G to J) or by a fixed point (Figs. 4E, 5B, and 7F) (45). Phosphorylated proteins were normalized to loading control and then to total abundance of the respective protein.
Actin cytoskeleton and focal adhesion staining
KURAMOCHI and COV362 cells were plated at a density of 2 × 105 cells per well in a six-well plate with an 18-mm2 glass coverslip inside each well. After treatment, cells were fixed and stained using the Actin Cytoskeleton/Focal Adhesion Staining Kit (Millipore, Sigma) following the manufacturer’s instructions. Anti-vinculin (1:500), tetramethyl rhodamine isothiocyanate (TRITC)–conjugated phalloidin (1:1000), and donkey anti-mouse Alexa Fluor 488–conjugated secondary antibodies (1:1000; Thermo Fisher Scientific) were applied to the coverslips for 1 hour at room temperature. After antibody incubation, coverslips were mounted on slides using ProLong Gold Antifade Reagent with DAPI (4′,6-diamidino-2-phenylindole) (Thermo Fisher Scientific) and allowed to set overnight. Images were taken with a Leica SP8 confocal microscope at ×40 magnification. Focal contacts were counted using the 3D Objects Counter tool in ImageJ.
Spheroid formation
Cells were plated at a density of 50,000 per well in 24-well ultralow attachment plates (Corning 3473). For drug treatment studies, drug was added to the cells at the time of plating, and images were taken 72 hours later on an EVOS digital inverted microscope. For siRNA studies, cells were transfected with siRNA in 2D culture plates 48 hours before the addition to the ultralow attachment plates, and images were taken 72 hours later (5 days after transfection). Spheroid area was quantified in ImageJ using the Analyze Particles tool after the color threshold was set to identify larger clusters of cells.
Drug synergy analysis
Drug synergy was determined using SynergyFinder (https://synergyfinder.fimm.fi) using the Bliss model and viability as the readout (46). Each drug combination was tested in triplicate.
Ovarian tissues and tumor specimens for IHC
Cancerous and benign ovarian specimens were obtained from patients who underwent surgical resection at FCCC. Seven ovarian tumor TMAs containing a total of 129 ovarian tumor cases in duplicate and one normal ovary and fallopian tube TMA containing a total of 24 normal ovaries and 11 fallopian tube cores in duplicate were provided by the Biosample Repository of FCCC (data file S5). All these samples were used with informed patient consent, and the study was approved by the Institutional Review Boards of FCCC.
Typically, tissue specimens were fixed in 10% buffered formalin, embedded in paraffin, sectioned, and stained with hematoxylin and eosin. All tumors were histologically classified according to the World Health Organization classification, and the surgical stages were determined according to the classification of International Federation of Gynecology and Obstetrics. Two-tiered histological grading system was also used for grading the serous carcinoma.
IHC and IHC evaluation
IHC staining was carried out according to the standard methods by the FCCC Histopathology Facility. Briefly, 5-μm formalin-fixed, paraffin-embedded TMA sections were deparaffinized and hydrated. Sections were then subjected to heat-induced epitope retrieval with 0.01 M citrate buffer (pH 6.0). Endogenous peroxidases were quenched by the immersion of slides in 3% hydrogen peroxide solution. The sections were incubated overnight with primary antibodies to MRCKA (Abcam, ab96659) (rabbit, 1:200; Abcam) (16) at 4°C in a humidified slide chamber. Immunodetection was performed using the Dako EnVision+ polymer system, and immunostaining was visualized with the chromogen 3,3′-diaminobenzidine. The sections were then washed, counterstained with hematoxylin, dehydrated with ethanol series, cleared in xylene, and mounted. Known positive cancer tissues were used as positive controls. As a negative control, the primary antibody was replaced with normal rabbit immunoglobulin G to confirm the absence of specific staining. All slides were viewed with a Nikon Eclipse 50i microscope, and photomicrographs were taken with an attached Nikon DS-Fi1 camera (Melville, NY). Immunoreactivity of MRCKA protein was evaluated by a pathologist, and IHC score was given based on the intensity of the staining (0: negative; 1: mild/weak; 2: moderate/intermediate; 3: strong/intensive).
Cell proliferation assays
For short-term growth assays, 1500 to 5000 cells were plated per well in 96-well plates and allowed to adhere and equilibrate overnight. Drug was added the following morning, and after 120 hours of drug treatment, cell viability was assessed using the CellTiter-Glo Luminescent cell viability assay according to the manufacturer (Promega). Student’s t tests were performed for statistical analyses, and P values ≤0.05 were considered statistically significant.
Quantitative reverse transcription polymerase chain reaction
GeneJET RNA Purification Kit (Thermo Fisher Scientific) was used to isolate RNA from cells according to the manufacturer’s instructions. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) on diluted complementary DNA was performed with inventoried TaqMan Gene Expression Assays on the Applied Biosystems 7500 Fast Real-Time PCR System. The TaqMan Gene Expression Assay probes (Thermo Fisher Scientific) used to assess changes in gene expression include MRCKA (assay ID: Hs00177522_m1) and PTK2 (assay ID: Hs00178587_m1).
RNAi knockdown studies
siRNA transfections were performed using 25 nM siRNA duplex and the reverse transfection protocol. A total of 1500 to 5000 cells per well were added to 96-well plates with media containing the siRNA and transfection reagent (Lipofectamine RNAiMax) according to the manufacturer’s instructions. In experiments where inhibitors were used, the inhibitor was added at the time of transfection. Cells were allowed to grow for 72 to 120 hours after transfection before CellTiter Glo (Promega) analysis. Two to three independent experiments were performed with each cell line and siRNA. Student’s t tests were performed for statistical analyses, and P values ≤0.05 were considered statistically significant. For Western blot studies, the same procedure was performed with volumes and cell numbers proportionally scaled to a 60-mm or 10-cm dish, and cells were collected 24 to 72 hours after transfection. siRNA product numbers and manufacturers are listed in data file S10.
FACS analysis of MRCKA knockdown
OVCAR4 cells were transfected with 25 nM siRNA and collected at the indicated time points. Cells were trypsinized, washed twice with phosphate-buffered saline, and fixed in ice-cold 70% ethanol for at least 24 hours. Before analysis, the fixed cells were washed once again with phosphate-buffered saline, resuspended in 300 μl of FxCycle PI/RNase (Thermo Fisher Scientific, F10797), and incubated for 30 min at room temperature. The cells were analyzed with a FACScan (BD) flow cytometer, and the data were analyzed using FlowJo (BD) at the Flow Cytometry Facility, FCCC.
Migration assay
Cell migration assays were performed using the xCELLigence Real-Time Cell Analyzer (RTCA) DP Instrument (ACEA Biosciences) in a CIM-Plate 16 (ACEA Biosciences) and the Cell Culture Facility, FCCC. Complete medium was added to the lower chambers of the CIM-Plate (or serum-free medium for negative control). In the upper chambers, 30 μl of serum-free medium was added to each well, and the CIM-Plate was placed in the instrument to equilibrate for 1 hour at 37°C. Following this equilibration step, a background measurement was taken. The CIM-Plate was removed from the instrument, and cells were added to the upper chambers at 50,000 cells per well in serum-free medium. The CIM-Plate was left at 25°C for 30 min to allow the cells to settle at the bottom of the well. The plate was then returned to the instrument, and cell index reads were taken once every 15 min for 24 hours. For the siRNA experiment, the cells were transfected 24 hours before the start of the assay. For the BDP9066 experiments, cells were pretreated with 1 μM inhibitor or DMSO 24 hours prior, and both upper and lower chambers contained 1 μM inhibitor or DMSO.
MIB preparation and chromatography
Experiments using MIB-MS were performed as previously described (13). Briefly, cells or tumors were lysed on ice in buffer containing 50 mM HEPES (pH 7.5), 0.5% Triton X-100, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 10 mM sodium fluoride, 2.5 mM sodium orthovanadate, 1× protease inhibitor cocktail (Roche), and 1% each of phosphatase inhibitor cocktails 2 and 3 (Sigma-Aldrich). Particulate was removed by centrifugation of lysates at 21,000g for 15 min at 4°C and filtration through 0.45-μm syringe filters. Protein concentrations were determined by bicinchoninic acid analysis (Thermo Fisher Scientific). To compare kinome signatures within and HGSOC tumors, we designed a super-SILAC method consisting of a cocktail of cancer cell lines that encompasses the activity of the “cancer kinome” to be used as a control for individual samples. Five cancer cell lines were selected from the NCI-60 panel (UACC257, MOLT4, COLO205, ACHN, and PC3) that differed in their origin, gene expression patterns, and mutation status. The cancer cell lines were labeled using SILAC and mixed equally providing a diverse SILAC {[13C6, 15N4]arginine (Arg10) and [13C6]lysine (Lys8)} heavy reference standard (s-SILAC) to be used repeatedly in kinome profiling assays. An equal amount of the s-SILAC reference (5 mg) lysate was added to our nonlabeled (5 mg) sample (cell or tumor tissue) and analyzed on MIB beads. Endogenous kinases were isolated by flowing lysates over kinase inhibitor–conjugated Sepharose beads (purvalanol B, VI16832, PP58, and CTx-0294885 beads) in 10-ml gravity-flow columns. After 2 × 10 ml column washes in high-salt buffer and 1 × 10 ml wash in low-salt buffer [containing 50 mM Hepes (pH 7.5), 0.5% Triton X-100, 1 mM EDTA, 1 mM EGTA, 10 mM sodium fluoride, and 1 M NaCl or 150 mM NaCl, respectively], retained kinases were eluted from the column by boiling in 2 × 500 μl of 0.5% SDS, 0.1 M tris-HCl (pH 6.8), and 1% 2-mercaptoethanol. Eluted peptides were reduced by incubation with 5 mM dithiothreitol (DTT) at 65°C for 25 min and alkylated with 20 mM iodoacetamide at room temperature for 30 min in the dark, and alkylation was quenched with DTT for 10 min. For MIB competition experiments related to Fig. 6C and fig. S5B, heavy-labeled (drug-treated or control) or light-labeled (drug-treated or control) lysates were run on MIB columns separately, eluted, and then mixed before reduction and alkylation. Samples were concentrated to about 100 μl with Millipore 10-kDa cutoff spin concentrators. Detergent was removed by chloroform/methanol extraction, and the protein pellet was resuspended in 50 mM ammonium bicarbonate and digested with sequencing-grade modified trypsin (Promega) overnight at 37°C. Peptides were cleaned with PepClean C18 spin columns (ThermoFisher Scientific), dried in a SpeedVac, resuspended in 50 μl of 0.1% formic acid, and extracted with ethyl acetate (10:1, ethyl acetate:H2O). Briefly, 1-ml ethyl acetate was added to each sample, vortexed, centrifuged at a maximum speed for 5 min, and then removed. This process is repeated four more times. After removal of ethyl acetate after the fifth centrifugation, samples were placed at 60°C for 10 min to evaporate residual ethyl acetate. The peptides were dried in a SpeedVac, and subsequent liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis was performed.
Nano–LC-MS/MS
Proteolytic peptides were resuspended in 0.1% formic acid and separated with a Thermo Fisher Scientific RSLC UltiMate 3000 on a Thermo Fisher Scientific Easy-Spray C18 PepMap 75 μm by 50 cm C-18 2-μm column with a 240-min gradient of 4 to 25% acetonitrile with 0.1% formic acid at 300 nl/min at 50°C. Eluted peptides were analyzed by a Thermo Fisher Scientific Q Exactive plus mass spectrometer using a top 15 methodology in which the 15 most intense peptide precursor ions were subjected to fragmentation. The automatic gain control for MS1 was set to 3 × 106 with a maximum injection time of 120 ms, the automatic gain control for MS2 ions was set to 1 × 105 with a maximum injection time of 150 ms, and the dynamic exclusion was set to 90 s.
Data processing and analysis
Raw data analysis of SILAC experiments was performed using MaxQuant software 1.6.1.0 and searched using Andromeda 1.5.6.0 against the Swiss-Prot human protein database (downloaded on 26 July 2018). The search was set up for full tryptic peptides with a maximum of two missed cleavage sites. All settings were default and searched using acetylation of protein N terminus and oxidized methionine as variable modifications. Carbamidomethylation of cysteine was set as fixed modification. The precursor mass tolerance threshold was set at 10 parts per million, and maximum fragment mass error was 0.02 Da. SILAC quantification was performed using MaxQuant by choosing multiplicity as 2 in group-specific parameters and Arg10 and Lys8 as heavy labels. The match between runs was used, and the significance threshold of the ion score was calculated on the basis of a false discovery rate (FDR) of <1%. For HGSOC tumor MIB-MS data analysis, MaxQuant normalized ratios were imported into Perseus software (1.6.2.3) for quantitation (data file S2). HGSOC tumor MIB-MS profiles were processed in the following manner: Normalized MIB-MS s-SILAC ratios were transformed 1/(x) to generate light/heavy ratios, followed by log2/(1/x) transformed. Columns were then filtered on the basis of a valid value of 150, and rows were filtered for presence in 70% of samples leaving 206 kinases. Imputation of missing values was performed as previously described (47), where, in the super-SILAC data, a width of 0.3 and a downshift of 0.5 were used. PCA (PC1 versus PC2, PC2 versus PC3, and PC1 versus PC3) was then performed to visualize kinome profiles among HGSOC tumors (Fig. 1, D and E, and fig. S1, A and B). Hierarchical clustering (Euclidean) of s-SILAC ratios was then performed, and column clusters were annotated selecting eight clusters at a distance threshold of 21.59 (fig. S1D). No imputation of missing MIB-MS s-SILAC values was used for clustering analysis. The prominent cluster (Cluster-27) of HGSOC tumors was then compared to other HGSOC tumors using a two-sample Student’s t test with the following parameters (S0 0.1 and Side, Both) using a P value of 0.05 (data file S4). Kinases statistically enriched in the main cluster were isolated (n = 82), and the s-SILAC ratios were hierarchical clustered and depicted in a heat map (Fig. 1G). For MRCKA knockdown MIB-MS data analysis, MaxQuant-normalized ratios were imported into Perseus software (1.6.2.3) for quantitation (data file S6). MIB-MS profiles were processed in the following manner: Normalized MIB-MS s-SILAC ratios were transformed 1/(x) to generate light/heavy ratios, followed by log2/(1/x) transformed. Rows were filtered for presence in 100% of samples leaving 168 kinases. No imputation of missing values was performed. Differences in s-SILAC ratios among cells treated with control siRNAs or MRCKA siRNAs were determined using a two-sample Student’s t test with the following parameters (S0 0.1 and Side, Both) using Benjamini-Hochberg FDR, FDR < 0.05 (data file S6). A volcano plot was then generated in RStudio 1.1.423 using ggplot2 and ggrepel libraries (Fig. 4B) to visualize differences in MIB binding among treatments. For BDP9066 and A-674563 MIB-MS data analysis, MaxQuant-normalized ratios were imported into Perseus software (1.6.2.3) for quantitation (data file S9). MIB-MS profiles were processed in the following manner: Normalized MIB-MS s-SILAC ratios were transformed 1/(x) to generate light/heavy ratios, followed by log2/(1/x) transformed. Rows were filtered for presence in 50% of samples leaving 216 kinases. No imputation of missing values was performed. Average SILAC ratio for kinases was calculated using Perseus software (1.6.2.3), and a line graph was generated to visualize the primary targets of BDP9066 (Fig. 6B). The MIB-MS data were deposited to the ProteomeXchange Consortium via PRIDE partner repository with the dataset identifier PDX015849.
Titanium dioxide enrichment and phosphoproteomics data analysis
To define kinase and kinase substrate phosphorylation at baseline or in response to kinase knockdown, we performed global phosphoproteomics analysis, as previously described (23). Using methanol/chloroform protein precipitation methods, 4 mg of lysate (2 mg of OVCAR4 si-NT2 or si-MRCKA and 2 mg of s-SILAC reference sample) was mixed and precipitated, resuspended in 360 μl of 50 mM tris-HCl (pH 8.5), and digested with 20 μl of 1% ProteaseMAX (Promega) plus 20 μg of Promega Trypsin/Lys-C mix for 4 hours at 37°C with vigorous shaking. Samples were reduced with 5 mM DTT at room temperature for 25 min and alkylated with 10 mM iodoacetamide at room temperature for 25 min in the dark, and alkylation was quenched with 5 mM DTT for 15 min. Samples were then digested with sequencing-grade modified trypsin (Promega) overnight at 37°C in an incubator shaker. Peptides were cleaned by subsequent elution from C-18 (Phenomenex) and Hypercarb/Hypersep PGC columns (Thermo Fisher Scientific) and dried via SpeedVac. Peptides were resuspended in 100 μl of 0.1% formic acid and quantified using a NanoDrop spectrophotometer at A280 absorbance. TiO2 beads (GL Sciences) were resuspended in binding buffer (2 M lactic acid in 50% ACN) at a concentration of 100 μg/μl and added to the peptides at a ratio of 4 mg of beads per 1 mg of peptide. Before the addition of the TiO2 beads, lactic acid and acetonitrile were added to the peptides for a final concentration of 2 M lactic acid and 50% ACN. Samples were vortexed and incubated for 30 min with end-over-end rotation. After incubation, samples were centrifuged for 1 min at 8000g, and the supernatant was added to fresh TiO2 beads for a second 30-min incubation. The beads from both incubations were combined and washed three times with 1 ml of binding buffer and then three times with 1 ml of 50% ACN. Peptides were eluted from the TiO2 beads with 600 μl of 5% ammonium hydroxide in 50% ACN by vortexing and passed through C-18 stage tips. One hundred microliters of 80% ACN was passed through the stage tip, and peptides were dried in a SpeedVac, and subsequent LC-MS/MS analysis was performed using the Q Exactive.
Raw data analysis of s-SILAC phosphoproteomics experiments was performed as described above, with phosphorylation of Ser, Thr, and Tyr being included as variable modifications, with a maximum number of modifications per peptide being 5. The MaxQuant normalized ratios “Phospho(STY).txt” were used for quantitation data analysis in Perseus software (1.6.2.3) (data file S7). s-SILAC phosphosite ratios were processed in the following manner: MaxQuant-normalized s-SILAC ratios were transformed 1/(x) to generate light/heavy ratios, followed by log2/(1/x) transformed. Phosphorylated serine, threonine, and tyrosine (pSTY) sites were filtered for only those that were confidently localized (class I, localization probability ≥0.75), followed by filtering for phosphosites identified in at least 70% of runs leaving 2906 phosphosites. Log2-transformed s-SILAC values were normalized to each column by subtracting its median. Imputation of missing values was performed the same as in MIB-MS analysis, where, in the s-SILAC data, a width of 0.3 and a downshift of 0.5 were used (data file S7). Quantitative differences in s-SILAC ratios among control siRNAs or MRCKA siRNA treatments were determined using a two-sample Student’s t test with the following parameters (S0 0.1 and Side, Both) using P value <0.05 (data file S7). KSEA (24) software (https://casecpb.shinyapps.io/ksea/) that uses preexisting phosphoproteomics databases to determine candidate active kinases was used to predict changes in kinase activity after MRCKA knockdown. A .csv file was generated from the two-sample Student’s t test comparing control siRNA versus MRCKA siRNA-treated cells Perseus export matrix (data file S7) that included Protein ID, Gene ID, Peptide, Residue.Both, P value, and fold change. Parameters used in KSEA were as follows: PhosphoSitePlus + NetworkKIN, Set NetworkKIN score cutoff = 2, [for plot] Set P value cutoff = 0.05, and [for plot] Set substrate count cutoff = 10. A volcano plot was then generated in RStudio 1.1.423 using ggplot2 and ggrepel libraries (Fig. 4C) using KSEA Kinase Scores (data file S7) to visualize kinases predicted to activated or inhibited. The phosphoproteomics MS data were deposited to the ProteomeXchange Consortium via PRIDE partner repository with the dataset identifier PXD015849.
Bioinformatics analyses of TCGA and CPTAC datasets
Analysis of MRCKA (CDC42BPA) CNA, protein, and mRNA alterations in HGSOC from TCGA and CPTAC studies was performed at the Biostatistics and Bioinformatics Facility, FCCC. The heat map presented in Fig. 3A was generated using R packages and shows the relationship between MRCKA (CDC42BPA) gene and protein expression with stage of tumors or subtype. MRCKA expression is not associated with stage of tumors (Fisher’s two-sided P = 0.9) or subtype (Fisher’s two-sided P = 0.17); however, a positive correlation was observed between mRNA and protein expression (Pearson’s correlation = 0.5, P < 0.001). Stage and subtype data were obtained from TCGA data repository. Overall survival was based on vital status and “days to death” from initial pathologic diagnosis. Individuals who were still alive at the time of the last follow-up were censored. Survival curves were compared with log-rank tests, and these calculations were done using the R “survival” package [T. M. Therneau and P. M. Grambsch Modeling Survival Data: Extending the Cox Model (Springer-Verlag, 2010)]. Survival data were obtained from TCGA data repository.
To assess the pathway profiles enriched in low and high MRCKA expressing tumors, we applied single-sample gene set variation analysis using RSEM gene expression measures for 307 HGSOC obtained from Broad Institute’s GDAC Firehose portal (https://gdac.broadinstitute.org/) and 12,915 gene sets obtained from MSigDB (48) as input. To identify enriched gene sets and pathways for low and high MRCKA expressing HGSOC cases, we split the 307 cases into high and low MRCKA expressors (Z score > 1.28 and Z score < 1.28), and Wilcoxon rank sum test was applied. Statistically significant pathways (C2 pathway sets of MSigDB) and gene sets (Benjamini-Hochberg FDR < 0.05%) enriched for low and high MRCKA cases (enrichment score > 0.3 or <−0.3) were selected. Resulting enrichment scores were plotted as heat map, and expression of MRCKA gene and protein was plotted as bars (Z score). Selected pathways were plotted as bar plots (P values were transformed into −log P values). Similarly, we also evaluated differentially expressed genes between low and high MRCKA expressors using DESeq2 algorithm with absolute gene counts for the same data as input. Selected differentially expressed genes (FDR < 0.05) were used as input to gene set enrichment analysis (49).
SUPPLEMENTARY MATERIALS
stke.sciencemag.org/cgi/content/full/13/619/eaax8238/DC1
Fig. S1. Quantitation of kinase abundance in HGSOC tumors using MIB-MS and s-SILAC.
Fig. S2. Characterization of MRCKA in HGSOC tumors and HGSOC cell lines.
Fig. S3. Exploring MRCKA knockdown in HGSOC cells using MIB-MS and phosphoproteomics.
Fig. S4. MRCKA promotes focal adhesion signaling in HGSOC cells.
Fig. S5. Evaluation of MRCKA kinase inhibitors in HGSOC cells.
Fig. S6. Characterization of proliferation, migration, apoptosis, and spheroid formation in BDP9066-treated HGSOC cells.
Data file S1. Raw and processed s-SILAC ratios from MIB-MS profiling of sections from a single HGSOC tumor tissue.
Data file S2. Characteristics of HGSOC primary and PDX tumors used in MIB-MS studies.
Data file S3. Raw and processed s-SILAC ratios from MIB-MS profiling of HGSOC primary and PDX tumors.
Data file S4. Prevalent MIB-MS kinome signature among HGSOC primary and PDX tumors.
Data file S5. IHC analysis of MRCKA protein levels in HGSOC tumor sections, normal fallopian tube, and ovarian surface epithelial tissues.
Data file S6. Raw and processed s-SILAC ratios from MIB-MS profiling of MRCKA knockdown in HGSOC cells.
Data file S7. Raw and processed s-SILAC ratios from phosphoproteomics analysis of MRCKA knockdown in HGSOC cells.
Data file S8. In vitro kinase inhibitor profiling report for BDP5290 and BDP9066.
Data file S9. Raw and processed SILAC ratios from MIB-MS profiling of OVSAHO cells treated with BDP9066 or A-674563.
Data file S10. Reagents used in the study, including small molecules, siRNAs, and antibodies.
This is an article distributed under the terms of the Science Journals Default License.
