Evidence for tankyrases as antineoplastic targets in lung cancer

We previously described clinically-relevant cyclin E-transgenic mouse lines that develop pulmonary pre-malignant lesions and lung adenocarcinomas [31]. For microarray analyses, adenocarcinomas and adjacent histopathologically normal lung tissues were each harvested from age- and sex-matched mice and immediately placed in RNAlater (Qiagen, Valencia CA). These specimens were isolated from human surfactant protein C (SP-C)-driven wild-type human cyclin E-transgenic mice (as previously described [31]). Normal non-transgenic lung tissue was harvested from age-and sex-matched FVB mice (NCI Frederick National Laboratory, Frederick MD). For qPCR analyses, malignant and adjacent normal lung tissues were isolated from additional transgenic mice of both wild-type and proteasome-degradation resistant human cyclin E-transgenic lines and snap frozen in liquid nitrogen.

Total RNA was isolated with TRIzol RNA isolation reagent (Life Technologies, Carlsbad CA). GeneChip Mouse Genome 430 2.0 Arrays were purchased (Affymetrix, Santa Clara CA), with 11-probe sets covering 39,000 transcripts within the mouse genome. Hybridizations were performed according to Affymetrix guidelines at the Dartmouth College Microarray Shared Resource using an Affymetrix GeneChip Workstation. Biotin-labelled cRNA was generated from 5 μg of total RNA and hybridized to the Mouse Genome 430 2.0 chip. A total of 12 hybridizations were performed comprising 12 independent biologic samples organized into three groups of four. Raw data from each hybridization was normalized by Robust Multichip Average (RMA), background corrected, and filtered for presentation using GeneSifter software (Geospiza Inc., Seattle WA). The remaining probe sets were then analyzed by GeneSifter software for species involved in the Wnt pathway. Raw and RMA data is available from the NCBI Gene Expression Omnibus (GEO) with accession number GSE45744.

A tissue bank accrued from consecutive cases over 8 years at Dartmouth-Hitchcock Medical Center containing paired human normal and malignant lung tissues was described [32]. Dartmouth’s Institutional Review Board (IRB) reviewed and approved the acquisition and analyses of these tissues.

Total mRNA was isolated using the RNeasy kit with on-column DNAse digestion (Qiagen). RT was performed with the High Capacity cDNA RT Kit (Applied Biosystems, Foster City CA) and a Peltier Thermal Cycler (MJ Research, Waltham MA). The qPCR assays were performed using iTaq Fast SYBR Green Supermix with ROX (Bio-Rad Laboratories, Hercules CA) and the 7500 Fast Real-Time PCR System (Applied Biosystems). All assays were performed in triplicate. Primers sequences are presented in Additional file 1: Figure S1.

Murine lung cancer cell lines studied included ED1, ED2, and ED1L (derived from a single-cell subclone of ED1), which were each previously described [32, 33]. The C-10 immortalized murine bronchial epithelial cell line, BEAS-2B immortalized human bronchial epithelial cell line, NCI-H522, Hop62, and A549 human lung cancer cell lines, and the 293T human embryonic kidney cell line were each purchased (ATCC, Manassas VA). All cell lines except BEAS-2B and 293T cells were cultured in RPMI 1640 medium (Corning, Manassas VA) supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific, Waltham MA) at 37°C in a 5% CO₂ humidity-controlled incubator. BEAS-2B cells were cultured in serum-free LHC-8 medium (Life Technologies) supplemented with 0.1% epinephrine. The 293T cell line was cultured in high glucose DMEM (Life Technologies) supplemented with 10% FBS and 4 mM L-glutamine (Life Technologies).

TNKS inhibitors XAV939 [20], IWR-1 endo, and IWR-1 exo [21] were purchased (Cayman Chemical, Ann Arbor MI) and dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich, St. Louis MO). Recombinant murine Wnt3a ligand was purchased (R&D Systems, Minneapolis MN) and dissolved in 1% bovine serum albumin (BSA; Sigma-Aldrich) in phosphate buffered saline (PBS; Corning).

For cell proliferation assays, ED1 (2 × 10³), ED1L (2 × 10³), ED2 (5 × 10³), C-10 (5 × 10³), BEAS-2B (5 × 10³), H522 (5 × 10³), and A549 (5 × 10³) were individually plated in growth medium in triplicate in individual wells of 12-well tissue culture plates (Corning) 24 hours before drug or vehicle treatments. Cell viability was measured 72 hours following these treatments using the CellTiter-Glo (Promega, Madison WI) luminescent cell viability kit and a TD-20/20 Luminometer (Turner Designs, Sunnyvale CA).

For clonal growth assays, ED1 cells were plated at a density of 200 cells per well in 6-well tissue culture plates (Corning) in triplicate 24 hours before drug or vehicle treatments. Colonies were stained after 7 days with DiffQuick (IMEB Inc, San Marcos CA) and counted using a Col Count instrument (Oxford Optronix, Oxford UK).

For washout studies, ED1 (3 × 10⁴) and A549 (7.5 × 10⁴) were independently plated in 10 cm tissue culture plates (Corning) in complete growth medium and individually treated 24 hours later with vehicle, XAV939, IWR-1 endo, or IWR-1 exo at 10 μM dose. Following 3 days of culture in drug, plates were trypsinized and replated at equal densities into 12-well plates in complete medium, as described for cell proliferation assays above. Cells were treated 6 hours later with vehicle, for control and washout wells, or the respective drug at 10 μM to maintain continuous treatment. Cell viability was assessed after 24 hours and 72 hours of treatment by CellTiter-Glo.

Cells were lysed in a modified radioimmune precipitation buffer, as before [34]. Protein concentrations were assayed using the BCA Protein Assay Reagent (Thermo Fisher Scientific). Twenty-five micrograms of protein were size-fractionated using 4-15% gradient sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) ReadyGels (Bio-Rad Laboratories) before electroblotting onto nictrocellulose membranes. Membranes were blocked with 5% nonfat milk in 0.1% Tween 20 (Sigma-Aldrich) tris-buffered saline (TBST), which was also the antibody diluent, except in the case of the activated β-catenin antibody (ABC), which was diluted in 1% milk TBST as in prior work [6, 35]. Antibodies and dilutions used are displayed in Additional file 1: Figure S1. Primary antibodies were detected with horseradish peroxidase-conjugated species-appropriate secondary antibodies (Santa Cruz Biotechnology, Santa Cruz CA and GE Healthcare Bio-Sciences Corp, Piscataway NJ) and visualized with the ECL Prime electrochemiluminescent detection reagent (GE Healthcare) and radiographic film.

For siRNA knockdown experiments, pairs of independent double-stranded siRNAs were purchased (Integrated DNA Technologies Inc., Coralville IA) that each targeted human or mouse TNKS1 or TNKS2. SiRNA sequences are presented in Additional file 1: Figure S1. A non-targeting scrambled siRNA was used as the control. ED1 (1 × 10⁴), ED2 (1.5 × 10⁴), A549 and (5 × 10⁴), Hop62 (5 × 10⁴) cells were independently plated in triplicate on 6-well tissue culture plates 24 hours before transfection. Transient transfections of each respective siRNA were accomplished with Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer’s protocol. Total RNA was collected and analyzed as described above to verify these knockdowns at 24 hours post-transfection, with cell growth assessed 72 hours post-transfection by CellTiter-Glo assay, as already described. Comparisons were made to cells transfected with the non-targeting scrambled control siRNA.

The 7TFP derivation of the pSuperTOPFlash vector and the EβP constitutively active β-catenin vector [36] were purchased (plasmids 24308 and 24313, respectively; Addgene, Cambridge MA). Two independent shRNA constructs targeting murine TNKS1 in a G418-selectable backbone, pLKO.1-CMV-Neo, and TNKS2 in a puromycin-selectable backbone, TRC2, were purchased, as well as scrambled controls in matched selectable backbones (Sigma-Aldrich). The sequences of these hairpin constructs are shown in Additional file 1: Figure S1.

Lentiviruses were generated with an optimized system [37] consisting of the transfer vector of interest and packaging plasmids pCMV-dR8.2 (plasmid 8455; Addgene) and pMD2.G (plasmid 12259; Addgene). These vectors were transfected into 293T cells using TransIT-LT1 transfection reagent (Mirus Bio, Madison WI) according to the manufacturer’s protocol. Lentiviral titers supplemented with 1% BSA were collected after 24 hours and used to infect murine ED1 lung cancer cells in the presence of 4ug/mL polybrene (Sigma-Aldrich). ED1 cells infected with 7TFP, EβP, or TNKS2 shRNA vectors were selected with 2.5 μg/mL puromycin (Life Technologies). ED1 cells infected with the TNKS1 shRNA vectors were selected with 1.5 mg/mL G418 sulfate (Corning). ED1 cells transduced with the combined controls or combined shRNAs received simultaneous drug selection at the above concentrations.

For luciferase assays, ED1 cells infected with the 7TFP vector (1 × 10⁴) were plated 24 hours before drug or vehicle treatments in triplicate in 12-well tissue culture plates. Treatments were with TNKS inhibitors or vehicle combined with a canonical Wnt activator or control (20 mM LiCl controlled by 20 mM NaCl or 25 ng/mL murine recombinant Wnt3a controlled by 1% BSA/PBS). Cells were lysed 16 hours after these treatments using the Reporter Lysis Buffer for the Luciferase Assay system (Promega). Luciferase activity was measured and normalized to protein concentrations.

The described animal protocols were reviewed and approved by Dartmouth’s Institutional Animal Care and Use Committee (IACUC). For both experiments, ED1 cells were infected with either both shRNA vector controls (dual shCTRL) or TNKS1 and TNKS2 combined knockdown (dual shTNKS) and selected, as described above. For xenograft studies, 1 × 10⁶ indicated cells were resuspended in 200μL of Growth Factor Reduced Matrigel (BD Biosciences, San Jose CA) and injected into the left flanks of 8 week old female NCr Nu/Nu athymic mice (NCI Frederick National Laboratory, Frederick MD). There were 10 mice in the dual control arm and 10 mice in the dual knockdown arm. Tumor diameters were measured twice weekly with vernier calipers by an investigator blinded as to the cell lines under analysis, and mice were sacrificed when mean tumor diameter reached 15 mm or when mice became moribund or cachexic, whichever arose first. Tumor volume was calculated as π/6 * Length * Width², where width was defined as the smaller of the cranial/caudal diameter or dorsal/ventral diameter [38].

For the syngeneic study, 1 × 10⁶ indicated cells were resuspended in 200μL PBS and injected into the tail veins of 8-week-old female FVB mice (NCI Frederick). There were 3 mice in the dual control arm and 5 mice in the dual shRNA arm. Mice were sacrificed 4 weeks post-injection and lung tissues were formalin-fixed, paraffin-embedded, and sectioned for histopathology, as before [31]. Hematoxylin and eosin staining was used and a pathologist blinded as to the treatment arms scored for lung tumor formation, as in prior work [39, 40].

Data shown represent at least three independent replicate experiments done in triplicate for the in vitro studies. Error bars indicate mean +/- standard deviation (SD), except in the case of the xenograft study, where error bars indicate mean +/- standard error of the mean (SEM). Statistical significance was determined by two-tailed t-test assuming unequal population variances in GraphPad InStat or Prism (GraphPad Software Inc, LaJolla CA) with significance set at P ≤ 0.05, except in the case of the syngeneic study, when a one-tailed t-test was used, and the xenograft study, where ANOVA was used to compare the growth curves and Kaplan-Meier analysis was used to compare survival to sacrificial endpoint. Multiple comparison in the proliferation studies was handled with ANOVA followed by Dunnett’s post-test. Microarray data were analyzed with the GeneSifter analysis suite using the embedded ANOVA function, with significance at P ≤ 0.05.

We previously reported a cyclin E-expressing murine transgenic lung cancer model [31] that recapitulated frequent features of human lung carcinogenesis [32, 39‐41]. We first asked whether this model deregulated the Wnt pathway, as occurs in human NSCLC [9, 16‐19]. Comprehensive microarray analyses compared non-transgenic murine lung, transgenic murine normal lung, and transgenic murine lung cancer. These analyses established Euclidean hierarchical clustering of each tissue by Wnt family member expression, with similar expression levels of these species detected in murine non-transgenic lung and transgenic normal lung tissue, but a different expression pattern in transgenic lung cancers (microarray data available at NCBI’s Gene Expression Omnibus, accession number GSE45744).

Five hundred probes on the array represent 258 unique genes that are functionally annotated in Gene Ontology (http://​www.​geneontology.​org) as members of the Wnt receptor signaling pathway. In our focused analysis of this pathway, 161 probes representing 117 unique genes (32% and 45% of the totals, respectively) were significantly (P ≤ 0.05 by ANOVA) up- or downregulated in comparison to the murine non-transgenic or adjacent transgenic normal lung controls (Figure 1A). Among the overexpressed mRNAs were: porcupine (PORCN), encoding an O-acyltransferase responsible for lipid-modifying Wnt ligands prior to secretion [42]; FZD2, encoding a Wnt/Ca²⁺ signaling-related Frizzled receptor [43]; and MYC. Repressed genes included those coding for Wnt antagonists such as WIF1 (discussed above) and PRICKLE1, a putative tumor suppressor that represses the Wnt pathway by destabilizing Dvl [44]. TNKS 1 and 2 were found to be moderately overexpressed, at 1.30-fold and 1.43-fold expression respectively, in comparison with transgenic normal adjacent lung.

The microarray results were independently validated by qPCR assays in murine lung cancers and their adjacent normal lung tissues. In a panel of 12 paired normal and malignant lung tissues derived from 7 wild-type cyclin E transgenes and 5 proteasome-degradation resistant cyclin E transgenes, 10 showed repression of WIF1 in the lung cancers versus adjacent normal lung tissues (Figure 1B). This result is consistent with prior published work from human NSCLC cases [18, 19]. Analysis of TNKS 1 and TNKS2 expression in the same panel showed overexpression of TNKS1 in 6 of 12 tumors and TNKS2 in 9 of 12 tumors, relative to adjacent normal lung tissues (Figure 1C). The qPCR analysis of WIF1 and TNKS expression in three cases of human lung adenocarcinoma revealed the expected repression of WIF1 and moderate overexpression of TNKS2 (as compared to adjacent normal lung tissues) in all examined cases, as well as for TNKS1 in 1 of 3 cases (Figure 1D).

Having established deregulation of Wnt pathway genes and TNKS overexpression in human and murine lung cancer, we next asked whether inhibition of the Wnt pathway with the TNKS inhibitors XAV939, IWR-1 endo, or IWR-1 exo would inhibit growth of murine or human lung cancer cell lines. Individual treatments with increasing doses of these three compounds (100nM to 50 μM) dose-dependently decreased proliferation of the ED1, ED1L, and ED2 murine lung cancer cell lines in comparison to vehicle controls after 3 days of treatment (Figure 2A with statistical analysis in Additional file 2: Figure S2). Independent treatments of the human lung cancer cell lines A549, Hop62, and H522 dose-dependently decreased proliferation of each of these lines versus vehicle controls after 3 days of culture (Figure 2B with statistical analysis in Additional file 2: Figure S2). The compounds also inhibited growth of the C10 murine immortalized bronchial epithelial cell line (Figure 2C, left panel). The BEAS-2B immortalized human bronchial epithelial cell line was found to be more resistant to growth inhibition by these agents, except for the IWR-1 endo compound (Figure 2C, right panel). Additionally, treatments of the ED1 cell line plated under colony-forming conditions with 10 μM of each TNKS inhibitor significantly inhibited 7-day colony formation relative to vehicle control, as shown with colony number quantification (Figure 2D).

Cells were growth inhibited by these compounds but did not show appreciable apoptosis (data not shown). In drug washout experiments, normal cell growth was restored after removal of the drugs, confirming that the effects of these compounds are reversible (Additional file 3: Figure S3).

We sought to validate whether the TNKS inhibitors exerted effects on Wnt signaling in human and murine lung cancer cell lines. Immunoblot analyses of ED1 and ED2 murine lung cancer cell lines (Figure 3A) and A549 and Hop62 human lung cancer cell lines (Figure 3B) following treatment with 10 μM of each inhibitor revealed stabilization of TNKS1, as expected due to inhibition of TNKS auto-PARsylation function [45]. Stabilization of TNKS2 was only apparent in ED1. We also observed accumulation of axin 1 at the protein level in all cell lines, a key component of the axin/APC/GSK3 β-catenin destruction complex and direct target of TNKS PARsylation. This relationship appeared to be dose-dependent, as evidenced in the ED1L cell line following 3 days of treatment with the inhibitors at increasing doses (Figure 3C).

Stable infection of ED1L (data not shown) or ED1 (Figure 3D) with a lentiviral vector containing a luciferase gene cassette under control of a 7× TCF binding site promoter [36] allowed for monitoring of the transcriptional activity of the Wnt pathway during TNKS inhibition. Wnt reporter activity was normalized to the total protein content of the cells to account for growth inhibitory effects. Basal (unstimulated) activity of the Wnt reporter was low in both cell lines, and was not significantly affected by tankyrases inhibition (data not shown). Co-treatment of either cell line with a Wnt activator (20 mM LiCl or 25 ng/mL recombinant murine Wnt3a) and a 10 μM dosage of each TNKS inhibitor led to a significant reduction in normalized Wnt-responsive luciferase activity versus vehicle controls at 16 hours post-stimulation (Figure 3D).

Because tool compounds such as XAV939 and IWR-1 likely have off-target effects, we sought to validate that the antineoplastic activity of these inhibitors was due to specific TNKS antagonism through independent genetic approaches. Transient TNKS knockdowns were accomplished with two siRNAs that independently target TNKS1 or TNKS2 in human or mouse cells. Transfection of these siRNAs alone or in combination led to significant (P ≤ 0.05) repression of TNKS1 or TNKS2 at the mRNA level in ED1 (Figure 4A, left panel) and ED2 (Figure 4B, left panel) murine cell lines at 24 hours, versus controls. Similar TNKS knockdown was achieved in the human lung cancer cell lines A549 (Figure 4C, left panel) and Hop62 (Figure 4D, left panel).

The consequences of TNKS knockdown on proliferation for each of the examined NSCLC cell lines was assessed after 3 days culture versus non-targeting control siRNA. Single knockdown of either TNKS1 or TNKS2 or dual TNKS knockdowns were growth inhibitory in the ED1, A549, and Hop62 lung cancer cell lines (Figures 4A, 4C, and 4D, respectively). In ED2 cells, only the combined knockdown achieved significant growth repression versus control (Figure 4B). The dual siRNA knockdown of TNKS1 and TNKS2 was not significantly more growth-suppressive than was either of the individual knockdowns in ED1, A549, or Hop62 cells. The growth inhibitory effects of TNKS knockdown were accompanied by stabilization of axin1 at the protein level, as shown in ED1 (Figure 5A) and Hop62 cells (Figure 5B).

To further show Wnt pathway specificity of in vitro antineoplastic effects of these TNKS inhibitors, we performed a rescue experiment. ED1 cells were stably infected with an empty vector or an activated form of β-catenin that cannot be phosphorylated by GSK3 and thus remains constitutively active. Expression of this construct was confirmed at the protein level (Figure 5C). Individual treatments of these stable cell lines with 10 μM TNKS inhibitors for 3 days showed differential growth inhibition between β-catenin expressing cells and controls, with a significant rescue of ~20% growth in each case (Figure 5D).

To confirm independently the transient in vitro knockdown results in the in vivo setting, we achieved stable genetic knockdown of the TNKS, alone or in combination, in the ED1 murine lung cancer cell line with the indicated TNKS shRNAs versus control shRNAs (Figure 6A). ED1 cells were infected with independent shRNAs targeting TNKS1 or TNKS2, or both species. Stable knockdown at the mRNA level was achieved following G418 (TNKS1 and pLKO.1-cmv-neo scrambled control) or puromycin (TNKS2 and TRC2 control) selection, or selection for both shRNA transductants in the dual knockdown study. Knockdown of TNKS1 caused an accumulation of axin 1, as did independent knockdown of TNKS2, although to a lesser extent; combined TNKS knockdown showed increased stabilization of axin 1 (Figure 6B). The in vitro growth inhibitory effects of stable TNKS knockdown are shown (Figure 6C) and are similar to those observed from siRNA-mediated transient TNKS repression.

ED1 murine lung cancer cells after dual TNKS1 and TNKS2 stable knockdown were selected for xenograft studies in nude mice or tail vein injection into syngeneic FVB mice. Mice bearing xenografts of the TNKS knockdown cells showed a decrease in tumor growth rate (Figure 6D) and increase in time to the specified endpoint (Figure 6E) as compared to controls. As expected from the in vitro results, shRNA-mediated knockdown of TNKS1 and TNKS2 led to a significant decrease in syngeneic tumor formation after 8 weeks (Additional file 4: Figure S4).