Gamma-Secretase Inhibitors Abrogate Oxaliplatin-Induced Activation of the Notch-1 Signaling Pathway in Colon Cancer Cells Resulting in Enhanced Chemosensitivity

Microarray data analysis

Gene chip microarrays were constructed from 308 normal and malignant colon tissues obtained at the time of surgical resection, including normal colon mucosa (10%), colonic polyps (15%), primary colon cancers (55%), liver metastases (13%), and lung metastases (6%), using Affymetrix U133A arrays (Affymetrix, Santa Clara, CA). Data preprocessing and statistical analysis were carried out in R (http://www.r-project.org) and Bioconductor (http://www.bioconductor.org). The expression intensities were normalized using the Robust Multiarray Average method (22), which includes background adjustment, quantile normalization across arrays, and probe-level expression measure summarization using median polish on the log2 scale, for each probe set. Differential expression analysis was performed to identify putative genes between sample groups. An Empirical Bayes t test was applied to each gene (23), as a p-value cutoff of 0.01 was used to select differentially expressed genes (p≤0.01). The following sample groups were compared respectively: normal mucosa vs. primary polyp vs. primary colon cancer, primary tumor stage I vs. II vs. III vs. IV, and primary tumor vs. lung metastasis.

Cell lines

The human colon adenocarcinoma cell lines HCT116, SW620, SW480, HT29, and LS513 (American Type Culture Collection, Manassas, VA) were maintained in McCoy's 5a medium (HCT116, HT29), Leibovitz's L-15 (SW480, SW620), or RPMI-1640 (LS513), as recommended by ATCC. Cultures were supplemented with 10% heat-inactivated fetal bovine serum and penicillin-streptomycin (100 units/ml) or 2 mM L-glutamine (SW480, SW620). All cell lines were maintained at 37° C in 5% carbon dioxide and were tested to rule out Mycoplasma contamination.

Drugs

GSI34, a sulfonamide analog, was derived from GSIs as described (24). GSI34 was dissolved in DMSO, stored at -20° C, and diluted in media before use so that the final concentration of DMSO was 0.1% or less in all experiments. The drugs oxaliplatin (Sanofi-Aventis, Bridgewater, NJ), and 5-FU (Pharmacia, Kalamazoo, MI) were obtained from the MSKCC Research Pharmacy. SN-38, the active metabolite of irinotecan, was generously provided by Dr. J. Patrick McGovren (formerly at Pharmacia and Upjohn, Peapack, NJ). Drugs were used at concentrations equal to or less than the IC₅₀ specific to each cell line.

Constructs and small interfering RNA (siRNA)

The pGL2-Hes-1-luciferase and pCS2-NICD constructs were generously provided by Dr. Raffi Kopan (Washington University, St. Louis, MO). The pGL2 and Renilla vectors were obtained from Promega (Madison, WI). siRNA to the following genes were used: Notch-1 from Santa Cruz Biotechnology (Santa Cruz, CA), nicastrin from Santa Cruz and Cell Signaling Technology (Danvers, MA), and Akt1/2 from Cell Signaling. A minimum of two different siRNA sequences were chosen for each gene. Commercially-available siRNA to random noncoding sequences were used for control transfections (Santa Cruz).

Protein immunoblot assays

Cell lines were treated with oxaliplatin (0.5 or 1 μM), SN-38 (2 to 20 nM), or 5-FU (1 to 10 μM), combined with either GSI34 (1 to 10 μM) or 0.1% DMSO (as a control) for 24 to 48 hours. Total protein lysates were prepared. For isolation of nuclear and cytoplasmic fractions, the Pierce NE-PER extraction kit was used (Rockford, IL). Proteins were probed with the following primary antibodies: Notch-1, cyclin D1, and Hes-1 (all from Santa Cruz); NICD, Phospho-Akt^Ser473, Akt, DNA-dependent protein kinase (DNA-PK), mammalian target of rapamycin (mTOR), Phospho-S6^{Ser235/Ser236} ribosomal protein, total S6 ribosomal protein, nicastrin, cyclin D1, and presenilin (all from Cell Signaling); and survivin and Bcl-X_L (Pharmingen-BD Biosciences, San Jose, CA). Equal protein loading was confirmed by probing for α/β-tubulin expression (Cell Signaling). Appropriate secondary antibodies conjugated to horseradish peroxidase were used, including anti-mouse or anti-rabbit IgG (GE-Healthcare, United Kingdom), and proteins were visualized with Amersham ECL Chemiluminescence (GE-Healthcare). Films were digitized with a Microtek scanner (Carson, CA), and images were processed with Photoshop software (Adobe, San Jose, CA).

Hes-1 luciferase assays

Cell lines were co-transfected with pGL2-Hes-1 luciferase reporter (1 μg/well) and the Renilla reporter pRL-CMV (0.1 μg/well) using Fugene (Roche, Switzerland). After 8 to 12 hours, cells were treated with oxaliplatin (0.5 or 1 μM), GSI34 (10 μM), or both drugs for 48 hours. Cells were also treated with SN-38 (2 or 20 nM), 5-FU (1 or 8 μM), GSI34 (10 μM) alone, or the combination of GSI34 with SN-38 or 5-FU for 48 hours. After 6, 12, 24, or 48 hours of drug treatment, total cell lysates were harvested using the Dual-Reporter luciferase assay kit (Stop-and-Glo, Promega), and luciferase activity was quantified on a luminometer (Turner Design, Sunnyvale, CA). Luciferase values were standardized by Renilla pRL-CMV co-transfection.

Clonogenicity assays

Cell lines, in single cell suspension, were plated and treated for 48 hours with oxaliplatin (0.5 to 2 μM), GSI34 (10 μM), or both drugs. Drug-containing media was then removed, and the cells were allowed to grow for a minimum of 2 weeks to form colonies. Colonies were stained with 0.01% crystal violet (Sigma) and quantified in an automated colony counter (ColCount, Oxford-Optronics, England).

Apoptosis assays

Apoptosis was assessed by quantitative confocal fluorescence microscopy. Briefly, following drug treatment, cells were fixed in 4% paraformaldehyde and stained with 4'-6-diamidino-2-phenylindole or DAPI (Sigma). Cells with fragmented nuclei under confocal fluorescence microscopy (40x magnification) were measured as apoptotic. A total of 500 nuclei from five different high-power fields were assessed for each condition.

Viability assays

HCT116 cells were plated in 96-well plates and treated with GSI34 (0 to 40 μM), oxaliplatin (0 to 2 μM), or both drugs for 24 to 72 hours. Viability was assessed with the sulforhodamine B (SRB) assay. Briefly, following drug treatment, cells were stained with 0.4% SRB (Sigma), fixed with 10% trichloroacetic acid, washed with 1% acetic acid, solubilized in 10 mM Tris-buffer, and absorbance was measured at 490 nM on a spectrophotometer (SpectraMax, Molecular Devices, Sunnyvale, CA).

Transfection of siRNA

Cell lines were transfected with siRNA for Notch-1 (Santa Cruz) at 60 pmol/L or with siRNA for Akt (Cell Signaling) at 50 nM using Oligofectamine (Invitrogen, Carlsbad, CA). Cells were also transfected with control siRNA consisting of random sequences (Santa Cruz) at 60 pmol/L. For protein immunoblotting, total cell lysates were harvested at 48 hours. For Hes-1 luciferase assays, cells were first transfected with the Hes-1 luciferase reporter for 12 hours and were then transfected with siRNA (Notch-1, Akt, or control) for 48 hours. For colony formation assays, cells were transfected with siRNA for 12 hours and then oxaliplatin (0.5 to 2 μM), SN-38 (0.2 to 20 nM), or media was added for 48 hours. Colonies were stained after 14 days.

Transfection of NICD

Colon cancer cell lines were transfected with pCS2-NICD (0.5 to 2 μg) or pCS2 vector (0.5 to 2 μg). To confirm protein expression of NICD, total cell lysates were harvested after 24 to 48 hours for immunoblotting. For luciferase experiments, the HES-1 luciferase reporter was transfected for 12 hours, and the cells were then transfected with NICD for 48 hours. For colony formation assays, cells were transfected with NICD for 48 hours, media was removed, and colonies were allowed to grow for 14 days.

Biostatistical analysis

All experiments were conducted in duplicate and were repeated at least twice. Statistical significance was analyzed with Student's T-test (p<0.05).

Overexpression of Notch-1 in colon cancer progression

To support our hypothesis that Notch signaling is important for colon cancer development, we analyzed the expression of Notch genes, along with known chemoresistance genes, in progressive stages of colon cancers. Gene chip microarrays were constructed from 308 normal and malignant colon tissues obtained at surgical resection, including normal colon mucosa (10%), colonic polyps (15%), primary colon cancers (55%), liver metastases (13%), or lung metastases (6%), using Affymetrix U133A arrays. Correlation analysis identified genes differentially expressed between sample classes among a list of 44 genes related to the Notch pathway (Supplemental Fig. 1). The expression of NOTCH1 increased from normal colon mucosa to Stage IV cancers with levels being highest in liver metastases, compared to normal colonic mucosa or liver parenchyma (trend test, p<0.001, Fig. 1A). In contrast, NOTCH2 did not similarly increase with disease progression (p<0.001, Fig. 1A). The pattern of NOTCH1 overexpression mirrored that of other genes involved in chemoresistance, including BCL2, BIRC5 or survivin, and CCND1 or cyclin D1 (Fig. 1A). In addition, the expression of the downstream target HES1, a target gene of Notch-1, increased with colon cancer progression (Fig. 1B). We then examined the levels of two genes that regulate Notch expression. GSK3B, (glycogen synthase kinase-3-beta), which positively regulates Notch-1 by preventing its phosphorylation (25), increases with colon cancer stage progression (p<0.001, Fig. 1C). In contrast, NUMB, a negative-regulator of Notch-1, is downregulated in advanced colon cancers (p<0.001, Fig. 1C). This Notch-1 overexpression may be ligand-independent, as the Notch ligand JAG1 was not increased with colon cancer progression (Fig. 1D).

Oxaliplatin induces Notch-1 protein and activity

Because Notch-1 expression is increased with colon cancer progression, we hypothesized that it may play a role in chemoresistance. First, we examined if Notch-1 signaling is affected by chemotherapy by treating colon cancer cells with oxaliplatin, a platinum-derived chemotherapy drug that damages DNA through the formation of DNA adducts. In two colon cancer cell lines, HCT116 and SW620, oxaliplatin induced NICD protein in a dose-dependent manner (Fig. 2A, top). The induction was independent of p53 status, as HCT116 has wild-type p53 and SW620 has mutant p53. It was then examined if the increase in NICD protein could augment Notch-1 signaling by measuring the activity of a downstream target, HES1. HCT116 and SW620 cells were transfected with a luciferase construct containing the promoter of HES-1 and were then treated with oxaliplatin. In both cell lines, similar to the induction of NICD protein, oxaliplatin induced Hes-1 activity in a dose-dependent manner, although the degree of induction in SW620 cells was five-fold higher (Fig. 2A, bottom). The effect was not limited to these two lines, as oxaliplatin also increased Hes-1 activity in three other colon cancer lines, HT29 (mutant p53), LS513 (wild-type p53), and SW480 (mutant p53), although the degree of Hes-1 induction again varied (Fig. 2B). We then examined if the increase in NICD by oxaliplatin activated the Notch pathway by measuring downstream targets. In HCT116 cells, oxaliplatin induced Hes-1 protein, as suggested by the luciferase data, as well as the Notch-1 targets cyclin D1 and survivin (Fig. 2C). Finally, it was determined if other chemotherapies used to treat metastatic colon cancer can also activate Notch signaling. Both the antifolate fluorouracil (5-FU), a pyrimidine analog, and SN-38, the active metabolite of the topoisomerase I inhibitor, irinotecan, increased Hes-1 activity (Fig. 2D).

Oxaliplatin induces gamma-secretase protein expression

We then examined the possible mechanism underlying the induction of NICD protein by chemotherapy. NICD is produced when the Notch-1 receptor is cleaved by the gamma-secretase complex, composed of four subunits: presenilin (PS1), nicastrin (NCT), anterior pharynx-defective (APH-1), and presenilin enhancer-2 (PEN-2). Therefore, we hypothesized that chemotherapy induced NICD protein by increasing gamma-secretase activity. Treatment of HCT116 cells with oxaliplatin induced the expression of two subunits, PS-1 and NCT (Fig. 3A). We then downregulated gamma-secretase production by using siRNA to target NCT. Transfection of HCT116 cells with NCT siRNA inhibited the protein levels of NCT (Fig. 3B, top), which, effectively decreased Notch signaling, as measured by Hes-1 activity (Fig. 3B, bottom). In HCT116 cells, suppression of nicastrin protein with siRNA prevented the induction of NICD and cyclin D1 protein usually observed after oxaliplatin (Fig 3C).

Gamma-secretase inhibitors abrogate Notch-1 induction by chemotherapy to enhance colon cancer chemosensitivity

We then hypothesized that the increase in Notch-1 by oxaliplatin could be blocked if the cell lines were co-treated with a gamma-secretase inhibitor (GSI). Our group used a novel sulfonamide analog, GSI34, synthesized as described (24). First, treatment of HCT116 cells with GSI34 effectively blocked the production of NICD protein at baseline (Fig. 4A, top). In the presence of either GSI34 or 0.1% DMSO as a control, HCT116 cells were then treated with increasing doses of oxaliplatin. At each dose level, the addition of GSI34 abrogated the induction of NICD protein by oxaliplatin (Fig. 4A, top). It was then examined if GSI34 could inhibit the activation of Hes-1 by oxaliplatin. Co-treatment with GSI34 abrogated the induction of Hes-1 by oxaliplatin at 0.5 μM but did not completely suppress the induction at the higher dose of 1 μM (Fig. 4A). GSI34 also decreased Notch-1 induction by oxaliplatin in four other colon cancer cells: HT29, LS513, SW480, and SW620 (Fig. 4B). Finally, co-treatment of HCT116 cells with GSI34 suppressed the induction of Hes-1 by other chemotherapies, including 5-FU alone (Supplemental Fig. 2A) or 5-FU combined with oxaliplatin (Supplemental Fig. 2B).

Because it was hypothesized that activation of Notch-1 may protect colon cancer cells from chemotherapy, we examined if inhibition of Notch-1 could decrease chemoresistance. Cell lines were treated with oxaliplatin combined with GSI34 or DMSO for 48 hours, and viability was then measured in a colony forming assay. In HCT116 cells, treatment with oxaliplatin or GSI34 as single agents decreased colony formation by 40%, but the combination of GSI34 and oxaliplatin decreased colony formation by 60% (Student's t-test, p=0.034, Fig. 4C). Similar reductions in colony formation by GSI34 and oxaliplatin were observed in SW620 and LS513 cells (Supplemental Fig. 2C). GSI co-treatment also enhanced the chemosensitivity of HCT116 to SN-38 and to 5-FU (p<0.05, Fig. 4C) Using isobologram experiments, we confirmed that the combination GSI34 and oxaliplatin synergistically decreased the IC₅₀ in HCT116 cells (Supplemental Fig. 2D). It was then examined if colon cancer cells co-treated with oxaliplatin and GSI34 were undergoing enhanced apoptosis. Co-treatment of HCT116 cells with GSI34 and oxaliplatin significantly increased the percentage of apoptotic cells, compared to either treatment alone, as determined by DAPI staining for nuclear fragmentation (p<0.05, Fig. 4D).

Notch-1 expression affects chemosensitivity

Because the gamma-secretase complex cleaves multiple transmembrane receptors besides Notch-1, including the other Notch receptors (Notch-2 to Notch-4), we wanted to confirm that the enhancement of chemosensitivity by GSI34 was mediated by Notch-1. We used multiple siRNA constructs to specifically inhibit Notch-1 to determine if the chemosensitivity observed with GSI34 could be replicated. Transfection of HCT116 cells with siRNA to Notch-1 for 24 hours decreased the expression of Notch-1 protein (Fig. 5A, top), which decreased the activity of Hes-1 (p=0.014, Fig. 5A, bottom), compared to cells transfected with control siRNA. Notch-1 siRNA also decreased the clonogenicity of HCT116 cells at 48 hours (Fig. 5A, right). It was then determined if Notch-1 siRNA could enhance chemosensitivity to oxaliplatin. Like the results with GSI34, the combination of Notch-1 siRNA and oxaliplatin decreased viability significantly, compared to either siRNA or chemotherapy alone (p=0.011, Fig. 5A, right).

Because inhibition of Notch-1 could enhance chemosensitivity, we then examined if overexpression of Notch-1 could instead increase colon cancer chemoresistance. HCT116 cells were transfected with an NICD construct for 24 hours, which significantly increased NICD protein (Fig. 5B, top), and activated Hes-1 (Fig. 5B, bottom). After transfection with NICD or empty vector for 12 hours, HCT116 cells were then treated with oxaliplatin for 48 hours. Overexpression of NICD did not significantly change the percentage of colonies, compared to vector-transfection (Fig. 5B, right). Although treatment with oxaliplatin decreased the percentage of colonies by 50%, overexpression of NICD protected against oxaliplatin, as no decrease in colony formation occurred (Fig. 5B, right). We then examined if Notch-1 expression could affect the chemosensitivity of HCT116 cells to SN-38. HCT116 cells were first transfected with either Notch-1 siRNA or with NICD, and the percentage of viable cells following treatment with SN-38 was then measured using the SRB viability assay. Again, Notch-1 siRNA enhanced chemosensitivity to SN-38, whereas overexpression of NICD increased resistance to SN-38 (Fig. 5C).

Gamma-secretase inhibitors decrease the induction of pro-survival factors, including Akt signaling, by oxaliplatin

We then investigated the mechanisms underlying how inhibition of Notch-1 signaling may enhance chemosensitivity. First, we examined if GSI co-treatment could decrease several pro-survival factors implicated in chemoresistance to oxaliplatin. Treatment of HCT116 cells with GSI34 decreased the baseline levels of Bcl-X_L and survivin and suppressed the induction of cyclin D1 by oxaliplatin (Fig. 6A, top). Second, because Notch-1 may activate PI3K/Akt signaling to protect against DNA damage (7), the expression of proteins in this pathway was examined. In HCT116 cells, co-treatment with GSI34 decreased the phosphorylation of Akt at Serine473 by oxaliplatin and also mildly decreased total Akt levels (Fig. 6A, bottom). We then examined two targets of Akt, mTOR and S6 ribosomal protein. Levels of total mTOR, phosphorylated S6 at Serine235/236, and total S6 protein levels were decreased following co-treatment with GSI34 and oxaliplatin (Fig. 6A, bottom). We then examined the expression of a kinase that may activate Akt, DNA-PK, as co-treatment with GSI34 and oxaliplatin decreased the protein levels of DNA-PK (Fig. 6A, bottom). Because intact Akt signaling may play a role in chemoresistance, we determined if suppression of Akt with siRNA would affect chemosensitivity to oxaliplatin. siRNA to Akt effectively suppressed total Akt protein levels (Fig. 6B, top), and decreased the viability of HCT116 cells following oxaliplatin (Fig. 6B, bottom). These results suggest that inhibition of Notch-1 signaling may sensitize colon cancer cells to oxaliplatin by preventing the activation of pro-survival pathways, including Akt, after DNA damage.