VAV2 drives EGFR-mediated Rac1 responses in prostate cancer

Abstract

The small G-protein Rac1 is a central player in cancer progression and metastatic dissemination. Rac1 has been established as a bona fide effector of receptor tyrosine kinases, acting as a signaling node for motility, invasiveness, mitogenesis, and gene expression. Previous studies demonstrated that Rac1 is hyperactivated in aggressive cellular models of prostate cancer. Here, we demonstrate that CRISPR/Cas9-Mediated knockout of Rac1 results in impaired proliferation and migration of prostate cancer cells. Rac1-null cells display profound alterations in transcriptional programs, particularly those associated with cell adhesion and extracellular matrix (ECM) regulation. Combined expression profiling and unbiased RNAi screening of Rac1 Guanine nucleotide Exchange Factors (Rac-GEFs) identified VAV2 as the foremost mediator of epidermal growth factor (EGF)-induced GTP loading onto Rac1 in prostate cancer cells. Depletion of VAV2 from prostate cancer cells significantly reduced their proliferative and migratory capacities without affecting the expression of Rac1-regulated genes, suggesting that VAV2 controls a discrete subset of Rac1-dependent cellular responses. Immunohistochemical assessment in human prostate biopsies showed significant VAV2 overexpression in tumor areas. Bioinformatic analysis revealed a strong correlation between VAV2 expression and poor clinical prognosis. In addition to uncovering a prominent role for VAV2-Rac1 as an effector pathway mediating EGFR-driven proliferative and migratory responses in prostate cancer cells, our findings underscore the potential prognostic value of VAV2 in human prostate cancer progression.

Introduction

Prostate cancer is the most common cancer and the second leading cause of cancer-related deaths among males in the US. While locally advanced or regional spread disease displays a five-year survival greater than 99%, the survival rate for advanced-stage disease with distal metastatic spread declines to approximately 34% (1). Androgen deprivation therapy by either pharmacological or surgical castration has been the standard of care for locally advanced or metastatic disease. Despite the initial response to androgen blockade, all patients eventually progress (either biochemical or clinical progression) to castration resistance in the setting of circulating testosterone castrate levels. Castration-resistant prostate cancer (CRPC) is associated with the acquisition of phenotypic pro-invasive cellular traits, leading to tumor cell growth and locoregional or distal spread (2–5).

The Rho GTPases are a family of small G-proteins that belong to the Ras superfamily. In response to extracellular cues, Rho GTPases promote major effects of actin cytoskeleton remodeling and cell polarity. Most Rho small G-proteins function as molecular binary switches, cycling between inactive GDP-bound and active GTP-bound states. Their activation is mediated by guanine nucleotide exchange factors (GEFs) that promote GTP loading, whereas inactivation is catalyzed by GTPase-activating proteins (GAPs) that accelerate GTP hydrolysis (6, 7). Rac1 has gained significant interest among the Rho small G-proteins due to its prominent role in cancer progression and metastasis. In its active, GTP-bound form, Rac1 constitutes an essential node in the signaling pathway, leading to migratory and invasive responses in cancer cells. In addition, Rac1 has been linked to cell proliferation, endocytosis, and the control of gene expression, among other cellular responses (8–11). The functional outcome of Rac1 activation is essentially cell context-dependent and is determined by the subcellular formation of distinctive Rac-GEF/Rac1 complexes (12–14). To date, more than 80 Rho GEF proteins have been reported in humans, with at least 43 members exhibiting exchange activity towards Rac1. The Rac-GEFs can be divided into two subfamilies. The Dbl-like class comprises 32 GEFs, characterized by a characteristic Dbl homology (DH) domain responsible for exchange activity and regulatory PH domain(s) primarily involved in intracellular targeting and autoinhibition. The second class of Rac-GEFs comprises DOCK GEFs (DOCK1 to 11), which contain a DOCK homology region (DHR2) responsible for catalytic activity and a targeting DHR1 domain. DOCK1-5 function as heterodimers with ELMO adaptor proteins (ELMO1-3) and display prominent specificity for Rac GTPases (11, 15, 16). The distinctive regulatory properties, intracellular location, and cell-type-specific expression patterns of Rac-GEFs highlight the extraordinary complexity of Rac1-driven signaling pathways and their mediated functional responses in cancer models.

It has been widely reported that Rac1 expression and/or signaling activity are aberrantly elevated in multiple cancer types. Such underlying alterations have been causally linked to cancer development and progression (8–11). While Rac1 mutations in cancer are rare, with the notable exception of the P29S somatic mutation in cutaneous melanoma (17), Rac1 hyperactivation in cancer is instead associated with elevated Rac-GEF activity, either due to abnormal expression or activating mutations or involves an excessive input from upstream membrane receptors or their effectors, namely PI3K (9, 11, 18–20). A well-known example of aberrant Rac-GEF expression is the up-regulation of P-REX1 in luminal breast cancer, which has been causally linked to elevated cell motility and metastasis (21, 22). Another example includes the overexpression of FARP1 and ARHGEF39 in non-small cell lung cancer (NSCLC). In all these cases, stimulation of receptor tyrosine kinases (RTKs), such as the epidermal growth factor receptor (EGFR), triggers the relocalization of the Rac-GEFs to actin-rich peripheral structures (e.g., ruffles), thereby activating Rac1 and conferring high migratory and pro-metastatic traits to cancer cells (23–25). We have previously established that aggressive androgen-independent prostate cancer cells display prominent Rac1 hyperactivation (26, 27). Still, the functional relevance of the Rac1 activation status in prostate cancer cells remains to be defined. Moreover, the Rac-GEFs contributing to Rac1 activation in response to RTK stimulation in prostate cancer models have yet to be established.

Using a Rac1 knockout cellular model, we established the critical roles of this small GTPase in cell motility, proliferation, and gene expression in prostate cancer cells. An unbiased screening identified VAV2 as the most prominent Dbl-like GEF involved in Rac1 activation (i.e., GTP loading) upon EGFR stimulation. Interestingly, VAV2 mediates only a subset of Rac1-driven responses in prostate cancer cells. In addition, immunohistochemical analysis of human specimens led us to establish a significant up-regulation of VAV2 in human primary prostate tumors. The observed association between VAV2 high expression and poor clinical outcome highlights its potential prognostic value in human prostate cancer.

Methods

Cell lines

Human prostate cancer cell lines LNCaP (RRID: CVCL_0395), LNCaP-C4 (RRID: CVCL_4783), LNCaP-C4-2 (RRID: CVCL_4782), DU145 (RRID: CVCL_0105), and PC3 (RRID: CVCL_0035) were purchased from ATCC (Manassas, VA). PC3-ML (RRID: CVCL_6E90) cells were provided by Dr. Alessandro Fatatis of Drexel University, as described previously (28). Cell lines were authenticated by IDEXX Bioanalytics (Columbia, MO) and used in passages 3–8. DAPI staining for Mycoplasma testing was performed twice a year.

RNAi depletion

Cells were transfected with 25 nM siRNA duplexes (Horizon Discovery, Cambridge, UK) using Lipofectamine RNAiMAX (13778150, Thermo Fisher Scientific) (29). The following siRNA duplexes from Horizon Dharmacon were used: ALS (Cat # J-014168-09), ARHGEF39 (Cat # J-015006-05), DOCK1 (Cat # J-011253-05), DOCK5 (Cat # J-018931-05), DOCK7 (Cat # J-031725-05), DOCK9 (Cat # J-014040-09), ECT2 (Cat # J-006450-05), ELMO2 (Cat # J-019222-05), FARP1 (Cat # J-008519-06), FARP2 (Cat # J-009237-06), PLECKHG2 (Cat # J-023690-05), PREX1 (Cat # J-010063-09), SPATA13 (Cat # J-015469-09), TRIO (Cat # J-005047-05), VAV2 (Cat # J-005199-05, J-005199-06 and J-005199-07) and Rac1 (Cat # J-003560-14-0005 and J-003560-15-0005). As a non-target control (NTC), we used an NTC Pool (Cat # D-001810-10-05).

Generation of Rac1 and VAV2 knockout cell lines

DU145 Rac1 knockout (KO) clones were generated using the gene KO kit #1 from Synthego Synthego (RRID: SCR_026304, Menlo Park, CA), following the manufacturer’s protocol. Briefly, a sgRNA targeting Rac1 exon 3 was complexed with recombinant Cas9 protein and transfected into cells using Lipofectamine CRISPRMAX (CMAX00001, Thermo Fisher Scientific. Clone #1 was generated with sgRNA #1 (CUGUUUGCGGAUAGGAU), and clones #2 and #3 were generated with sgRNA #2 (CCUUACUGUUUGCGGAU). Scramble clones were generated with a non-targeting sgRNA (GUACGUCGGUAUAACU). Two days post-transfection, cells were seeded as single cells for clonal expansion, and KO was subsequently confirmed by Western blotting and Sanger sequencing. A similar approach was used to generate Rac1 PC3 KO cells. The generation of DU145 and LNCaP C4-2 VAV2 KO clones was performed following the same protocol, using the Synthego gene KO kit #2, which utilizes a pool of three sgRNAs (CGAUGGAGCCGGGGGAG, CGCGCAGCGCCUGCGCC and CGGGCGCCAUGGAGCAG), each targeting different sites. All KO clones were confirmed by sequencing analysis and/or Western blot.

Active Rac1 and Cdc42 pull-down assays

Pull-down assays and detection of active small G-proteins were performed as described previously (30, 31). Briefly, cells were serum-starved for 16 hours and then stimulated with EGF (100 ng/mL, 1 minute). After lysis and pull-down with GST-fused PAK binding domain (PBD), active Rac1 and Cdc42 were detected by Western blotting using specific antibodies. For RNAi depletion experiments, pull-downs were performed 48 hours after transfection.

Proliferation assays

Cell proliferation was assessed in triplicate in 96-well plates using an automated Cytation^™ 5 (Cytation 5 Cell Imaging Multi-Mode Reader (RRID: SCR_019732, BioTek^® Inc.) imaging system. Images were captured every 24 h, and cell counts were determined. Alternatively, cell counting was performed using a hemocytometer.

Morphology assessment

Cells were dissociated using a non-enzymatic cell dissociation solution (Cat # C5914, Merck) and seeded onto glass coverslips in complete media to assess morphology. After 24 h, cells were fixed in 3.6% formaldehyde, permeabilized in 0.1% Triton-X100/PBS, and blocked in 10% goat serum/PBS before staining with both Alexa Fluor 488-conjugated phalloidin and Hoechst 33258 to label filamentous actin and DNA, respectively. Wide-field images were acquired using a Nikon Eclipse TE2000 inverted microscope system (RRID: SCR_023161). Cell and nuclear areas were quantified using ImageJ (RRID: SCR_003070).

Migration assays

Cell migration assays were performed using Transwell inserts with 8 μm pore polyester membranes (uncoated, #3464, Corning Incorporated, MA, USA). After 24 hours of serum starvation, 5 × 10^4 cells were seeded, and the lower chamber was supplemented with either 10% FBS or 200 ng/ml EGF. Cells were incubated at 37°C and 5% CO2 for 16 to 24 hours. Migrated cells were fixed with 10% formaldehyde for 15 min and then stained with 0.2% crystal violet (Cat # 23-750-025, Thermo Fisher Scientific, MA, USA) for 15 min at room temperature. Cells on the top surface of inserts were removed by wiping. Micrographs of migrated cells were obtained from five random fields by phase contrast microscopy (Nikon Eclipse TE2000 inverted microscope) Nikon Eclipse TE2000 inverted microscope system (RRID: SCR_023161), and cells were counted using CellProfiler Image Analysis Software (RRID: SCR_007358). Experiments were performed in triplicate.

Western blot

Western blot was performed as described previously (32). Primary antibodies were as follows: anti-Rac1 clone 23A8 (Millipore Cat# 05-389, RRID: AB_309712); anti-Cdc42 clone 11A11 (Cell Signaling Technology Cat# 2466, RRID: AB_2078082), anti-vimentin (Cell Signaling Technology Cat# 3390, RRID: AB_2216128), anti-phospho-ERK (Cell Signaling Technology Cat# 9101, RRID: AB_331646), anti-VAV2 (Cell Signaling Technology Cat# 2848, RRID: AB_2213746), anti-P-cadherin (Cat. #1429S), anti-caspase-3 (Cat. 9662S), anti-Zeb1 (Cell Signaling Technology Cat# 3396, RRID: AB_1904164), anti-Twist2 (Cat. #69366), anti-SNAIL (Cat. #3879) (Cell Signaling Technology Cat# 3879, RRID: AB_2255011), anti-SLUG (Cell Signaling Technology Cat# 9585, RRID: AB_2239535), anti-PKCα (Cell Signaling Technology Cat# 2056, RRID: AB_2284227), anti-AXL (Cell Signaling Technology Cat# 8661, RRID: AB_11217435), and anti-E-cadherin (R&D Systems Cat. #8505-EC-050). As loading controls, we used anti-actin (Sigma-Aldrich Cat# A5441, RRID: AB_476744) and anti-vinculin (Sigma-Aldrich Cat# V9131, RRID: AB_477629). As secondary antibodies, we used anti-mouse and anti-rabbit HRP antibodies (Bio-Rad Cat# 172-1017, RRID: AB_11125758, and Bio-Rad Cat# 170-6515, RRID: AB_11125142, respectively). Bands were visualized using enhanced chemiluminescence with the LI-COR Odyssey Fc dual-mode imaging system.

Rac-GEF expression and Q-PCR assays

To quantify the expression of Rac-GEFs, we used a custom-made Q-PCR array (ThermoFisher Scientific) encompassing 32 Dbl-like and 11 DOCK family Rac-GEFs (23, 33). The array included UBC and B2M housekeeping genes for normalization. Briefly, RNA was extracted from cultured cells using the RNeasy kit, as directed by the manufacturer (QIAGEN, Valencia, CA). cDNA was obtained with the Taq-Man Reverse Transcription Kit (QIAGEN). Results were expressed as ΔCt, calculated as the difference in Ct values between the Rac-GEF of interest and the average of housekeeping genes.

For single-gene Q-PCR amplification, we used an ABI PRISM 7300 Detection System in a total volume of 10 μl containing TaqMan^™ Fast Advance Master Mix (ThermoFisher Scientific), target primers (300 nM), fluorescent probe (200 nM), and cDNA. UBC was used as a housekeeping gene.

RNA-Seq and transcriptomic analysis

Parental DU145 cells, three Rac1 CRISPR KO DU145 clones, and two CRIPSR scrambled control clones were processed for RNA-Seq. Briefly, RNA was extracted from the cells using TRIzol, followed by a chloroform treatment, and then purified using the Zymo Research RNA Clean and Concentrator kit. Library preparation was performed using SMARTScribe reverse transcriptase (Takara bio) with 160 ng of RNA following the manufacturer’s protocol. Library amplification was performed using Advantage 2 DNA polymerase (Takara Bio). After cDNA fragmentation, the library underwent tagmentation and was purified with SPRI beads (Beckman Coulter). The final library concentration was quantified by Q-PCR.

An Illumina NextSeq 550 System (RRID: SCR_016381) was used for sequencing the library with 2 × 80 bp paired-end reads. The sequence output was aligned using STAR (RRID: SCR_004463). Differential expression was assessed using DESeq2 (RRID: SCR_015687), comparing KO samples to WT and scramble CRISPR control clone samples. Genes were considered to be differentially expressed with a log2(fold-change) ≥ 0.5 or ≤ −0.5, and an adjusted p-value of < 0.05 was used for statistical significance.

In silico analysis of Rac-GEF expression profiles in human prostate cancer

Pre-processed Rac-GEF RNA-seq expression levels among normal and primary prostate carcinomas and follow-up data were retrieved from The Cancer Genome Atlas Prostate Adenocarcinoma (TCGA-PRAD) dataset through the UCSC Xena (RRID: SCR_018938 (http://xena.ucsc.edu/). In addition, pre-processed VAV2 mRNA expression profiles among four prostate cancer microarray-based datasets, GSE3325 (34), GSE6099 (35), GSE21034 (36), and GSE35988 (37), were obtained from GEO and analyzed using R software. Pearson correlation analysis of Rac-GEFs mRNA expression levels with Rac1 cell motility activity predicted by PARADIGM among primary carcinomas (n=488) was performed with data retrieved from UCSC Xena browser (cohort: TCGA Prostate Cancer – PRAD; dataset: pathway activity - z score of 1387 constituent PARADIGM pathways). Primary prostate carcinomas were divided into two groups based on their Rac-GEF expression levels, categorized as low or high, according to the median value. The Wilcoxon rank-sum test was used to compare these two groups for Rac1 cell motility signaling pathway activity.

Tumor microarray staining and analysis

Immunohistochemistry was performed essentially as described in (38). We utilized prostate cancer tissue microarrays (TMAs) constructed at the Department of Pathology and Laboratory Medicine, University of Pennsylvania, for immunohistochemical analysis of human specimens. All cases were anonymized, and no association was found with patient information. For H-score calculation, the intensity of staining was graded on a scale of 0 to 3, with 0 indicating no staining and 3 indicating the most potent staining. The percentage of tumor cell staining was calculated over 10 high-power fields in the region over the entire tumor. The staining was normalized to compare different tumors by multiplying the intensity of staining with the percentage of tumor cells staining. The data were analyzed in a blinded manner concerning the core tissue samples. The VAV2 antibody was purchased from Abcam (Cat. # ab52640).

Statistical analysis

Statistical significance was determined by Student’s t-test or ANOVA using GraphPad Prism version 10.2.3 GraphPad Prism (RRID: SCR_002798). For survival analysis using TCGA-PRAD, statistical analysis was performed using the R packages survival (RRID: SCR_021137) and survminer (RRID: SCR_021094).

Data availability

The data generated in this study are available within the article and its supplementary data files. Gene expression profile data in this study were obtained from Gene Expression Omnibus (GEO) GSE3325, GSE6099, GSE21034, GSE35988, and TCGA-PRAD. The gene expression data generated in this study are publicly available in GEO at GSE292590.

Results

Rac1 regulates the morphology and migration capacity of DU145 prostate cancer cells.

We have previously reported robust constitutive Rac1 activation in androgen-independent prostate cancer cell lines (26, 27). Rac1 expression is not altered in prostate tumors compared to normal (Supplementary Fig. S1A). However, the PARADIGM algorithm, which integrates Rac1 expression with pathway analysis in TCGA-PRAD, revealed that the “High Rac1 cell motility activity” pathway predicts a poor progression-free interval (PFI) (Supplementary Fig. S1B). This suggests that Rac1 activity rather than expression is associated with disease progression.

To address the functional relevance of this pathway, we generated Rac1 KO DU145 prostate cancer cell lines using a gene-editing-based CRISPR/Cas9 system. Rac1 KO clones were generated using two different sgRNA sequences (sgRNAs #1 and #2), and 3 clones were selected: Rac1 KO #1 (from sgRNA #1) and Rac1 KO #2 and #3 (from sgRNA #2; see Methods). A scrambled sgRNA (Scr) was used as a control. Rac1 depletion in all KO clones was confirmed by Western blot (Fig. 1A). Despite the lack of gross changes in cellular shape, Rac1 KO clones display higher total cell volume with a preserved nuclear-to-cytoplasmic ratio (Supplementary Fig. S2). Loss of Rac1 caused a significant reduction in cell proliferation. Indeed, Rac1 KO cells display ~50% reduction in cell number relative to control cells at 72 h after seeding (Fig. 1B). As expected from the well-established role of Rac1 in cell motility and consistent with previous studies in other cellular models (25), Rac1 KO DU145 cells have reduced migratory capacity (Fig. 1C). The impaired motility was confirmed in DU145 subjected to transient Rac1 depletion using RNAi (Supplementary Fig. S3A) and in PC3 KO cells (Supplementary Fig. S3B). To further confirm the role of Rac1 in prostate cancer cell migration, we examined the effect of three different Rac inhibitors: MBQ-167, MBQ-168, and EHop-097. MBQ-167 and MBQ-168 act primarily by preventing GTP binding to the small GTPase. EHop-097 has been identified as an inhibitor that interferes with the binding of Rac1 to VAV GEFs (39–42). Analysis of DU145 cell migration using a Transwell assay showed a concentration-dependent inhibition by all three compounds. Interestingly, the VAV/Rac inhibitor EHop-097 was approximately 100 times more potent than MBQ-167 and MBQ-168. Indeed, IC₅₀’s for MBQ-167 and MBQ-168 were ~ 10 μM, whereas IC₅₀ for EHop-097 was ~ 0.1 μM (Fig. 1D).

Figure 1.

Rac1 KO DU145 cells have impaired migratory and proliferative capacities. A, Representative Western blot showing Rac1 depletion in Rac1 knockout (KO) DU145 cells. Scr scrambled sgRNA. B, Cell number was determined at 24 h, 48 h, and 72 h after seeding. Results are expressed as fold-change relative to time=0 (mean ± SEM, n=4). C, Boyden chamber migration assay. Left panel, representative membrane image. Middle panel, representative micrographs. Right panel, quantification of migratory cells (mean ± SD, n=12 fields). A second experiment gave similar results. D, Migration of parental DU145 cells was determined in the presence of Rac inhibitors MBQ-167, MBQ-168, or EHop-097 (0.1–10 μM). Left panel, representative images. Right panel, quantification of migratory cells. Results were expressed as migration percentage relative to vehicle-treated cells (mean ± SEM, n=3). ns, not significant, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Rac1 controls gene expression in prostate cancer cells.

Rac1 regulators and effectors are known to control gene expression in cancer cells by influencing transcriptional programs associated with tumorigenesis and metastasis (43–49). To elucidate the global regulation of gene expression by Rac1 in prostate cancer cells, we conducted a transcriptomic analysis using RNA sequencing (RNA-Seq). For this experiment, we used three Rac1 KO DU145 cell lines (#1, #2, and #3) and three DU145 control cell lines (one parental and two scrambled sgRNA cell lines). Principal component analysis revealed a clear difference between control and Rac1 KO cells (Fig. 2A). Using a cutoff of Log2(FC) ≥ .5 or ≤ −.5 relative to parental cells, we identified 358 differentially regulated genes in Rac1 KO cells vs. control cells (269 up-regulated and 89 down-regulated), as portrayed in the heatmap in Fig. 2B and the volcano plot in Fig. 2C. A complete list of Rac1-regulated genes is presented in Supplementary Table S1. Validation by Q-PCR for 4 up-regulated genes (MPZL2, SAT1, SLPI, and SPINT1) recapitulated the results of the transcriptome analysis (Supplementary Fig. S4). KEGG pathway gene set enrichment analysis showed pathways associated with cell adhesion, ECM functions, migration, proliferation, inflammation, and immune response, which are established processes regulated by Rac1 (Fig. 2D and Supplementary Table S2). The highest KEGG enrichment in Rac1 KO cells was for the “Cell adhesion molecules” gene set. A combined analysis using KEGG, Reactome, and Gene Ontology (GO) revealed the enrichment of pathways associated with similar cellular processes (Supplementary Table S2). These results suggest that Rac1 plays a crucial role in regulating gene expression and controlling fundamental cellular functions in prostate cancer cells.

Figure 2.

Rac1 regulates gene expression in prostate cancer cells. RNA-Seq analysis of three DU145 Rac1 KO clones compared to parental (P) and two scrambled sgRNA (Scr) control clones was carried out. A PCA plot for parental scrambled and Rac1 KO DU145 cells. B, Heatmap of differentially expressed genes with a log2(FC) ≥ 0.5 or ≤ −0.5 and adjusted p-value ≤ 0.05. C, Volcano plot of Rac1 KO differential gene expression. Blue dots, genes of the KEGG cell adhesion molecules or KEGG ECM receptor interaction gene set lists (see Fig. 3). D, Plot showing enriched KEGG gene sets in Rac1 KO cells.

To further delineate Rac1-regulated genes associated with cell adhesion, a heatmap of genes from the “Cell adhesion molecules” and “ECM receptor interactions” gene sets that were differentially expressed in Rac1 KO cells was generated. This includes sets of genes for integrins (ITGA2, ITGB3, ITGB4), cadherins (CDH1, CDH3) and fibronectin (FN1). As shown in Fig. 3A, there was a prominent increase in the expression of genes associated with these processes. A clear up-regulation in CDH1 and CDH3 mRNA expression was detected by Q-PCR in Rac1 KO cells, thus confirming results observed in the RNA-Seq (Fig. 3B and 3C). Furthermore, Western blot analysis revealed significant up-regulation (3–6-fold) of CDH1 and CDH3 gene products (E-cadherin and P-cadherin, respectively) for all Rac1 KO DU145 clones (Fig. 3D). Despite the elevated levels of the epithelial marker CDH1, Rac1 KO cells still retain the expression of markers associated with a mesenchymal state, namely vimentin, SNAIL, SLUG, ZEB2, TWIST2, AXL and PKCα (Fig. 3E). Thus, despite the negative regulation on E-cadherin expression by Rac1, inhibition of Rac1 expression was unable to reverse the characteristic mesenchymal phenotype of DU145 cells (50, 51).

Figure 3.

Rac1 regulates the expression of E-cadherin and P-cadherin in prostate cancer cells. A, Heatmap of genes in the KEGG “cell adhesion molecules” or KEGG “ECM receptor interaction” gene lists. P, parental; Scr, scrambled sgRNA. B, Expression of CDH1 and CDH3 according to the RNA-Seq analysis. C, CDH1, and CDH3 up-regulation in Rac1 KO DU145 cells, as determined by Q-PCR. Results are expressed as fold-change relative to Scr. D, E-cadherin and P-cadherin up-regulation in Rac1 KO DU145 cells. Upper panel, representative Western blot. Lower panel, densitometric analysis of 4 independent experiments (mean ± SEM). E, Expression of EMT markers in DU145 cells. A representative experiment is shown. Similar results were observed in two additional experiments. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001 vs. Scr.

Determination of Rac-GEF expression in prostate cancer cells

The mechanisms leading to Rac1 activation in prostate cancer are poorly understood. There is very limited knowledge about the expression of Rac-GEFs and their functional relevance in prostate cancer models. Owing to the large size of the Rac-GEF family, we chose to use an unbiased approach to determine their expression in prostate cancer cell lines. mRNA was purified from androgen-dependent cell lines LNCaP and LNCaP-C4, its androgen-independent derivative LNCaP-C4–2, as well as from androgen-independent DU145, PC3 and PC3-ML cell lines. Rac-GEF expression was subsequently determined with a customized Q-PCR array that comprises 43 known Dbl-like and DOCK Rac-GEFs and the 3 ELMO adaptors (23, 33). Results are depicted in the heatmap in Fig. 4A and Supplementary Table S3 revealed a comparable expression pattern among the different prostate cancer cell lines. The top-expressed Rac-GEFs were ECT2, VAV2, TRIO, PLEKHG2, FARP1, FARP2 and PREX1. Prostate cancer cells express the ELMO2 adaptor protein, whereas ELMO1 and ELMO3 expression was low.

Figure 4.

Identification of VAV2 as a mediator of EGFR-induced Rac1 activation in prostate cancer cells. A, Heatmap for the expression of Rac-GEFs and ELMO adaptors in prostate cancer cell lines, as determined with a Q-PCR array. Expression is shown as ΔCt for each Rac-GEF relative to the average of UBC and B2M housekeeping genes. B, Correlation analysis of mRNA expression levels with Rac1 signaling pathway and Rac1 cell motility signaling pathway activities for 43 Rac-GEFs as predicted by PARADIGM among TCGA primary prostate carcinomas (n=488). Gene expression levels and pathway activities were retrieved from the UCSC Xena browser. Statistically significant positively correlated Rac-GEFs (p<0.05) were divided into low and high median Rac-GEF expression levels and compared to RAC1 cell motility signaling pathway activity using the Wilcoxon rank-sum test. C, C4–2 prostate cancer cells were transfected with the indicated Rac-GEF siRNA duplexes or a non-target control RNAi (NTC). After 48 h, cells were serum starved and treated with EGF (100 ng/μl, 1 min). Rac1-GTP levels were determined with a PBD pull-down assay. Upper panel, representative Western blots. Lower panel, densitometric analysis. Results, normalized to total Rac1 levels in each case, were expressed as percentage relative to parental cells, + EGF (dotted line) (mean ± SEM, n=3–11). ** p<0.01. D, Similar experiments as in C using DU145 cells (mean ± SEM, n=3–10). ** p<0.01.

Rac-GEF mRNA expression levels were correlated to Rac1 constituent integrated pathway motility activities predicted by PARADIGM, specifically for prostate carcinomas derived from the TCGA-PRAD project (n=488). The PARADIGM algorithm integrates pathway, expression, and copy number data to infer activation of pathway features within a superimposed pathway network structure extracted from NCI-PID, BioCarta, and Reactome. Fig. 4B shows the correlation coefficients between mRNA expression for each Rac-GEF and the Rac1 cell motility pathway activity. These results identified several Rac-GEFs to correlate with Rac1 cell motility activity in prostate adenocarcinomas positively. The top correlation was observed for VAV2, followed by ECT2, KALRN, FARP2, ARHGEF16, DOCK4 and DOCK7.

Unbiased identification of VAV2 as the main Rac1 activating protein downstream of EGFR in prostate cancer cells

EGF is a well-established stimulus for Rac1 activation (21, 23, 30, 52, 53) and has demonstrated roles in prostate cancer progression (54–57). Using two different cellular models (C4–2 and DU145 cells), we observed that EGF triggers Rac1 activation (GTP loading), as determined by a Pak-binding domain (PBD) pull-down approach (Fig. 4C and Fig. 4D). The top 14 expressed Rac-GEFs, as ranked by the Rac-GEF array, were individually silenced using specific siRNA duplexes both in C4–2 and DU145 cells. This includes 10 Dbl-like Rac-GEFs and 5 DOCK GEFs. The knockdown efficiency for the corresponding targeted genes was 70–90% in most cases, as determined by Q-PCR (Supplementary Fig. S5). ERK phosphorylation was used as a control for EGFR stimulation. Interestingly, this unbiased screening revealed VAV2 as the only GEF implicated in EGFR-driven Rac1 activation in C4–2 and DU145 cells. On the other hand, silencing other Dbl-like and DOCK Rac-GEFs, even those displaying high expression, did not cause any statistically significant changes in Rac1 activation by EGF in either cell line. Consistent with the lack of involvement of DOCK1–5 GEFs, depletion of the ELMO2 adaptor does not affect EGF-induced Rac1 activation (Fig. 4B and Fig. 4C).

To confirm the relevance of the VAV2 requirement for Rac1 activation by EGF, as well as to minimize the chances of misinterpretation of results due to the off-target effect of RNAi, we used three different VAV2 siRNA duplexes, which in all cases caused significant VAV2 depletion, as determined by Western blot. As shown in Fig. 5A and 5B, Rac1 activation by EGF was impaired upon VAV2 depletion with all three siRNA duplexes in C4–2 and DU145 cells. The involvement of VAV2 in Rac1 activation by EGF was confirmed in VAV2 KO C4–2 and DU145 cells generated using CRISPR. Consistent with the additional role of VAV2 as a Cdc42 GEF, VAV2 KO cells also displayed reduced Cdc42 activation in response to EGF (Supplementary Fig. S6).

Figure 5.

VAV2 is required for the migration and proliferation of prostate cancer cells. A, C4–2 cells were transfected with the indicated VAV2 or non-target control (NTC) siRNA duplexes. After 48 h, cells were serum starved and treated with EGF (100 ng/μl, 1 min). Rac1-GTP levels were determined with a PBD pull-down assay. Left panel, representative Western blot. Right panel, densitometric analysis. Results, normalized to total Rac1 levels in each case, were expressed as fold-change relative to parental cells + EGF (mean ± SEM, n=3–4). B, Similar experiments as in (A) using DU145 cells (mean ± SEM, n=3–4). C, Boyden chamber migration assay, 48 h after transfection with the indicated VAV2 siRNA duplexes or an NTC RNAi. EGF (100 ng/μl) was added to the lower chamber for the duration of the assay (16 h). Left panel, representative micrographs. Right panel, quantification. Results were expressed as fold-change relative to parental cells, - EGF (mean ± SEM, n=4). D, Similar experiments as in (C) using DU145 cells. E, DU145 cells were transfected with the indicated VAV2 siRNA duplexes or NTC RNAi. Cell number was determined for 24–96 h after seeding. Results are expressed as fold-change relative to time=0 (mean ± SEM, n=3). F, DU145 cells were transfected with the indicated VAV2 siRNA duplexes or NTC RNAi. Viability was determined 72 h after cell seeding using an MTT assay. Results are expressed as fold-change relative to parental (dotted line) (mean ± SEM, n=3). G, Representative Western blot for caspase-3 in control and VAV2-depleted DU145 cells. H, VAV2, CDH1, and CDH3 mRNA levels were determined by Q-PCR 48 h after transfection with VAV2 siRNA duplexes or NTC RNAi. Results are expressed as fold-change relative to parental cells (dotted line) (mean ± SEM, n=3). I, Expression of E-cadherin and P-cadherin in DU145 cells, 48 h after transfection with VAV2 siRNA duplexes or NTC RNAi. Left panel, representative Western blot. Right panel, densitometric analysis of 3 independent experiments. Results are expressed as fold-change relative to parental cells (dotted line) (mean ± SEM). ns, not significant, ** p<0.01, *** p<0.001, **** p<0.0001.

To further characterize the role of VAV2 in Rac1-mediated responses, we examined the migratory behavior of prostate cancer cells in response to EGF as a chemoattractant. Notably, silencing VAV2 led to a significant defect in the migratory capacity (as determined with a Boyden chamber) of C4–2 cells (Fig. 5C) and DU145 cells (Fig. 5D). Interestingly, VAV2-deficient DU145 cells also display impaired proliferation capacity, as determined by cell counting (Fig. 5E) and MTT assay (Fig. 5F). No caspase-3 cleavage could be observed in VAV2-depleted cells, suggesting lack of apoptosis in this case (Fig. 5G). Unlike the effect observed in Rac1 KO cells, VAV2 depletion could not cause any measurable up-regulation in E-cadherin and P-cadherin at mRNA and protein levels (Fig. 5H and 5I). CDH1, CDH3, MPZL2, SPINT1, and SAT1 did not change their expression in VAV2-depleted cells and were not regulated by EGF (Supplementary Fig. S7). Therefore, while VAV2 mediates migratory and proliferative responses in prostate cancer cells, it does not control the expression of genes regulated by Rac1, thus suggesting that it is only required for a subset of Rac1-dependent responses.

Immunohistochemical analysis of VAV2 in human prostate tumors

While VAV2 expression is readily detected in cultured human prostate cancer cell lines, there is no information available about the expression of this GEF in human prostate cancer specimens by immunohistochemistry. To address this issue, we stained a prostate cancer tissue microarray (TMA) that comprises 126 cores with a validated anti-VAV2 antibody. We found prominent VAV2 staining in tumor areas, whereas staining in stromal tissue was negative (Fig. 6A and 6B). Determination of average H-scores for each core revealed a statistically significant increase in VAV2 staining in tumor vs. non-tumoral glands (p<0.0001). Remarkably, ~90% of tumor areas display H-scores of 200–300 (Fig. 6C).

Figure 6.

VAV2 expression is elevated in human prostate cancer. A, Immunohistochemical staining of VAV2 in prostate cancer specimens. Representative cores from the TMA are shown. Rectangles, enlarged areas in B. Bar, 100 μm. B, Representative VAV2 staining in normal and tumor areas. Enlarged micrographs from selected areas in (A). Bar, 50 μm. C, Left panel, H-score for normal and tumor areas. Right panel, percentage of samples according to H-scores. **** p<0.0001.

Lastly, we conducted a bioinformatics analysis for Rac-GEFs expressed in prostate cancer using TCGA-PRAD. This dataset revealed that 5 of the 32 Dbl-family Rac-GEFs (VAV2, ECT2, ARHGEF16, FARP1, and FARP2) and 2 of the 11 DOCK Rac-GEFs (DOCK4 and DOCK6) display statistically significant up-regulation in tumors vs. normal. VAV2 was the top-up-regulated Rac-GEF (Fig. 7A, a complete list of Rac-GEFs in Supplementary Table 4). VAV2 was found to be a negative predictor for disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI) among the up-regulated Rac-GEFs (Fig. 7A; Kaplan-Meier analysis shown in Fig. 7B). Fig. 7C shows the VAV2 up-regulation in tumor vs. normal both in TGCA and in the Taylor dataset (36). VAV2 expression was higher in Gleason score tumors 8 to 10 relative to Gleason scores 6–7 (Fig. 7D), indicating a correlation of VAV2 expression with the aggressiveness of the tumor. Interestingly, analysis of available datasets providing expression on metastatic samples (34, 35, 37) revealed higher VAV2 expression in metastasis vs. primary tumors (Fig. 7E). The observed association between VAV2 high expression and poor clinical outcome highlights its potential prognostic value in human prostate cancer.

Figure 7.

VAV2 expression analysis from databases. A, Up-regulated Rac-GEFs in prostate cancer according to TCGA-PRAD. Left panel, heatmap. Right panel, p values for up-regulation in tumor vs. normal, disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI). B, Kaplan-Meier analysis for DSS, DFI, and PFI. C, VAV2 mRNA transcript expression in normal vs. tumor tissues for TCGA-PRAD and Taylor et al. cohorts. D, Expression of VAV2 mRNA transcript according to Gleason scores for TCGA-PRAD. E, Expression of VAV2 mRNA transcripts in primary prostate adenocarcinoma vs. metastatic prostate carcinoma from Varambally, Tomlins, and Grasso databases. ns, not significant, * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

Discussion

Rac1 is a well-established player in cellular responses associated with tumorigenesis and metastasis. This small G-protein has been extensively linked to the dynamics of actin cytoskeleton reorganization that occurs in response to extracellular cues in normal and cancerous cells, therefore playing a central role as a mediator of migratory responses (7–11, 23). Towards addressing the relevance of the Rac1 signaling pathway in cellular responses associated with prostate cancer progression, a largely understudied area, we pursued a thorough functional characterization of this small GTPase and the large family of Rac1 activating proteins, the Rac-GEFs, in cellular prostate cancer models. Our study shed light on the fundamental contribution of Rac1 in prostate cancer cell migratory and proliferative capacities and unambiguously established the impact of this small GTPase on the control of gene expression networks. A second major finding in our study is the unbiased identification of VAV2 as a mediator for the activation of Rac1 and a subset of Rac1-mediated responses downstream of EGFR, overtly revealing a signaling link contributing to the migration and proliferation of prostate cancer cells.

Dysregulation of Rac1 signaling is often considered a hallmark of malignant cells, including prostate cancer cellular models (8–11, 26). While mutations in Rac1 are rare and limited to few cancers (e.g., cutaneous melanoma) and other diseases (17, 58, 59), aberrant Rac1 activation more frequently implies the hyperactivation of membrane receptors or their effectors (such as PI3K), up-regulated expression or mutations of Rac-GEFs, or down-regulation of Rac-GAPs (9, 11, 18–20). Importantly, Rac1 hyperactivation in cancer cells has been often associated with deregulated tumor growth and metastasis (8–11, 18–20). A previous study from our laboratory reported robust constitutive Rac1 activation in androgen-independent cell lines. The precise nature of the mechanisms leading to the overactive Rac1 status in these cellular models remains elusive, particularly considering their unexpected insensitivity to Rac-GAP-mediated inactivation (26). By generating a Rac1-null DU145 cellular model, we could provide evidence that hyperactive Rac1 is a major driver of migratory and proliferative responses. Notably, knocking out Rac1 significantly alters the transcriptional profile of prostate cancer cells, a finding that is consistent with the demonstrated role of the Rac1 small GTPase in gene expression control (8, 9). Our analysis identified Rac1 as a major regulator of genes linked to cell adhesion, ECM, and cytokine function due to its prominent role in controlling fundamental processes associated with tumor cell growth and invasiveness. However, Rac1 did not seem to influence the characteristic mesenchymal phenotype of DU145 cells, as determined by the absence of changes in the expression of EMT transcription factors upon Rac1 KO. Although there have been no reported gene expression studies using Rac1 null cells, fluctuations of Rac1 activity have been linked to transcriptional heterogeneity, whereby high Rac1 activity correlates with the expression of invasion-associated gene signatures (60). More recently, single-cell transcriptome sequencing followed by functional clustering and pathway analysis in lung adenocarcinoma identified major associations of Rac1 with adhesion, ECM, and angiogenic signaling signatures. In line with this finding, inhibition of Rac1 in lung cancer cells attenuates proliferation, migration, and invasion (23, 42, 30). Furthermore, the characterization of Rac-GEF gene signatures in cancer models also provided similar strong associations with cancer progression pathways. One such example is the identification of P-Rex1-regulated gene networks involved in controlling ECM organization, migration, chemotaxis, and metastasis in breast cancer cells (44). The gene expression networks underscored in Rac1 KO prostate cancer cells strongly advocate for the foremost influence of Rac1 in controlling key cancer-related transcriptional signatures whilst highlighting the potential of Rac1 hyperactivation in influencing biological processes through transcriptionally driven mechanisms.

Mounting evidence suggests that membrane receptors, including RTKs, signal via specific GEFs to promote the activation of Rho family GTPases (12, 23–25, 29). However, the large number of GEFs encoded by the human genome (> 80 Rho family GEFs, with nearly half of them acting on Rac1), together with their distinctive cell-type expression pattern, makes the pairing of the individual receptor with specific Rac-GEFs exceedingly difficult. We developed a Q-PCR array for effectively assessing the relative expression of the 43 known Rac-GEFs belonging to the Dbl-like and DOCK families together with the ELMO adaptors, which proved to be highly resourceful for the identification and subsequent functional characterization of Rac-GEFs in several cancer cell models (23, 33, 29). Taking advantage of this resource, we could establish a rank of Rac-GEF expression in prostate cancer cells that helped narrow down candidate RTK effector Rac-GEFs. Specifically for our study, we used EGF as the stimulus for Rac1 activation for several reasons. First, EGF is a well-established stimulus for Rac1 activation in cancer cells (21, 23, 30, 52, 53). Second, EGFR has demonstrated roles in prostate cancer progression, including metastatic dissemination of prostate cancer cells (54–57). Lastly, previous analyses of EGF-mediated Rac-GEF/Rac1 activation in lung and breast cancer cells would allow us to establish commonalities and differences in Rac-GEF utilization among the different cancer models. Our unbiased screening revealed ECT2, VAV2, TRIO, PLEKHG2, FARP1, FARP2, and PREX1 as the top-expressed Rac-GEFs in prostate cancer cells regardless of their androgen-dependence traits. Quite remarkably, while many of these GEFs are also highly expressed in other cancer cell types, their requirement for Rac1 activation by EGF differs among the different models. For example, P-REX1 was the most relevant Rac-GEF downstream of EGFR in luminal breast cancer cells, whereas it was dispensable in the prostate (this study) and lung cancer cells (21). This is congruent with the reported nonessential role of P-REX1 in migratory and invasive activities of androgen-independent prostate cancer cells (26). The discovery of VAV2 as a mediator of EGF-induced Rac1 activation in prostate cancer cells also strikingly differs from the Rac-GEF requirement found in lung adenocarcinoma cellular models, wherein FARP1, ARHGEF39, and TIAM2 turned out to be the dominant EGFR effector Rac-GEFs (23–25). Whereas we cannot rule out that compensatory mechanisms preclude the identification of other Rac-GEFs as mediators of Rac1 activation by EGF in prostate cancer cells, our results unambiguously point to VAV2 as a fundamental player in this setting. Our results also revealed that DOCK1, DOCK5, DOCK7, and DOCK9 are expressed in prostate cancer cells. The DOCK GEFs were found to be dispensable for EGFR-mediated Rac1 activation in prostate cancer cells, which is strikingly contrasted with the requirement observed in other cancer models (61, 62). Taken together, our findings strongly support a model for the unique RTK coupling to the Rac-GEF/Rac1 pathway in a strict cell context manner. Mechanistically, this specificity may be dictated by the distinctive coupling of EGFR to adaptors and effectors in each cell type. As an example, EGF-mediated Rac1 activation in lung adenocarcinoma cells is mediated by EGFR coupling to adaptors GRB2/GAB1 and PI3K as the main effector (23). An attractive hypothesis is that alternative coupling via other EGFR adaptors (e.g., NCK, SHC1) and effectors (e.g., PLCγ, SHP2, SOS1) may occur in different cellular models. Additionally, the differential subcellular localization of Rac-GEFs according to the cell type may influence their access to the plasma membrane where nucleotide exchange activity primarily occurs.

It is essential to highlight that only a subset of Rac1-mediated responses, namely proliferation and migration, depend on VAV2. In contrast, Rac1-dependent gene expression seems unaffected by VAV2 RNAi depletion. A likely scenario is that VAV2 primarily acts as an effector for EGFR activation but not in response to other stimuli, including other factors in the serum. We postulate that different classes of receptors (e.g., RTKs vs. GPCRs), or even different receptors within each class, operate via distinct subsets of Rac-GEFs. This conclusion is predicted based on the complex and unique coupling mechanisms of individual receptor types. Indeed, it would be interesting to determine if other RTKs are relevant in prostate cancer progression, such as FGFR (63), and signal through the VAV2/Rac1 pathway in prostate cancer cells. An additional level of complexity entails the selectivity of GEFs towards different members of the Rho GTPase family. VAV2 is an archetypical non-selective GEF with the intrinsic ability to promote nucleotide exchange activity towards multiple Rho family members (64). We speculate that the selectivity of GEFs towards specific Rho GTPases is dictated by the nature of the stimulated receptor and their specific coupling mechanisms. A noteworthy recent example of the divergent activation of small GTPases depending on stimuli is FARP1, which promotes nucleotide activity towards Rac1 upon EGFR stimulation but solely acts on Cdc42 in response to GPCR stimulation (23, 29). Untangling the mechanistic insights behind this complex scenario would require wide-ranging screenings with multiple stimuli in different cancer cell models.

The prominent reliance on EGF-induced Rac1 activation and prostate cancer cell migration on VAV2 strongly supports the relevance of the EGFR→VAV2→Rac1 pathway as a major signaling cascade. Reliable validation of this pathway in prostate cancer cells was attained with EHop-097, a Rac-GEF/Rac inhibitor with selectivity for VAV GEFs. Indeed, this compound was nearly 100-fold more potent as an inhibitor of prostate cancer cell migration than other Rac inhibitors. While the selectivity of EHop-097 towards different members of the Dbl-like Rac-GEF family remains to be completed, the preeminent action of this inhibitor observed in our studies matches with results observed using VAV2 RNAi approaches.

Another salient observation in our study is the up-regulation of VAV2 in human prostate tumors. While analysis of publicly available databases (also reported by Burnstein and coworkers) (65) predicted that this was to be the case, our immunohistochemistry inquiry using a prostate cancer TMA unequivocally revealed elevated VAV2 staining in tumor areas compared to non-tumoral areas, with negative staining in stromal tissue. VAV2 overexpression has been reported in other human tumor types, including gastric, adrenocortical, breast cancer, and head and neck cancer. Moreover, high VAV2 tumor levels are associated with tumor invasion, metastasis, recurrence, and poor overall patient survival (48, 66–68). The fact that prominent differences in DSS, DFI, and PFI were observed depending on VAV2 expression largely supports the potential prognostic value of this GEF in human prostate cancer. While database analysis predicts elevated VAV2 expression in high Gleason prostate tumors as well as in those from patients who developed metastatic disease (see also 65), we were unable at this stage to prove this in our cohort based on the available patient information.

In summary, our study identified a signaling link leading to Rac1 activation in prostate cancer cells mediated by VAV2. The remarkable complexity of signaling mechanisms leading to Rac1 activation in cancer cells warrants a deep analysis of the Rac-GEF requirement for receptors and oncogenic inputs, as described for other cancer models (21–23, 29, 33. 43, 44). Considering the reported antitumorigenic and antimetastatic activities of Rac inhibitors in preclinical studies, our study provides a solid rationale for developing VAV2/Rac1 inhibitors as effective approaches for prostate cancer therapeutics.

Implications:

This study highlights the central role of VAV2 in prostate cancer cell proliferation and migration, as well as its potential prognostic value in disease progression.