Introduction
The p21-activated kinases (PAKs) have generated significant interest as therapeutic targets in cancer [
1,
2]. The PAK family is comprised of six members and is subdivided into two groups (Groups I and II) based on sequence and structural homology. PAKs are currently amongst the most well-characterized effector proteins of the Ras-related C3 botulinum toxin substrate 1 (Rac) and cell division control protein 42 (Cdc42). These GTPases stimulate PAK catalytic activity by relieving an intramolecular interaction between the kinase and autoinhibitory domains. The kinase domains of Group I versus II PAKs share approximately 50% identity and also share homology with additional members of the sterile-20 (STE20) subfamily of the kinome that are upstream activators of mammalian mitogen-activated protein kinase (MAPK) pathways.
PAK1 signaling has been shown to be important for regulating cytoskeletal organization and cell migration via both its catalytic activity and protein-protein interactions. For instance, PAK1 modulates the activity of myosin II (an actin interacting motor protein that can drive cell contractility), LIM-kinase (involved in actin polymerization through inactivation of cofilin family proteins) and filamin A (a large actin-binding protein that induces membrane ruffling) [
3]. PAK1 is also involved in the phosphorylation of proteins that control microtubule dynamics such as stathmin, which destabilizes microtubules by binding tubulin dimers to inhibit tubulin polymerization and promote microtubule disassembly [
4]. In addition, PAK1 phosphorylates tubulin cofactor B to augment heterodimerization of tubulin [
5] as well as dynein light chain 1 which is a component of the cytoplasmic dynein complex that moves along with microtubules [
6]. To date, the evidence for the role of PAK1 in microtubule remodeling comes primarily from overexpression and genetic studies. For instance, PAK1
−/− mouse embryonic fibroblasts display decreased microtubule regrowth and polymerization compared with wild-type cells, and the reciprocal phenotypic was observed using MCF7 breast cancer cells overexpressing PAK1 [
7]. The contribution of PAK1 catalytic activity to microtubule dynamics has yet to be thoroughly explored.
In addition to its role in regulation of the cytoskeleton, PAK1 has been implicated in cellular processes that directly contribute to tumorigenesis, including growth factor pathways, cell proliferation, and pro-survival signaling [
8]. PAK1 is also an effector of well-established oncogenes, such as the Ras small monomeric GTPase which is mutated in approximately 30% of human tumors. Given that Rac and Cdc42 lie downstream of Ras [
9,
10], several groups have evaluated the contribution of PAKs to Ras-driven cellular transformation and
in vivo tumorigenesis [
11,
12]. For instance, PAK1 deletion in a mouse model of Ras-driven cutaneous squamous cell carcinoma led to markedly decreased tumorigenesis and progression, which was accompanied by attenuated signaling through MAPK and cytoskeletal pathways [
11].
In terms of direct dysregulation in cancer, PAK1 is amplified, overexpressed or hyperactivated in several tumor subtypes [
1,
13]. Of note, focal genomic amplification of PAK1 at 11q14.1 has been reported for hormone receptor-positive breast carcinoma [
14,
15]. Analysis of breast cancer cell lines with PAK1 genomic copy number gain using RNA interference approaches revealed dependence on PAK1 expression for cell survival [
14] and transformation [
16]. Consistent with these findings, functional studies using transgenic mouse models have also demonstrated that overexpression of PAK1 in the mammary gland promotes the formation of preneoplastic lesions and breast tumors [
17] and that PAK1 contributes to human endothelial growth factor receptor 2 (HER2)/Neu-driven tumorigenesis [
18].
However, given this emerging body of work, a detailed assessment of PAK1 copy number alteration and validation experiments using small molecule inhibitors to evaluate PAK1 catalytic inhibition in breast cancer are still lacking. Moreover, the potential efficacy of PAK1 inhibition in combination with additional inhibitors of cytoskeletal organization has yet to be examined. Herein, we demonstrate that PAK1 gene amplification and protein overexpression are associated with poor clinical outcome in a large collection of luminal breast cancers. We also introduce a novel ATP-competitive small molecule inhibitor of group I PAKs, FRAX1036, and demonstrate sensitivity of PAK1-amplified breast cancer cells to this compound. Taken together, these results suggest that further investigation of PAK1 as a therapeutic target in breast cancer is warranted. Given that PAK1 regulates the cytoskeleton and microtubule inhibitors are used as standard-of-care chemotherapy in advanced breast cancer, we explored the molecular and cellular mechanisms for this therapeutic combination and showed increased anti-tumor efficacy in breast cancer cells.
Materials and methods
Materials, cell culture and viability assays
FRAX1036 was synthesized by Afraxis, Inc. (La Jolla, CA, USA) and docetaxel was purchased from Selleck Chemicals (Houston, TX, USA). Antibodies used for immunoblotting (p-MEK1-S298, p-CRAF-S338, Cleaved PARP, Cyclin D1, p-Stathmin-S16, p-β-catenin-S675, MCL-1, BCL-xL, p-Bad-S112 and PAK1) were purchased from Cell Signaling Technology (Danvers, MA, USA); anti-Actin was purchased from Sigma (St Louis, MO, USA). Cell lines were acquired from the American Type Culture Collection (ATCC; Manassas, VA, USA) and maintained at 37°C and 5% CO2 in RPMI 1640 media with 10% fetal bovine serum and 2 mM L-glutamine. U2OS-red fluorescent protein (RFP)-Tubulin cells (Marinpharm, Luckenwalde, Germany) were stably transduced with a plasmid expressing green fluorescent protein (GFP)-histone H2B. Cell transfections and treatments were performed using short interfering RNA oligonucleotides for PAK1 from Dharmacon RNAi Technologies (Chicago, IL, USA). Cellular viability was assessed via ATP content using the CellTiter-Glo Luminescent Assay (Promega, Madison, WI, USA) and results represent mean ± standard deviation from three experiments.
PAK1/CCND1 survival analysis
Breast tumors from the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) dataset [
15] with survival and DNA copy number data were selected, yielding 980 patients. DNA copy number was calculated using Affymetrix SNP6.0 arrays and a modified version of the PICNIC algorithm [
19], published recently [
20]. Samples were identified as having amplification of either
PAK1 or
CCND1 if the absolute copy number of the respective gene was >5 copies. The Kaplan-Meier plot and log-rank test were performed using the censored survival values (days since diagnosis) provided with the METABRIC dataset and our calculated
PAK1 amplification status using the R language [
21], version 3.1, and the R package “survival”, version 2.37-7.
A Cox proportional hazard model was constructed using the METABRIC censored survival data, Nottingham prognostic index (NPI), patient age, and patient PAM50 breast cancer subtype classification in addition to the interaction of CCND1 and PAK1 amplification statuses. More specifically, the model “survival ~ NPI + age + PAM50 + CCND1 * PAK1” was fit using the “coxph” R package, where ccnd1 and pak1 are binary variables, as discussed above. The forest plot was produced using the coefficients from this model and their P-values. The whiskers on this plot represent ±1.96 × the standard error for each coefficient. The coefficient for amplification of both CCND1 and PAK1 (dual amplification) in the same sample was calculated as the sum of the coefficients “pak1Amplified”, “ccnd1Amplified”, and the coefficient for the interaction term for these two terms.
Bliss analysis
Cellular viability was assessed via ATP content using the CellTiter-Glo Luminescent Assay (Promega, Fitchburg, WI, USA) after a 4-day incubation period, and results represent mean ± standard deviation from three experiments. Total luminescence was measured on a Wallac Multilabel Reader (Perkin-Elmer, Waltham, MA, USA). Cells were treated simultaneously with FRAX1036 (dose range = 0 to 5 μM) or docetaxel (dose range = 0 to 0.4 nM) in an 8 × 10 matrix of concentrations. Combination synergy of FRAX1036 and docetaxel was determined by Bliss independence analyses. A Bliss expectation for a combined response C was calculated by the equation: C = (A + B) - (A × B) where A and B are the fractional growth inhibitions of given doses of drug A and B. ΔBliss scores were summed across the dose matrix to generate a Bliss sum. Bliss sum = 0 indicates that the combination effect is additive while Bliss sum >0 indicates synergy effect and Bliss sum <0 indicates antagonism effect. Statistical analysis comparing the Bliss sums for each cell line was conducted by the Student’s t test.
Biochemical assays
The activity/inhibition of human recombinant PAK1 (kinase domain), PAK2 (full length) or PAK4 (kinase domain) was estimated by measuring the phosphorylation of a FRET peptide substrate (Ser/Thr19) labeled with Coumarin and Fluorescein using Z’-LYTE™ assay (Invitrogen, Carlsbad, CA, USA). The 10 μL assay mixtures contained 50 mM HEPES (pH 7.5), 0.01% Brij-35, 10 mM MgCl2, 1 mM EGTA, 2 μM FRET peptide substrate, and PAK enzyme (20 pM PAK1; 50 pM PAK2; 90 pM PAK4). Incubations were carried out at 22°C in black polypropylene 384-well plates (Corning Costar, Corning, NY, USA). Prior to the assay, enzyme, FRET peptide substrate and serially diluted test compounds were preincubated together in assay buffer (7.5 μL) for 10 minutes, and the assay was initiated by the addition of 2.5 μL assay buffer containing 4× ATP (160 μM PAK1; 480 μM PAK2; 16 μM PAK4). Following the 60-minute incubation, the assay mixtures were quenched by the addition of 5 μL of Z’-LYTE™ development reagent, and 1 hour later the emissions of Coumarin (445 nm) and Fluorescein (520 nm) were determined after excitation at 400 nm using an Envision plate reader (Perkin Elmer). An emission ratio (445 nm/520 nm) was determined to quantify the degree of substrate phosphorylation.
Immunoblotting
Protein extracts were prepared at 4°C with RIPA Lysis Buffer (EMD Millipore Corporation, Billerica, MA, USA), 1 mM phenylmethylsulphonyl fluoride, Phosphatase Inhibitor Cocktail 2/3 and protease inhibitor cocktail (Sigma-Aldrich). For Western blot analysis, proteins were resolved by 4 to 12% SDS-PAGE and transferred to nitrocellulose membranes (Life Technologies, Grand Island, NY, USA). Immunoblotting was performed using the indicated primary antibodies and analyzed using secondary antibodies for enhanced chemiluminescence.
IncuCyte apoptosis assays
For caspase 3/7 activation apoptosis assays, cells were plated at 10,000 cells/well in 96-well Corning plates for 24 hours prior to treating with DMSO, FRAX1036, and/or docetaxel. Caspase 3/7 reagent was added at a 1:1000 dilution (Essen Bioscience No. 4440, Ann Arbor, MI, USA). Cells were imaged at 10× magnification in an IncuCyte Zoom Live-content imaging system (Essen Bioscience) at 37°C, 5% CO2. Images were acquired every 2 hours or 4 hours for 36 to 72 hours, two images/well. Data was analyzed using IncuCyte analysis software to detect and quantify green (apoptotic) cells/image. Each condition was performed in triplicate. Averages with SEM at each time point were plotted in Excel (Microsoft, Redmond, WA, USA). A t-test was performed for the final time point comparing the combination of FRAX1036 and docetaxel with each single agent in Prism (Graphpad, La Jolla, CA, USA). The apoptotic index was calculated from the apoptosis assays by dividing the final apoptotic cell count by the total cell count. Averages with SEM were plotted in Excel (Microsoft), and a t-test was performed comparing the combination of FRAX1036 and docetaxel with each single agent in Prism (Graphpad).
Live-cell microscopy and image analysis
U2OS cells stably expressing GFP-Histone H2B and RFP-Tubulin were cultured in Dulbecco’s modified Eagle’s high glucose medium, 10% fetal bovine serum, NEAA, at 37°C and 5% CO2. For live imaging experiments, U2OS cells were plated in a 24-well glass bottom, black-walled plate (Sensoplate #662892, Greiner Bio-one, Monroe, NC, USA). On the following day, cells were treated with DMSO, FRAX1036, and/or docetaxel and imaged every 10 minutes for 72 hours with a 40× ELWD Plan Fluor objective (NA: 0.6, Nikon, Tokyo, JP) at 37°C and 5% CO2. Imaging was performed on a Nikon Ti-E perfect focus inverted microscope equipped with a spinning disk confocal CSU-X1 (Andor, Oxford Instruments, Abingdon, Oxfordshire, UK), motorized X,Y stage (Nikon), environmental chamber (OkoLab, Burlingame, CA, USA) and iXon3 897 EMCCD camera or Clara interline CCD camera (Andor, Oxford Instruments), all controlled by NIS-Elements software (Nikon, Tokyo, JP). Time-lapses were analyzed in NIS-Elements, and supplemental movies were generated in Quicktime Pro (Apple, Cupertino, CA, USA). For high-resolution imaging of microtubule organization, U2OS cells were imaged after 20 hours treatment with a 60× Plan Apo objective (NA: 1.4, Nikon). For immunofluorescence, MDA-MB-175 cells were methanol fixed, permeabilized in TBS-0.5% TritonX-100 and blocked in 2% bovine serum albumin, 0.1% Triton X-100. Microtubules were probed with primary antibody rat-anti tubulin (1:250, Serotec clone YL1/2), secondary antibody Alexafluor488 anti-rat (1:500, A-11006, Life Technologies, Grand Island, NY, USA), and mounted with Prolong Gold with DAPI (Life Technologies, Grand Island, NY, USA).
For the duration of mitosis/mitotic arrest and cell fate measurements, cells were monitored from the time they began to round up from the plate to the time when they were observed to divide, slip out of mitosis with micronuclei, or apoptose. Observations were made using the phase morphology of the cells as well as chromosome condensation/decondensation and mitotic spindle morphology in the fluorescent channels. Cells that divided or slipped were monitored for the remainder of the 72-hour movie, and subsequent cell events were recorded. Duration of mitosis/mitotic arrest was graphed in Prism (GraphPad), and significance was determined by one-way analysis of variance with multiple comparisons to compare each condition to one another. A t-test was performed on the two significant but close conditions (docetaxel and FRAX1036 + docetaxel).
Discussion
The advent of high-throughput techniques for genetic and epigenetic characterization of tumor specimens has led to an exponential increase in our understanding of molecular events underlying the process of carcinogenesis. This is especially true for breast cancer, an indication in which tumor tissues can be successfully obtained and analyzed with high frequency. Typically, novel putative driver genes for breast cancer have been preliminarily evaluated using genetic and knockdown approaches. However, more comprehensive and rigorous assessment of intracellular targets for therapeutic intervention requires selective, potent and cell-active small molecules with good biochemical and cellular properties. FRAX1036 displays selectivity for PAK1-3 relative to group II PAK members as well as other kinases (Additional file
2: Table S1) and can be used as a tool compound for
in vitro target validation experiments.
Interestingly, PAK1 genomic amplification or protein overexpression are strongly associated with poor outcome for luminal (or estrogen receptor-positive) breast cancer patients (Figure
1; Additional file
1: Figure S1). A subset of breast carcinomas without genomic amplification also display high mRNA and protein expression of PAK1 (Figure
1A; Additional file
1: Figure S1B). The molecular mechanisms underpinning dysregulated PAK1 expression in the absence of genomic amplification are not well characterized, although regulation by microRNAs [
28] and gene translocation (Peter Haverty, unpublished data) have both been observed.
PAK1 copy number alterations have also been observed in other tumor indications, such as ovarian cancer and melanoma [
16,
29] and further validation efforts are necessary to apply the findings reported here to these other indications.
Hormone receptor-positive breast cancer patients with localized disease receive front-line treatment with endocrine therapies, such as tamoxifen or aromatase inhibitors. There is some evidence that PAK1 may directly phosphorylate estrogen receptor-α [
30] or components of the estrogen receptor multi-protein complex [
31]. However, the potential roles for PAK1 inhibition in combination with later lines of therapy, such as taxanes, have yet to be explored. Given the evolutionarily conserved role of PAK1 in regulating cytoskeletal dynamics and the common use of microtubule inhibitors in later lines of breast cancer treatment, we evaluated the mechanism and potential therapeutic benefit of FRAX1036 combination with docetaxel. We show that signaling changes elicited by FRAX1036 and docetaxel potentiate apoptosis of breast cancer cells and that microtubule morphology is affected by both pathways (Figure
4A). Combination treatment of FRAX1036 significantly diminished time in docetaxel-induced mitotic arrest (Figure
4C; Additional file
5: Figure S5), pushed cell fate from mitotic slippage to apoptosis (Figure
4B) and accordingly increased the kinetics of breast tumor cell apoptosis (Figure
3C-F; Additional file
3: Figure S2). Given that luminal breast cancer patients generally do not respond durably to chemotherapy, combination of FRAX1036 with taxanes may help address unmet needs for patients with advanced and metastatic disease.
Competing interests
CCO, SG, CKC, MSa, WZho, AMJ, LS, MSc, WFF, PMH, TOB, LSF, HK and JR own stock in Hoffmann-La Roche. KPH is a shareholder of Blueprint Medicines. DAC and SGD own shares in Afraxis. CC receives research funding from Genentech and is a member of the Scientific Advisory Board at AstraZeneca. CP, WZhe, SD, MS, NY, MLB, WH, JCM, JC, SFC, EAR, ARG and IOE declare that they have no competing interests.
Authors’ contributions
CCO, SG, CP, MSa, JR and KPH conceived and designed the experiments. CKC, WZho and LS performed in vitro experiments. WFF and PMH conducted bioinformatics and statistical analyses. AMJ, HK, SFC, CC, EAR, ARG and IOE provided tumor tissues, associated molecular data and conducted histological analyses. SGD, DAC, WZhe, SD, MSh, NY, MLB, WH and JCM developed FRAX1036; MSc, TOB, LSF and JC provided key reagents and essential guidance on the development of cell-based assays and interpretation of data. These efforts included the apoptosis assays to demonstrate the activity of FRAX1036 in cells with PAK1 genomic amplification as well as the cell cycle analyses using cells engineered with reporter constructs. The manuscript was written by CCO, SG, JR and KPH. CCO, SG, MSa, PMH, EAR, ARG, JR and KPH revised the manuscript. All authors approved the submitted version of the manuscript.