Small Molecule Inhibition of Group I p21-Activated Kinases in Breast Cancer Induces Apoptosis and Potentiates the Activity of Microtubule Stabilizing Agents

Christy C Ong; Sarah Gierke; Cameron Pitt; Meredith Sagolla; Christine K Cheng; Wei Zhou; Adrian M Jubb; Laura Strickland; Maike Schmidt; Sergio G Duron; David A Campbell; Wei Zheng; Seameen Dehdashti; Min Shen; Nora Yang; Mark L Behnke; Wenwei Huang; John C McKew; Jonathan Chernoff; William F Forrest; Peter M Haverty; Suet-Feung Chin; Emad A Rakha; Andrew R Green; Ian O Ellis; Carlos Caldas; Thomas O'Brien; Lori S Friedman; Hartmut Koeppen; Joachim Rudolph; Klaus P Hoeflich


Breast Cancer Res. 2015;17(59) 

In This Article


PAK1 Amplification and Overexpression Are Associated With Poor Outcome in Luminal Breast Cancer

The genomic and transcriptomic architecture of 2,000 breast tumors was recently characterized as part of METABRIC.[15] Clustering analysis of joint copy number and gene expression data from the cis-associated genes revealed 10 novel molecular subgroups for breast cancer, including an estrogen receptor-positive subgroup composed of amplification at 11q13/14. This amplified region includes CCND1 (11q13.3) and PAK1 (11q14.1). Consistent with previous reports,[14] PAK1 mRNA expression was correlated with copy number gain (Figure 1A) and elevated in luminal breast cancer subtypes (Figure 1B) in METABRIC samples. To determine the prognostic significance of PAK1 in breast cancer, gene amplification was correlated with clinical outcome using a Cox proportional hazard model constructed with METABRIC censored survival data, patient age, NPI,[22] PAM50 breast cancer subtype classification, and CCND1/PAK1 amplification status. As expected, the NPI score, patient age and certain PAM50 subtypes were associated with worse hazard ratio (Figure 1B). High level, focal PAK1 amplification was significantly associated with poor patient outcome (P = 1.29 × 10−4) (Figure 1B; Additional file 1: Figure S1A, although this was not noted for either focal gain of CCND1 or broad amplification of both CCND1 and PAK1. Furthermore, elevated PAK1 mRNA expression was also correlated with poor survival in the same tissue samples (data not shown).

Figure 1.

p21-Activated kinase (PAK)1 copy number and expression is elevated and associated with poor clinical outcome in breast tumors analyzed by the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC). (A) Illumina mRNA expression of PAK1 is correlated with copy number alteration and breast cancer subtype in METABRIC tissue samples. "Amplification" is defined as gene amplification greater than or equal to 5 copies, while "Gain" is defined as >2 and <5 copies. (B) PAK1 focal amplification is associated with poor patient prognosis. Hazard ratio coefficient (log2 scale) is plotted for clinical and molecular parameters. Hazard ratio is significantly higher for PAK1 focal amplification relative to CCND1 focal amplification or dual PAK1/CCND1 amplification in luminal breast tumors (n = 980). P values are shown. CCND1, cyclin D1; Decade, patient age at time of diagnosis; Lum A, luminal A; Lum B, luminal B; NPI, Nottingham prognostic index.

The association of PAK1 dysregulation with survival of breast cancer patients was evaluated at the protein level in an independent sample set of 1,108 estrogen receptor-positive, early-stage breast tumors. PAK1 immunohistochemistry was validated previously[14] and breast tumor tissues were analyzed on a standard histology score of 0 to 3. Patients whose tumors had the lowest protein expression of PAK1 (staining intensity 0 to 1) displayed better overall survival than patients whose tumors had high PAK1 expression levels (staining intensity 2 to 3) (Additional file 1: Figure S1B; two group hazard ratio = 0.80). Taken together, these results indicate that PAK1 genomic amplification and overexpression are correlated with poor patient outcome in luminal breast cancer.

FRAX1036 Combines With Docetaxel to Alter Signaling to Microtubule Regulators and Induce Apoptotic Markers in Luminal Breast Cancer Lines

The small molecule pyridopyrimidinone inhibitor FRAX1036 (Figure 2A) was derived from chemical optimization of arylamino pyridopyrimidinone PAK inhibitors, as represented by FRAX597.[2,23] FRAX1036 is devoid of the characteristic arylamino moiety of earlier generation compounds, resulting in improved kinase selectivity (Additional file 2: Table S1 and general drug properties. Its biochemical potency (Ki) against PAK1 and PAK2 is 23.3 and 72.4 nM, respectively, with high selectivity against PAK4 (Ki = 2.4 μM; Figure 2B). In order to test the cellular activity of FRAX1036, breast cancer cell lines with known PAK1 gene amplification status were tested for levels of PAK1 expression and activity (Additional file 3: Figure S2 Potent cellular inhibition of group I PAK substrate phosphorylation (MEK1-S298 and CRAF-S338) was observed at 2.5 to 5 μM concentrations of FRAX1036 in PAK1-amplified MDA-MB-175 cells (Figure 2C). Consistent with previous reports evaluating PAK1 function in breast cancer cell lines via genetic approaches,[9] dose-dependent inhibition of PAK1 effector signaling was correlated with poly(ADP-ribose) polymerase (PARP) cleavage.

Figure 2.

FRAX1036 inhibition of group I p21-activated kinase (PAK) isoforms. (A) Chemical structure of the group I PAK inhibitor, FRAX1036. (B) Concentration-response analysis of FRAX1036 against PAK1, PAK2 or PAK4. Concentration response curves were generated in duplicate and represent one of at least three experiments for PAK1 and PAK2 with similar results. Data shown for PAK4 represent one of two experiments with similar results. Each curve is normalized to zero and 100% based on no enzyme or DMSO, respectively. (C) Pharmacodynamic changes induced by FRAX1036 dose–response. MDA-MB175 cells were treated with increasing concentrations of FRAX1036 for 24 hours. Cell lysates were immunoblotted with antibodies against biomarkers involved in PAK1 effector and survival signaling.

The PAK effector, stathmin, is a microtubule destabilizing protein and phosphorylation at serine 16 by PAK and other kinases, and regulates stathmin-tubulin binding.[18,24] We therefore hypothesized that PAK1 inhibition in combination with microtubule stabilizing chemotherapeutic agent taxanes, such as docetaxel (Taxotere, DTX), could synergistically alter microtubule dynamics in breast cancer cells leading to greater cell death.[25] FRAX1036 and docetaxel combination treatment of PAK1-amplified lines, MDA-MB-175 and HCC2911, elevated a major apoptotic marker (cleaved PARP) and attenuated a cell cycle regulator (cyclin D1) (Figure 3A). Single-agent docetaxel treatment increased stathmin-S16 phosphorylation indicative of the accumulation of cells in mitosis.[26] To further validate the observation of a combined effect of PAK1 inhibition and microtubule perturbation, the combination of docetaxel with PAK1 short interfering RNA knockdown was also investigated (Figure 3B). PAK1-dependent phosphorylation of stathmin was observed in breast cancer cells following selective knockdown (Figure 3B) or FRAX1036 treatment (Additional file 4: Figure S2B Comparable apoptotic signaling changes were observed for either PAK1 small molecule inhibition or knockdown in combination with docetaxel. Since combined inhibition altered signaling to caspase substrates (cleaved PARP), we further examined cell viability by monitoring apoptosis using high content time-lapse imaging. HCC2911 and MDA-MB-175 breast cancer cells treated with the combination of FRAX1036 and docetaxel showed increased kinetics of apoptosis compared with either single agent (Figure 3C,E). After 72 hours of treatment, the apoptotic index (ratio of apoptotic cells/total cells) was analyzed to account for cell proliferation (Figure 3D,F). FRAX1036 and docetaxel combination treatment resulted in a significantly greater apoptotic index over FRAX1036 or docetaxel alone in both cell lines. Synergistic modulation of cell viability was also demonstrated via FRAX1036 and docetaxel dose ranging and Bliss independence analysis (Additional file 4: Figure S3 Taken together, these data indicate that combinatorial increases in tumor cell killing are observed with FRAX1036 and docetaxel treatment.

Figure 3.

FRAX1036 and docetaxel (DTX) combine to alter stathmin phosphorylation, induce the apoptotic marker cleaved PARP and increase kinetics of apoptosis. (A) MDA-MB-175 and HCC2911 cells were treated with DMSO, 5 μM FRAX1036, 0.2 μM docetaxel and a combination of 5 μM FRAX1036 and 0.2 μM docetaxel for 24 hours. Cell lysates were immunoblotted with apoptotic and PAK1 downstream markers. (B) MDA-MB-175 cells were treated with DMSO or 0.2 μM docetaxel for 48 hours after non-targeting control short interfering RNA (siRNA) or PAK1 siRNA transfection for 72 hours. Cell lysates were harvested and subjected to immunoblot analysis for apoptotic markers and microtubule regulators. The molecular weight of the lower band from the phospho-stathmin immunoblot corresponds to total stathmin. The efficacy of knockdown by PAK1 siRNA was 47% (lane 2) and 80% (lane 4) as determined by densitometry. (C) Kinetic apoptosis assay. HCC2911 cells were plated in 96-well plates and were untreated (control) or treated with DMSO, 2.5 μM FRAX1036, 0.2 μM docetaxel, or a combination of 2.5 μM FRAX1036 and 0.2 μM docetaxel. Apoptosis was assayed by counting the number of green caspase 3/7-positive objects at each time point (Essen Cell player kinetic caspase 3/7 assay). (D) Apoptotic index. The number of apoptotic cells was normalized to the total number of cells at the final time point in (C) to account for cell proliferation. (E,F) The same as (C,D) with MDA-MB-175 cells. The average and SEM of three replicates are shown and a t-test performed at the final time point and on the apoptotic index (*P < 0.03, **P < 0.003, ***P ≤ 0.0001).

FRAX1036 and Docetaxel Combination Alter Microtubule Organization, Duration of Mitotic Arrest and Kinetics of Apoptosis

Both PAK1 signaling and docetaxel were previously reported to affect microtubule dynamics and progression through mitosis.[25,27] Regulation of microtubule dynamics ultimately affects microtubule length and organization of microtubule arrays. To directly visualize the effects of FRAX1036 and docetaxel treatments on microtubules, we utilized a U2OS osteosarcoma cell line that stably expresses RFP-Tubulin and GFP-Histone H2B. The flat and spread morphology of U2OS cells was more amenable to high-resolution microscopy, allowing us to visualize microtubules in live cells without fixation. Analysis of pharmacodynamic markers and apoptosis (Additional file 5: Figure S5A,B confirmed that U2OS cells are similarly affected by FRAX1036, docetaxel and combination treatment as the breast cancer lines examined in this study. In DMSO-treated cells, microtubule arrays were organized by the microtubule organizing center and radiate uniformly to the cell periphery. After 20 hours of treatment with FRAX1036, microtubules were disorganized and were not evenly distributed throughout the cytoplasm, between the microtubule organizing center and the periphery (Figure 4A, arrow). As expected, docetaxel stabilized microtubules resulting in elongated bundles of microtubules that curved around the cytoplasm. Cells treated with both FRAX1036 and docetaxel had shorter, straight microtubules, suggesting a change in the regulation of microtubule dynamics from docetaxel treatment alone. In addition, immunofluorescence of fixed MDA-MB-175 cells treated with FRAX1036 and docetaxel showed similar effects on microtubule organization (Additional file 6: Figure S4 To further probe the relationship between cell cycle progression and apoptosis we imaged RFP-Tubulin and GFP-Histone H2B U2OS cells over a 72-hour treatment period using spinning disk confocal microscopy (Additional file 7: Movie 1 We tracked individual cells for each of the treatment conditions and visualized mitotic spindle formation to analyze timing and fate (division, slippage or cell death) after entering mitosis (Figure 4B,C; Additional file 5: Figure S5C FRAX1036-treated cells completed normal mitoses with the majority of apoptosis occurring during interphase (66.7%). Because the cells are not synchronized it was not clear from our analysis whether a completed mitosis was required for apoptosis. Docetaxel-treated cells arrested in mitosis five-fold longer than control cells slipped out of mitosis (71.4%) without completing cell division and formed micronucleated cells that later died (Figure 4A; Additional file 5: Figure S5C, + symbol In contrast, when FRAX1036 was combined with docetaxel, there was a small decrease in the duration of mitosis and the majority of these cells died during mitotic arrest (65.9%), possibly accounting for the increased rate of cell death. To confirm these results in breast cancer cells, we imaged MDA-MB-175 cells by phase-contrast microscopy (Additional file 8: Movie 2 Without clear visible markers of entry into mitosis, MDA-MB-175 cells were difficult to track and quantitate due to the densely packed and rounded morphology. However, the overall trends in cell fate could be observed: docetaxel treatment resulted in slipped cells with micronuclei that later died, while dead cells accumulated more quickly in the FRAX1036 and docetaxel combination. The dependency of FRAX1036 and docetaxel combination effects on the order of drug treatment was also determined. A pronounced decrease in cell viability was observed by simultaneous treatment of compounds and when docetaxel was dosed prior (4 hours) to FRAX1036 (Additional file 9: Figure S6

Figure 4.

FRAX1036, docetaxel (DTX) and their combination affects microtubule organization, mitosis and cell fate. (A) Spinning-disk confocal images of live U2OS cells expressing red fluorescent protein (RFP)-Tubulin (red) and green fluorescent protein-Histone H2B (green) bottom panel. The top panel is RFP-Tubulin channel alone with an individual cell outline by a dotted line for each condition. Arrows highlight changes in microtubule organization that are characteristic of each treatment. A micronucleated cell is indicated by +. Cells were treated with DMSO, 2.5 μM FRAX1036, 0.2 μM docetaxel, or a combination of 2.5 μM FRAX1036 and 0.2 μM docetaxel for 20 hours before imaging. Scale bar = 20 μm. (B) Duration of mitosis/mitotic arrest of cells treated with DMSO, 2.5 μM FRAX1036, 0.2 μM docetaxel, or a combination of 2.5 μM FRAX1036 and 0.2 μM docetaxel. Cells were followed from time of entering mitosis to the time of division, slippage or apoptosis. Each grey symbol represents a single cell and black bars represent the average. N = 42 mitotic cells imaged from five fields of view. Data is from one of two experiments with similar results. One-way analysis of variance with multiple comparisons showed that all averages are significantly different except for DMSO:FRAX1036. A t-test was performed on FRAX1036 + DTX combination vs DTX alone (P = 0.0002). (C) Distribution of major cell fates after entry into mitosis of U2OS cells treated with FRAX1036, docetaxel and their combination. N = 42 mitotic cells for each treatment condition.