Depleting tumor-NQO1 Potentiates Anoikis and Inhibits Growth of NSCLC

Abstract

The fundamental role that NAD(P)H/quinone oxidoreductase 1 (NQO1) plays, in normal cells, as a cyto-protective enzyme guarding against stress induced by reactive oxygen species (ROS) is well documented. However, what is not known is whether the observed overexpression of NQO1 in neoplastic cells contributes to their survival. The current study discovered that depleting NQO1 expression in A549 and H292 lung adenocarcinoma cells caused an increase in ROS formation, inhibited anchorage-independent growth, increased anoikis sensitization and decreased 3-D tumor-spheroid invasion. These in vivo data further implicate tumor-NQO1 expression in a pro-tumor survival role, since its depletion suppressed cell proliferation and decreased lung tumor xenograft growth. Finally, these data reveal an exploitable link between tumor-NQO1 expression and the survival of lung tumors since NQO1 depletion significantly decreased the percentage of ALDH^(high) cancer cells within the tumor population.

Introduction

Lung cancer is the leading cause of cancer related deaths in the U.S. (1). Over the past decade some improvement has been made toward the goal of increasing overall survival in lung cancer patients. These improvements have mostly been due to technological advances allowing early diagnosis of lung cancer as well as improved molecular based therapeutic approaches (2). However, with 5-year survival rates at 15% or less, novel mechanism based therapeutic approaches are still desperately needed.

NADPH quinone oxidoreductase-1 (NQO1) is an inducible two-electron oxidoreductase that is highly overexpressed in many solid tumors including breast, pancreas and non-small cell lung cancer (NSCLC) (3-10). NQO1 is an essential phase II detoxification gene and as such plays a critical role in both detoxification and bio-activation of many DNA damaging quinones (5). As a chemo-preventive gene, NQO1 has been shown to detoxify a broad spectrum of quinone substrates and it plays a role in reactive oxygen species (ROS) scavenging by generating antioxidant forms of alpha tocopherol (5, 11).

In our past investigations we demonstrated that NQO1 bioactivated several anticancer quinones including β-lapachone (12) and more recently deoxynyboquinone (DNQ) (13). Our previous in vitro studies determined that NQO1 is a viable target for developing personalized lung cancer therapy since tumor-NQO1 levels are often 5-20 fold greater in lung tumors as compared to the levels of NQO1 observed in associated normal tissues (9). Thus, targeting NQO1 with anticancer quinones has become a feasible option for preclinical anticancer studies. Furthermore, our in vivo studies with anticancer quinones and novel drug delivery formulations, has led to a surge in interest in NQO1-bioactivated anticancer quinones (13, 14), resulting in clinical trials for treatment of various solid tumors. However, there is still very little known as to why NQO1 levels are so vastly overexpressed in solid tumors. More specifically, no studies have addressed whether reducing tumor-NQO1 levels affects processes critical to tumor survival and proliferation, including anchorage-independent growth, escape from apoptosis and the ability to invade and metastasize.

In the current study we hypothesized that depleting NQO1 expression levels in NSCLC tumors would have deleterious effects on cell proliferation and survival. Our rationale for this hypothesis stemmed from numerous reports suggesting that cancer cells must regulate oxidative stress levels to prevent death from toxic levels of ROS created in their microenvironment as part of a host defense response (15). Thus, one strategy to protect tumor cells from lethal levels of ROS stress is to activate, or hijack, pathways that regulate the expression levels of antioxidant genes. Importantly, a primary regulator of oxidative stress is the transcription factor Nrf2 whose role is to activate antioxidant gene expression; and its own overexpression has been associated with enhanced tumorigenesis (16-18). One of the many transcriptionally activated antioxidant genes regulated by Nrf2 is NQO1, and numerous studies have shown that NQO1 levels in various tumors are elevated in comparison to associated normal tissues (3, 6, 9). Here we show that depletion of NQO1 expression levels, in various NSCLC cell lines, decreased the tumor cells ability to form colonies in anchorage-independent growth assays. The inability of NQO1-depleted NSCLC cells to form tumor colonies in anchorage-independent assays correlated with increased reactive oxygen species formation, an increase in anoikis sensitization and a decrease in cell proliferation rates. Our data also show that depletion of NQO1 expression levels inhibited the ability of NSCLC cells to invade in 3D-tumor spheroid assays. Our in vivo data show that loss of tumor-NQO1 expression in NSCLC cells inhibited tumor growth as compared to controls. Finally, we show that NQO1 knockdown decreases the percentage of ALDH^(high) cancer cells, suggesting that the depletion of NQO1 decreases tumorigenicity by eliminating the cancer stem cell population within the tumor. Together these novel findings illuminate the role of NQO1 in tumors, and suggest that depleting tumor-NQO1 levels disrupts the protective barrier against ROS provided to cancer cells by elevated tumor-NQO1 expression levels. Thus, NQO1 depleted tumor cells are more susceptible to oxidative stress and their overall growth and survival is inhibited due to increased cell death, and reduced proliferation of the cancer stem cell population.

Materials and Methods

Reagents

NQO1 activity assay kit (Abcam), Cell death detection ELISA kit (Roche Applied Sciences), Seaplaque agarose, SeaKem agarose, 1N Sodium Hydroxide and Rat tail collagen type I (Fisher Scientific), Noble agar (Becton, Dickinson), 10X DPBS (Hyclone), Cyquant cell proliferation assay kit and 2’, 7’-dichlorodihydrofluorescein diacetate, acetyl ester, DCFDA (Lifetechnologies). The NQO1 inhibitor Mac220 was a generous gift from Dr. David Ross, University of Colorado Anschutz Medical Center.

Cell growth and maintenance assays

H292, HCC1171 and non-transformed, non-tumorigenic human bronchial epithelial (HBEC) cell lines were a generous gift from the laboratory of Dr. John D. Minna, UTSW Medical Center at Dallas. A549 and H596 cells were previously described (9). A549, H596, H292 and HCC1171 cell lines were cultured in DMEM (Lonza) containing 10% fetal bovine serum (FBS) and 1% L-glutamine. HBEC cells were cultured in Keratinocyte Serum-Free Media with supplements (Invitrogen). All cell lines were incubated at 37°C at 10% CO_2.

Western Blotting

Protein lysates were separated by 10% SDS-PAGE and transferred onto a PVDF membrane. Membranes were blocked with 5% milk in PBST for 1 hour at room temperature, and then incubated overnight with β-actin (1:5000 in 3% BSA, Santa Cruz Biotechnology) at 4°C. Blots were washed in PBST and incubated for 1 hour with 1:5000 dilution of goat-antimouse IgG-HRP in 5% milk in PBST. The process was repeated using a 1:5000 dilution of monoclonal NQO1 antibody (clone A-180, Santa Cruz Biotechnology). Pierce ECL western blotting substrate (Thermo Scientific) was used to visual bands on Hyblot-CL autoradiography film (Denville Scientific). For PARP-1 cleavage assays, A549 (shCtr-R and shNQO1) tumors were harvested post mortem and sonicated in PARP-lysis buffer as described previously (12, 19). Extracts were resolved in 10% SDS-PAGE gels and transferred to a PVDF membrane. PARP-1 protein was visualized using a monoclonal PARP-1 antibody (Santa-Cruz, clone F-2) at a 1:1000 dilution.

Patient survival analysis

Three separate survival analyses were performed on overall survival data from NSCLC patients obtained from 3 TCGA data sets (https://tcga-data.nci.nih.gov/tcga/) (20-22). Gene expression values used were reported from Affymetrix U133A Microarray data. Patient data with valid gene expression levels were used to estimate medians and bounds for upper and lower quartiles. Patients were categorized into two groups based on whether the values of gene expression were above the upper quartile bound and below the lower quartile bound. Kaplan–Meier survival graphs were plotted, and log-rank tests were performed using GraphPad Prism.

Transient and stable NQO1 protein knockdown assays

The human shRNA-NQO1 retroviral vector (RHS1764-9691437) was purchased from Open Biosystems. A stable shRNA knockdown cell line (A549-shNQO1) and vector control (A549-shCtr-R) was generated by infecting A549 cells with polybrene-supplemented medium obtained from Phoenix packaging cells transfected with the human retrovirus vector targeting NQO1 or non-silencing control vector as described previously (7). Human shRNA-NQO1 lentiviral particles (sc-37139) and controls were purchased from Santa Cruz Biotechnology. Stable shNQO1 lentiviral knockdown (A549-shNQO1-B) and control (A549-shCtr-L) lines were generated by infecting A549 cells with polybrene-containing culture medium (5 μg/mL) and 10 μL of the lentiviral particles were added directly to the culture medium. Medium was changed 24 hours after transfection. After 48 hours shNQO1 containing cells were isolated by limited dilution in media containing puromycin (2 μg/mL) and screened for NQO1 expression levels by Western blot. Similar experiments were performed with H292 cell lines to create H292-shNQO1-B and H292-shCtr-L cell lines. H596 cells, which are NQO1 null, were infected with retrovirus particles from a retroviral control (LPC-X) or retroviral NQO1 (LPC-NQO1) vector as described previously (9). For transient NQO1 knockdown, siRNA-NQO1 or scramble control siRNA (Santa Cruz Biotechnology) was transiently transfected into HCC1171 or H596 cell lines (Lipofectamine 2000, Life Technologies) using the Life Technologies protocol. Cells were harvested after 48 hours and analyzed for NQO1 protein expression or enumerated using a hemocytometer for use in invasion assays.

NQO1 activity assays

To analyze endogenous NQO1 levels we used an NQO1 activity assay kit (Abcam). Briefly, cell pellets, containing 2X10⁷ cells, were collected for each cell line. Pellets were solubilized in 1X extraction buffer on ice for 20 minutes. Samples were then centrifuged at 18,000 × g for 20 minutes at 4°C. Supernatants were transferred to new eppendorf tubes and aliquots were stored at −80°C. Protein concentration was determined using the Bio-Rad protein assay method. Samples were diluted to 2X the working concentration of 5 μg/mL with supplemented buffer. Two wells were prepared for each sample (one well with and one well without inhibitor). 50 μL of each cell line was plated in triplicate in 96 well plates provided with the kit. The reaction buffer and the reaction buffer plus inhibitor were prepared according to the manufacturer's calculation table. The reaction buffer plus inhibitor were added to the samples first. Reaction buffer without the inhibitor were added last. Absorbance was measured at 440 nm every 20 seconds for 5 minutes using the Synergy-H1 Hybrid microplate reader. The plates were shaken before and after each reading.

In vitro survival assays

Long-term survival assays based on DNA content after 7-10 days of growth were conducted in 48 well dishes as previously described (9, 12). Cells were treated with vary doses of ARQ-761 (aka β-Lapachone) in the presence or absence of the NQO1 inhibitors dicoumarol or 5 μM Mac 220.

Cell Proliferation Assays

To determine cell proliferation rates we used the CyQUANT cell proliferation assay kit (Life Technologies) and followed the manufacturers protocol. Briefly, standard curves were generated by pelleting 1X10⁶ cells for all cell lines. Pellets were resuspended in 1 mL of CyQUANT GR/cell-lysis buffer and vortexed briefly. A dilution series in one row of a 96 well microplate ranging from 50 to 50,000 cells per cell line in total volumes of 200 μL were plated along with a 200 μL control well with no cells and incubated for 5 minutes in the dark at room temperature. Using a Synergy-H1 Hybrid Reader (Bio Tek), fluorescence was measured at excitation 480nm and at 520nm emission. For proliferation, cells were plated out in 6 wells at 5,000 cells per well in a total of 200 μL of a 96 well plate. Multiple plates were seeded using the same starting concentration and cultured at 37°C and 10% CO₂ until desired time to collect the plates. Plates were collected at 0, 24, 48, and 72 hours by inverting the plates and blotting on a paper towel to remove medium from the wells. Plates were stored at −80°C until all plates were collected. Plates were thawed at room temperature and 200 μL of CyQUANT GR/cell-lysis buffer was added. Plates were incubated for 5 minutes in the dark at room temperature.

Anchorage-independent growth assays

For A549 cell lines, a 1.5% SeaPlaque Agarose (SPA) mixture was made by slowly adding SPA to PBS and autoclaving. 0.5% SPA was created by diluting the 1.5% stock SPA 1:3 with culture media. 1 mL of the 0.5% SPA mixture was added to each well of a 6 well plate to create a bottom layer and allowed to solidify at room temperature for 15-20 minutes. Cells were counted and suspended at 750 cells/mL in a separate 0.5% SPA mixture. 2 mL were added to each well on top of the bottom layer and allowed to solidify for 30-45 minutes at room temperature to create a cell layer. A 0.3% SPA mixture was created by diluting the 1.5 % stock SPA 1:5 with culture media. 1 mL of the 0.3% SPA mixture was added to each cell layer and allowed to solidify for 20-30 minutes at room temperature to create a top layer. 250-500 μL of culture media was added onto the top layer to prevent from drying out. A similar method was used where SeaKem Agarose was substituted for SeaPlaque Agarose yielding similar results. Plates were wrapped in parafilm and placed at 37°C. 250-500 μL of new culture media was added every week. Plates were imaged after 3 weeks (A549 cells) and 6 weeks (H292 cells) using and Epson V700 photo scanner. The enumeration of colonies present in each dish was quantified using imageJ software.

Cell Death Elisa (CDE, anoikis) assays

A cell death detection ELISA kit was used to determine the level of detachment induced cell death (anoikis). Briefly, cells were seeded at a density of 250,000 cells in a 10cm dish for 48 hours prior to plating for assay. 2.0 mL of 0.5% methylcellulose and culture medium mixture was added to poly-HEMA coated plates and allowed to equilibrate in the incubator for 1 hour at 37°C. 150,000 cells were seeded per well in the 0.5% methylcellulose mixture and incubated at 37°C for the 24 and 48 hour time points. For the zero hour time point cells were placed directly into Eppendorf tubes and placed on ice. Cells were lysed with 100 μL of CDE lysis buffer at 4°C for 20 minutes. Cells were pelleted for 12 minutes at 4°C at 13,000 rpm. 75 μL of the supernatant was transferred to a new tube and stored at −80°C until all time points were collected. All samples were processed using the cell death detection ELISA kit with the manufactures protocol. Absorbance values were recorded using the Synergy-H1 Hybrid reader (Bio Tek) at 405 nm. Values were calculated by subtracting the zero hour time point from the 24 and 48 hours time points.

Invasion assays

Invasion assays were performed as described previously (23). Briefly a 1.5% noble agar/PBS mixture was made and then autoclaved. Using a multi-channel pipette, 100 μL of the noble agar mixture was added to each well of a 96 well plate. Cells were counted and suspended at a density of 50,000 cells/mL. 200 μL of this suspension was added to the 96 well plates once the noble agar was solidified. Plates were allowed to sit for 1-3 days depending on cell line until spheroids were formed. Once spheroids were formed, a 400 μL base layer mixture of 10X DPBS, 1N sodium hydroxide, sterile water and rat tail collagen was added to a 24 well plate and allowed to solidify at 37°C for 30 minutes. Using a nucleofector pipette, spheroids were added one at a time to Eppendorf tubes. 3 spheroids were added to each tube. Spheroids were centrifuged at 1,000 rpm for 3 minutes. The media was removed from each tube using a nucleofector pipette. A collagen cell layer mixture was made and 500 μL were carefully added to each tube, one at a time, and spheroids were added to the 24 well plates. The cell layer was incubated at 37°C for 1 hour. 1 mL of culture media was added to the top of the cell layer to prevent the collagen from drying out. Once media was added, zero hour images were taken at 5X (A549 and HCC1171 cells) or 10X (H292) magnification using an inverted microscope (Axiovert 200M, Zeiss). Images were taken every 24 hours until invasion distance was outside of the focal field of view. The scale bar in each image represents 100 microns. Using the AxioVision software, the invasion distance was calculated by subtracting the initial spheroid radius from the invasive distance at the final time point.

DCFDA staining for ROS studies

500,000 cells were plated in 10 cm dishes containing 7 mL of 0.5% Methylcellulose mixture. Cells were incubated for 24 hours at 37°C and 10% CO_2. After 24 hours pellets were collect by centrifuging at 1,000 rpm for 2 minutes. Pellets were washed 1X with PBS and centrifuged at 1,000 rpm for 2 minutes. Pellets were resuspended in DMEM/F-12 without phenol red, which contained either 5μM DCFDA (stained) or DMSO (unstained control), and transferred to flow cytometry tubes. Cells were incubated at 37°C and 10% CO₂ for 30 minutes. Cells were then pelleted again at 1,000 rpm for 2 minutes. Pellets were resuspended in 500 μL of DMEM/F-12 without phenol red and placed at 37°C and 10% CO₂ for 15 minutes to equilibrate and then samples were read using the LSR Fortessa (BD Biosciences).

In vivo tumorigenesis and survival assays

For tumorigenesis assays 4-6 week old female athymic nude mice (Charles River) were subcutaneously injected on the right flank with A549 shCtr-R and A549 shNQO1 cells on the left flank. Mice were stratified into 3 groups containing initial tumor injection concentrations of 5 million, 2.5 million or 1 million cells of each cell type. 10 mice were used for each group. Tumor growth rates were monitored by caliper measurements using the formula (L × W²/2). Statistical significance between A549-shNQO1 and A549-shCtr-R tumor growth rates, at each concentration, was calculated using an unpaired, two tailed Students t-test.

In separate studies, to compare the effect that NQO1 depletion had on overall survival, 4-6 week old athymic female nude mice (Charles River) were subcutaneously injected with A549-shCtr-R or A549-shNQO1 cells into their flanks. Tumor growth rates were monitored by caliper measurements using the formula (L × W²/2). Tumor growth was assessed until the tumors reached the set volume of 1000 mm³. Post mortem, tumors were collected for evaluation of NQO1 expression. Log-rank test were applied to survival analyses (Kaplan-Meir). All statistical significance assessments were conducted using Graph-Pad Prism 6 software.

All animal studies were performed in accordance with the animal care policies of West Virginia University and were approved by the West Virginia Animal Care and Use Committee.

Aldefluor activity assays

The percentage of cells that were ALDH^(high) within the various cell populations assayed was determined using the Aldefluor Kit (Stemcell Technologies). The kit was used according to the manufactures protocol. Briefly, two flow cytometry tubes, per cell line, were labeled as control or test. Cells were trypsinized and pelleted at 1,000 rpm for 1-2 minutes. 1 X 10⁶ cells were then counted out for both shCtr-R and shNQO1 cell lines. Pellets were washed 1X with PBS and resuspended in 1mL of the Aldefluor Assay Buffer provided and transferred to the “test” tube. 5μL of the DEAB (ALDH inhibitor) was added to the “control” tube and was immediately recapped. 5μL of the Aldefluor reagent was added to the “test” tube and was vortexed immediately. After the “test” tube had been vortexed, 500μL of the Aldefluor Assay Buffer was transferred from the “test” tube to the “control” tube and the “control” tube was immediately vortexed. That procedure was repeated for each cell line. Once all cells line were stained, the “test” and “control” tubes were incubated at 37°C for 30 minutes. After the incubation period, tubes were centrifuged at 250 rpm for 5 minutes. Supernatant was removed and pellets were resuspended in 500μL of Aldefluor Assay Buffer and samples were placed on ice. A549 samples were assessed using a Fortessa flow cytometer and Mia PaCa and PC3 cell lines were analyzed using a Facs Calibur flow cytometer (Becton Dickinson). Final data analysis was performed using FCS Express software.

Cell viability assays

HBECs were transiently transfected with siRNA-NQO1 or scramble control according to the protocol described above. After the 48 h transfection period, cells were enumerated and seeded at a density of 20,000 cells/ well in 96 well plates (white). The following day, the viability of cells in each group (8 wells/ group was assessed by adding 100 μL of CellTiter-Glo (promega) to each well. Luminescence was detected using a Synergy-H1 Hybrid reader (Bio Tek).

Statistical analysis

Statistical differences were determined by using Student's t tests, and p values were reported. All statistical analyses were performed using Graph-Pad Prism 6 software, and considered significant when p values were < 0.05.

Results

Elevated NQO1 expression predicts poor survival in NSCLC patients

In previous studies our laboratory, as well as other investigators, reported that NQO1 expression levels were highly elevated in lung cancer patient tumor versus associated normal lung tissue (9, 24). Elevated tumor-NQO1 levels have provided a distinct advantage for developing NQO1-directed anticancer therapeutics such as β-lapachone and deoxyniboquinone (13, 14). However, in more recent retrospective investigations of patient outcomes, a strong correlation between elevated tumor-NQO1 expression levels and poor patient survival in various cancer types including breast and ovarian cancers has emerged (25, 26). Thus, we sought to determine whether elevated NQO1 expression in lung tumors also confers a survival disadvantage in lung cancer patients. We analyzed gene expression and survival data from NSCLC (lung adeno- and squamous cell carcinoma) patients within The Cancer Genome Atlas (TCGA) (20-22). Patients were stratified into high and low NQO1 expressers based on a quartile bound cutoff. With this cutoff, a total of 191 patients were identified as high expressers and 244 patients were identified as low expressers. Our data, in three separate analyses, show that lung cancer patients with high tumor-NQO1 expression levels have worse overall survival (Figure 1A and Supplemental Figure 1). Our data are consistent with the aforementioned reports of poor overall survival in breast and ovarian cancer patients whose tumors had high NQO1 expression levels (25, 26). These data, as well as our laboratory observations that NQO1 levels increase during the process of transformation, suggested that cancer cells increase NQO1 expression levels as part of a pro-survival strategy during tumorigenesis, and that depleting NQO1 levels could possibly eliminate this survival advantage.

Figure 1. Elevated NQO1 levels in NSCLC patient tumors decreases their overall survival.

In A, Kaplan-Meier analysis of patient survival based on tumor NQO1 expression levels from a TCGA data set (21). Patients were grouped into NQO1 low and NQO1 high expression groups as described in “Materials and Methods”. In B, Western blot analysis of A549 and H292 cells stably transduced with retroviral (shNQO1) or lentiviral (shNQO1-B) NQO1 constructs. In C and D, A549 and H292 NQO1 knockdown cell models from (B) were assayed for NQO1 enzyme activity, and activity was expressed as nMoles/min/μg of protein. In C, p values for A549 shCtr-R vs A549 shNQO1 (p <0.0001) and for A549 shCtr-L vs A549 shNQO1-B. (p = 0.0002) In D, p values for H292 shCtr-L vs shNQO1-B (p = 0.0114).

Depleting tumor-NQO1 levels inhibits anchorage-independent growth and invasion of NSCLC

To investigate whether depleting NQO1 would alter the growth of lung cancer cells we used NQO1 shRNAs to establish stable knockdown of NQO1 in A549 and H292 NSCLC cell lines. Our data show that shNQO1 knockdown in A549 cell lines using a retroviral vector (shNQO1) or lentiviral vector (shNQO1-B) caused a significant decrease in NQO1 protein expression levels (Figure 1B and Supplemental Figure 2A), which correlated with loss of NQO1 activity (Figure 1C). Similar results are shown for NQO1 knockdown in H292 cells (Figures 1B lower panel and 1D). The parental A549 cells have nearly 12 fold higher levels of NQO1 activity as compared to the levels of NQO1 activity detected in H292 cells. Thus, these two cell lines with their distinct differences in NQO1 activity levels, also serve as an internal comparison to determine whether patients with lower NQO1 levels in their tumors could also benefit from therapeutic strategies aimed at depleting NQO1 expression.

A hallmark of oncogenic transformation is the newly acquired ability of a transformed cell to grow in an anchorage-independent environment. This acquired phenotype also increases the invasive and metastatic potential of transformed cells. In previous reports A549 and H292 cell lines served as metastatic models for in vivo studies (27, 28). Thus, we hypothesized that stable shRNA knockdown of NQO1 in A549 and H292 cells would be sufficient to determine whether NQO1 depletion affected tumor growth in anchorage-independent colony forming assays (also referred to as soft agar assays). In Figures 2A-D and Supplemental Figure 3 our data show that stable depletion of NQO1 significantly inhibits the growth of A549 and H292 cells in soft agar assays. Interestingly, the inhibition of NQO1 expression in A549 cells using the shNQO1 vector was substantially greater than what was observed with the shNQO1-B vector. Thus, the higher NQO1 activity observed in shNQO1-B A549 knockdown cells (Figure 1 C) correlated with more colony growth (Figure 2C and Supplemental Figure 2B). These data suggest that the degree of NQO1 activity loss affects the ability of cells to grow in soft agar. In addition to NQO1 shRNA knockdown studies we examined the effect of dicoumarol and Mac 220, NQO1 inhibitors that mimic the co-factor NAD(P)H which is required for NQO1 activity (29), on colony growth. Our data show that the NQO1 inhibitors dicoumarol and Mac220 also significantly inhibited the growth of A549 cells in soft agar colony forming assays (Supplemental Figures 4 and 5), further confirming that anchorage-independent growth in these cells is affected by the loss of NQO1 activity.

Figure 2. Depleting NQO1 expression levels inhibits growth of NSCLC cells in soft agar.

In A, A549 shCtr-R and A549 shNQO1 cells were analyzed for their ability to form colonies and grow in soft agar. Photomicrographs shown are representative of experiments performed in sextuplet. In B- D, graphical representation of enumerated colonies for A549 shNQO1, A549 shNQO1-B and H292 shNQO1-B cells versus A549 shCtr-R, A549 shCtr-L and H292 shCtr-L cells. In B, p values for A549 shNQO1 vs A549 shCtr-R (p (<0.0001). In C, p values for A549 shNQO1-B vs A549 shCtr-L (p = 0.0041). In D, p values for H292 shNQO1-B vs H292 shCtr-L (p=0.0114).

To further confirm the role of NQO1 in anchorage-independent growth we examined the effect of NQO1 overexpression in H596 cells, which are NQO1 null due to the *2 polymorphism. Our data show that NQO1 overexpression in H596 cells (H596-LPC-NQO1) caused significantly more colony growth as compared to vector only H596-LPCX cells (Supplemental Figure 6). In contrast, our data show that transient knock down of NQO1 in H596-LPC-NQO1 cells results in significant loss in their ability to grow in soft agar (Supplemental Figure 7). These data further indicate that NQO1 plays a role in the survival of cells in an anchorage-independent environment.

The ability to survive in an anchorage-independent environment is uniquely tethered to a cancer cells ability to invade and metastasize (30). Thus, we employed an in vitro 3-dimensional spheroid invasion assay (23, 31) to address the role that NQO1 plays in the process of tumor cell invasion. Our data show that stable knock down of NQO1 in A549 cells decreased the overall area of lung tumor spheroids (Supplemental Figure 8A) and inhibited the invasive progression of lung tumor spheroids in A549 and H292 cells (Figures 3 A-D and Supplemental Figure 8B). In addition to A549 and H292 cells, our data show that transient depletion of NQO1 expression using siRNA also inhibited invasion in HCC1171 lung cancer cells (Supplemental Figure 9). In contrast to NQO1 knockdown studies, when NQO1 was overexpressed in H596 cells a significant increase in invasion was observed (Supplemental Figure 10). Together these data demonstrate that NQO1 levels are critical for anchorage-independent growth and the invasion of lung cancer cells.

Figure 3. Loss of NQO1 expression inhibits invasion of NSCLC.

In A, and C, 3D- tumor-spheroid invasion assays were performed on A549 shNQO1, H292 shNQO1-B and A549 and H292 shCtr-L cell lines as described in “Materials and Methods”. Shown are photomicrographs of representative spheroids from each cell line at 0 and 72 (h). In B and D, graphical presentation of the tumor-spheroid invasion distance migrated by A549 shNQO1 and H292 shNQO1-B cells as compared to A549 and H292 shCtr-L cell lines. In B, p values for A549 shNQO1 vs A549 shCtr-R (p<0.0001). In D, p values for H292 shNQO1 vs H292 shCtr-L (p<0.0001)

NQO1 depletion elevates ROS levels and sensitizes cells to anoikis

Previous investigations have shown that NQO1 can act as a scavenger of ROS (32), thus we hypothesized that depleting NQO1 in our lung cancer models would increase endogenous levels of ROS. As expected, our data show that depletion of NQO1 in A549 cells caused an increase in oxidative stress as indicated by the increased DCFDA staining (a general ROS indicator including H₂O₂ levels (Figures 4 A-B), supporting our hypothesis that endogenous ROS levels are increased in lung cancer cells when tumor-NQO1 levels are depleted.

Figure 4. Depleting tumor-NQO1 levels increases ROS formation and sensitizes NSCLC to anoikis.

In A and B, A549 shCtr-R, A549 shNQO1, H292 shCtr-L and H292 shNQO1-B cell lines were stained with 5 μM DCFDA (Life technologies) to detect endogenous ROS levels (H₂O₂). In C and D cell death ELISA assays (Roche Applied Sciences) were performed on A549 and H292 NQO1 knockdown and control cells as described in “Methods” to detect cells that had undergone detachment induced cell death (anoikis). In A, p values for shCtr-R vs A549 shNQO1 (p = 0.004). In B, p values for shCtr-L vs H292 shNQO1-B (p = 0.0012). In C, p values for A549 shCtr-R vs A549 shNQO1 (p < 0.0001). In D, p values for H292 shCtr-L vs H292 shNQO1 (p = 0.0001).

In transformed cells the intracellular production of ROS is tightly regulated to prevent programmed cell death (33). Excessive ROS production can lead to apoptotic catastrophe, and cells that escape apoptosis are resistant to detachment induced cell death, also known as anoikis (34). Anoikis resistant cells are capable of continued proliferation and distant tumor formation (35). Thus far, our data show that depleting NQO1 prevents anchorage-independent growth and increases ROS stress levels. Therefore, we further hypothesized that the inability of NQO1 depleted cells to grow in an anchorage-independent environment was linked to anoikis sensitization caused by increased levels of ROS. To test this hypothesis we performed cell death ELISA assays on NQO1 knockdown cell models. Our cell death assays show that loss of NQO1 in A549 and H292 cells significantly increased sensitization to anoikis (Figures 4C-D). These data suggest that depletion of NQO1 expression in lung cancer cells increases oxidative stress and potentiates detachment induced cell death.

Depleting tumor-NQO1 expression levels decreases cell proliferation and in vivo tumor growth

Uncontrolled proliferation is a hallmark of malignant neoplastic cells, thus novel approaches to reduce uncontrolled cell proliferation are of paramount importance in the development of anticancer strategies. Our current data show that loss of NQO1 decreases tumor growth in soft agar and increases sensitization to anoikis. Thus, we hypothesized that depletion of tumor-NQO1 levels would significantly decrease the ability of cells to proliferate. Our data show that A549 and H292 NQO1 knockdown cells had significantly lower rates of proliferation as compared to their respective controls (Figures 5 A-B). These data suggest that depletion of NQO1 inhibits cell growth by inducing apoptosis caused by detachment induced cell death.

Figure 5. Loss of NQO1 expression inhibits cell proliferation and in vivo tumor growth.

In A and B, A549 and H292 cells were assayed for proliferation rates at 0, 24, 48 and 72 h using the CyQuant cell proliferation kit (Life Technologies). In C, A549 shNQO1 cells (open symbols) and A549 shCtr-R (closed symbols) were subcutaneously injected into flanks of athymic mice at varying concentrations ((1.0 black), (2.5 blue) or (5.0 red) × 10⁶) cells. Tumor growth was assessed bi/weekly using calipers. In D, A549 shNQO1 and A549 shCtr-R cells were injected subcutaneously into flanks of athymic mice and tumors were measured bi weekly by caliper measurements until a volume of 1000 mm³ was reached. Kaplan-Meir survival analysis was conducted using GraphPad Prism 6 software. In E, representative photomicrograph of mice in C where mice were injected on the Left flank (L) with A549 shNQO1 cells or on the right flank (R) with A549 shCtr-R cells at the indication concentration of cells. Shown are mice whose tumors were photographed after 32 days. In F, Western Blot for tumor-NQO1 expression and PARP-1 cleavage. Samples were harvested in PARP-lysis buffer as described in “Materials and Methods”. In A, p values for A549 shNQO1 vs A549 shCtr-R cells at 24 h (p = 0.0031), 48 h (p < 0.0001) and 72 h (p = 0.0004). In B, p values for H292 shNQO1-B vs H292 shCtr-L cells at 24 h (p = 0.0070), 48 h (p = 0.0011) and 72 h (p < 0.0001). In C, p values for A549 shNQO1 vs A549 shCtr-R cells at 5 × 10⁶ cells (p <0 .0006), 2.5 × 10⁶ cells (p < 0.0006) and 1.0 × 10⁶ cells (p<0.006). In D, p values for A549 shCtr-R vs A549 shNQO1 (p = 0.0083)

To address the role that NQO1 depletion plays in vivo tumor growth, varying concentrations (1, 2.5 and 5 million) of A549 shNQO1 and A549 shCtr-R cells were implanted subcutaneously into athymic mice and tumor growth and overall survival rates were evaluated. Our in vivo xenograft data clearly show that a significant growth disadvantage is observed in A549-shNQO1 cells at each concentration of cells implanted as compared to A549-shCtr-R cells (Figures 5 C and E). In addition to significantly reducing in vivo tumor growth rates, the depletion of tumor-NQO1 expression levels in animals bearing A549-xenografts increased their overall survival as compared to animals bearing A549-shCtr-R xenografts (Figure 5 D). We also observed that tumor-NQO1 levels remained depleted in A549-shNQO1 xenografts as illustrated by our in vivo western-blot analysis for NQO1 protein expression (Figure 5 F). Interestingly, a substantial difference in PARP-1 proteolysis was observed in A549-shNQO1 tumors as compared to A549-shCtr-R tumors, further supporting our anoikis data that suggest that loss of NQO1 leads to increased apoptosis (Figure 5 F and Supplemental Figure 11).

NQO1 depletion reduces the percentage ALDH^hgh cells in the tumor cell population

We have shown that knockdown of NQO1 expression in lung cancer cells decreased clonogenic growth in vitro and tumor growth in vivo. Numerous studies have reported that cancer stem cell populations are responsible for increased tumorigenicity and resistance to therapeutics. Thus, we sought to determine if NQO1 affected this critical population of cells. Previous work has shown that one of the most reliable cancer stem cell markers is aldehyde dehydrogenase (ALDH) (36, 37). Although several isoforms of ALDH exists, a common assay used to define the ALDH^(high) stem cell population is the Aldeflour assay (36). In our studies we tested the hypothesis that NQO1 depletion caused less tumor growth due to depletion of ALDH^(high) cells. Our data clearly show that there is a significant decrease in the ALDH^(high) population in A549 shNQO1 cells as compared to A549 shCtr-R cells (Figure 6). Interestingly, this phenomenon was also discovered to be true in MiaPaCa (Pancreas) and PC3 (Prostate) cancer cells (Supplemental Figures 12 and 13).

Figure 6. NQO1 depletion causes a decrease in ALDH^high activity.

In A, representative flow cytometry tracing of A549 shNQO1 and A549 shCtr-R cells analyzed for ALDH^(high) activity using the Aldefluor Assay Kit from “Stem Cell Technologies”. Cells were assayed according to the manufacturers protocol as described in “Materials and Methods”. DEAB was used as an inhibitor of ALDH^(high) activity. The percentages shown in each tracing indicate the population of cells staining for ALDH^(high) activity. In B, graphical presentation of A549 shNQO1 and A549 shCtr-R cells assayed for ALDH^(high) activity. The graph represents experiments repeated at least 5 times in duplicate.

Discussion

Normal cells are under continuous bombardment from intracellular and extracellular oxidative stress in the form of ROS (38). Damage caused by uncontrolled oxidative stress from lethal levels of ROS can lead to DNA strand breaks, mutation events and even cell death (39). Thus, mechanisms that facilitate control over ROS levels are uniquely important to cell proliferation and survival. Importantly, normal levels of ROS are needed in various cell-signaling events involving cell proliferation as well as programmed cell death. It is also of note that specific ROS are critical to many disease processes such as aging and cancer (38, 39). Thus, various defense mechanisms have evolved to regulate exposure to endogenous and exogenous ROS. These mechanisms include the transcription factor Nrf2 that transcriptionally activates the expression of numerous down stream target genes that modify and regulate the duration and exposure level to ROS (40). The genes activated by Nrf2 include glutathione peroxidase, catalase and NQO1. The down stream targets of Nrf2, such as NQO1, regulate exposure to ROS from both exogenous and endogenous sources and play a critical regulatory role in cell survival and cell death.

Cancer cells, just as in normal cells, must regulate ROS levels and have adapted to exposure to high levels of ROS through the altered expression of specific ROS regulatory genes that aid in their survival (41). Catalase for example, is normally expressed at high levels in normal tissues, however in tumors its levels are relatively low (42). The down-regulation of catalase expression in tumors is not clearly understood. However, catalase suppression has been associated with specific tumor promoting signaling pathways and resistance to chemotherapeutics (8, 43). Interestingly, studies with breast cancer and lung cancer cells have shown that re-expression of catalase modifies their exposure to ROS levels from their tumor microenvironment and ultimately enhances tumor cell death (44, 45). This would imply that reversing the expression of specific ROS regulatory genes in cancer cells could potentiate a tumor specific cell death.

Previous studies have implicated NQO1 as a prognostic marker that negatively affects patient survival (46, 47). In most of these studies, the poor patient outcome is attributed to the existence of 2 prominent NQO1 polymorphisms, referred to as *2 and *3. The *2 mutation is more common and involves a C to T point mutation at nucleotide position 609. These polymorphisms exist at varying, but small percentages, within the population (10). Those patients whose tumors were identified to have the homozygous *2 mutation in NQO1, were found to be more susceptible to issues involving chemotherapeutic toxicity when exposed to NQO1-detoxified therapeutics such as epirubicin (47). In contrast, recent retrospective analyses have shown that elevated NQO1 expression in patient tumor versus normal tissue predicts poor patient survival (25, 26). We hypothesized, from the latter studies, that tumors elevate NQO1 to enhance survival and that reduction of NQO1 could potentially ameliorate the negative effects of tumor-NQO1 overexpression on patient outcome.

In the current study we focused on determining whether decreasing elevated tumor-NQO1 levels in lung cancer cells would inhibit tumor survival. We chose specific readouts, such as anchorage-independent growth, anoikis and in vivo tumorigenesis assays, to make a connection between tumorigenic processes and the role that NQO1 played in each. Our data clearly show that loss of NQO1, by stable shRNA knockdown, significantly affected anchorage-independent growth of lung cancer cells, which is hallmark of tumorigenesis. These data suggested that NQO1 overexpression is intimately involved in the survival and proliferative capacity of lung cancer cells that overexpress NQO1. These data were corroborated by dicoumarol and Mac220 studies that showed that treatment with NQO1 inhibitors significantly decreased the growth of lung cancer cells in soft agar. In addition to inhibiting growth in soft agar, we showed that loss of NQO1 potentiated anoikis, suggesting that cells that were NQO1 depleted were more susceptible to detachment induced cell death. This was further supported by the increase in ROS that was found in shNQO1 cells versus our control cells which correlates with increased anoikis. We also showed that loss of NQO1 decreased both cell proliferation and invasion suggesting that knocking down NQO1 decreases the tumorigenic potential of lung cancer cells. In contrast to cancer cells, our in vitro data show that transient knockdown of NQO1 in non-transformed, non-tumorigenic human bronchial epithelial cells (HBECs) did not reduce their short-term viability or long-term survival (Supplemental Figure 14).

Our in vivo studies confirmed that stable NQO1 depletion increased long-term survival in mice since shNQO1 tumors were significantly smaller than control tumors, and survival of mice bearing shNQO1 tumors was significantly enhanced as compared to mice bearing control tumors. Finally, we show that loss of NQO1 substantially reduced the ALDH^(high) population in lung, pancreas and prostate cancers. In previous reports it has been demonstrated that ALDH^(high) activity within a tumor population is a reliable marker for cells that have a cancer stem cell phenotype in a number of malignancies (36, 37). Interestingly, knockdown of specific ALDH isoforms has been linked to the loss of stemness and tumorigenicity in lung cancer (48). Our studies show that loss of NQO1 reduces the population of cells with ALDH^(high) activity, suggesting that the loss in tumorigenicity seen in NQO1 knockdown cells (Figure 5 C-E) is attributable to the loss of the ALDH ^(high) subpopulation of cells (Figure 6).

In summary, we report for that NQO1, a gene found overexpressed in many solid tumors, including NSCLC, can be directly targeted for therapy since it plays a critical role in the overall growth, invasive potential and survival of lung cancer. We hypothesize that NQO1 expression is increased in tumors to thwart ROS stress, and that reversing the elevated expression of tumor-NQO1 leads to increased susceptibility to ROS and reduced tumor burden due to anoikis and the loss of the ALDH^(high) cell population (Figure 7). These results suggest that NQO1 depletion may be an important link in eliminating cancer stem cell populations not only in lung cancer, but in other malignancies as well. Finally, these data establish the potential for a new clinical approach that targets NQO1 in lung cancer patients whose tumor-NQO1 expression levels are often found to be 5-20 times more elevated than the NQO1 levels in their adjacent normal lung tissue (9).

Figure 7. Model depicting therapeutic approach where reduction of NQO1 levels in tumors leads to decreased tumor burden in the lungs of cancer patients.

In A, NQO1 (blue squares) is expressed at normal levels in lung in response to oxidative stress. In B, NQO1 (blue squares) is overexpressed in tumor cells (black circles) within the lung due to increased necessity to inhibit ROS stress. In C, targeting NQO1 in lung tumors leads to a decrease in tumor burden in patients with elevated NQO1 levels in their lung tumors. In B, cancer stem cells within a tumor will have increased ALDH^(high) activity. In C, NQO1 knockdown reduces the population of tumor cells with ALDH^(high) activity.

Implications.

Loss of tumor-NQO1 expression inhibits tumor growth and suggests that novel therapeutics directed at tumor-NQO1 may have clinical benefit.