|Home | About | Journals | Submit | Contact Us | Français|
Carcinoids are neuroendocrine (NE) tumors with limited treatment options. Notch activation has been shown to suppress growth and hormone production in carcinoid cells. In this study, we provide a process for identifying Notch activating compounds using a high throughput screen (HTS), and we validate the effects of the strongest hit from the 7,264 compounds analyzed by HTS, resveratrol (RESV). Treatment of carcinoid cells with RESV resulted in upregulation of the Notch signaling pathway as measured by suppression of its downstream target achaete-scute complex-like 1 (ASCL1). Luciferase reporter assays incorporating the centromere-binding factor 1 binding site also confirmed the functional activity of RESV-induced Notch. As activation of the Notch pathway has been shown to suppress carcinoid proliferation, RESV treatment of carcinoid cells led to a dose-dependent inhibition of cellular growth. Immunoblotting revealed phosphorylation of cdc2 (Tyr15) and upregulation of p21Cip1/Waf, markers of cell cycle arrest, with RESV treatment. Flow cytometry confirmed the mechanism of RESV-induced growth inhibition is S phase cell cycle arrest. Furthermore, as Notch has been shown to inhibit bioactive hormone production from NE tumors, RESV also suppressed expression of the NE peptides/hormones chromogranin A and serotonin. RNA interference assays demonstrated that the hormone suppressing capacity of RESV was due to upregulation of the Notch2 isoform. Thus, this HTS can be utilized to identify novel Notch activating compounds which may have the potential to suppress carcinoid tumor growth and the associated endocrinopathies.
The Notch signaling network is an intercellular signaling pathway involved in cell fate determination, stem cell potential and lineage commitment 1–3. The Notch genes encode 300kDa multifunctional transmembrane receptors. Binding of one of the Notch ligands promotes a sequence of proteolytic cleavage events mediated by γ-secretase and convertase enzymes resulting in the activation of Notch intracellular domain (NICD) 4, 5. The active carboxyl-fragment of Notch translocates to the nucleus where it interacts with the DNA-binding protein complex CSL (centromere-binding factor 1 [CBF-1/RBPjκ], suppressor of hairless and Lag-1) to transactivate target genes which are transcriptional repressors of basic helix-loop-helix (bHLH) factors like achaete-scute complex (ASCL1)4, 6, 7.
The biologic function of Notch signaling is cell-context dependent, with the oncogenic role of truncated, active Notch1 being first identified in T cell acute lymphoblastic leukemia 8 and subsequently being shown to stimulate cell proliferation and prevent apoptosis in breast carcinoma 9, 10, melanoma 11, lung adenocarcinoma 12, and kidney epithelial cells 13, 14. Notch signaling may have opposing actions on cellular growth and differentiation, as demonstrated by the tumor suppressing function in epidermal keratinocytes 15, 16. Loss-of-function experiments in these cells demonstrate that the Notch1 homologue and p21WAF1/Cip1, a downstream Notch target in keratinocytes, suppress Wnt ligand expression and negatively regulate tumorigenesis 17, 18.
Previous studies have demonstrated conservation of the Notch signaling pathway in gastrointestinal (GI) and pulmonary carcinoid tumors 19. We and others have demonstrated that the downstream target of Notch signaling, ASCL1 transcription factor, can modulate the NE phenotype in carcinoid tumors 19–22. We hypothesize that regulation of the Notch pathway by small molecule activators may exploit a key cellular signaling mechanism regulating NE cell growth and may represent a novel therapy for patients with carcinoid disease. In the present study, we sought to develop a cell based assay to screen for activators of Notch signaling utilizing high throughput screening (HTS) technology. The strongest hit from this screen was resveratrol (trans-3,4’,5-trihydroxystilbene; RESV), a dietary polyphenol which has been shown to have chemopreventive activity against a variety of cancers. Further experiments validate that the effects of RESV on human carcinoid tumor cells are consistent with Notch pathway activation, and support the concept of utilizing this HTS to identify compounds which may have anti-proliferative and anti-hormone producing properties in carcinoid cancer cells.
Human GI carcinoid tumor cells (BON)—provided by Mark Evers and Courtney Townsend (Galveston, TX)—and human pulmonary carcinoid cells (NCI-H727), obtained from ATCC (Manassas, VA), were maintained in DMEM/F12 and RPMI 1640 respectively, supplemented with 10% fetal bovine serum 100 IU/ml penicillin and 100 Ag/ml streptomycin.
To measure functional Notch activity, a BON cell line stably expressing a CBF1/luciferase reporter plasmid was created. JH23A plasmid (Diane Hayward, Baltimore, MD) was digested with kpn1/hindIII to remove a 4xCBF1 binding site. This was subsequently ligated into pGL4.20 (Promega, Madison, WI). BON cells transfected with the CBF1/luciferase plasmid were selected with puromycin. The resultant clone (BON-CBF1-luc) was validated by treatment with Valproic acid or suberoyl bishydroxamic acid (SBHA), and luciferase activity (luminescence) was measured. Luminescence was assessed using a Monolight 3010 luminometer (San Diego, CA).
A library of 7,264 compounds was utilized to perform HTS. The KBA01 Library consists of 3 commercially available collections totaling 4,160 compounds. KBA01 consists of 880 high purity compounds with known safety and bioavailability profiles in humans from Prestwick Chemicals (Washington, D.C.), 2,000 diverse FDA-approved drugs and natural products from Microsource Discovery Systems (Gaylordsville, CT) and 1,280 compounds from the Library of Pharmacological Active Compounds representing marketed drugs, failed developments, and “gold standards” with well-characterized activities. Additional compounds were obtained from the NCI Developmental Therapeutics Program (NCI-DTP), consisting of 1990 molecules selected from 140,000 compounds. The mechanistic diversity set consists of 879 molecules that represent a broad range of activities in the NCI -60 cell line screen performed at the NCI-DTP was also screened. The Natural Products Collection from the NCI-DTP consists of 235 natural products selected from the open repository of 140,000 compounds. BON-CBF1-luc cells were plated in 384-well microtiter dishes using a µFill non-contact reagent dispenser (Biotek, Winooski, VT). 0.5µL library compounds were added using a Biomek FX liquid handler (Beckman-Coulter, Brea, CA). KBA was screened at a final compound concentration of 1uM, and the NCI compounds were screened at a final concentration of 10uM Assays were read on a Safire 2 microplate reader (Tecan, Männedorf, Switzerland).
Percent bioluminescence of the positive control was calculated, and the Z-Factor and Z-prime values were determined for each plate. The Z-Factor and Z-prime averaged 0.63, and 0.77 respectively, indicating that the assay was robust. Compounds that showed over 150% increase in signal were used to calculate the Z-factor (n=112), whereas VPA which showed an average increase of 300% was used to calculate the Z-prime (n=16). All data was analyzed in the UW-Small Molecule Screening Facility database, Activitybase (IDBS Inc., Alameda, CA), containing over 50 sets of screening data on the compounds screened in this project, to rule out non-specific promiscuous inhibitors.
Carcinoid tumor cell proliferation was measured by MTT assay. The cells were then treated with RESV (Biomol International, Plymouth Meeting, PA) in concentrations of 0–125µM.
Denatured cellular extracts (20–40µg) were resolved by 10%–12% SDS-PAGE (Invitrogen), transferred onto nitrocellulose membrane (Bio-Rad, Hercules, CA), blocked in milk, and incubated overnight in the appropriate primary antibody. The antibody dilutions were as follows: 1:1,000 for phospho-cdc2 (Tyr15) (Cell Signaling, Beverly, MA), and mammalian achaete scute homologue-1 (MASH1) for ASCL1 (BD PharMingen, San Diego, CA); 1:2,000 for p21Waf1/Cip1, CgA, cyclic B1, and CDK2 (Cell Signaling); and 1:10,000 for glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Trevigen, Gaithersburg, MD). Horseradish peroxidase-conjugated goat anti-rabbit or goat anti-mouse secondary antibodies (Pierce) were used. Immunstar (Bio-Rad) or SuperSignal West Femto (Pierce) kits were used.
BON and NCI-H727 cells were treated with or without RESV for 48 h. Cellular viability was determined using trypan blue dye exclusion. For DNA content analysis, cells (1 × 106) were fixed with cold 70% ethanol, washed with PBS, incubated with 0.2mg/ml RNAse-A and stained with 10µg/ml PI staining solution. FACS analysis was performed on the FACSCalibur flow cytometer (BD Biosciences) and results were analyzed with ModFitLT3.2 software (Topsham, ME).
To determine 5-HT levels in cellular extracts of carcinoid cells treated with RESV (0–100µM), we utilized a 5-HT ELISA kit ( Concord, MA). 5-HT values were quantified relative to the control. Two independent experiments in triplicate were analyzed.
siRNA against Notch1 or Notch2 and nonspecific siRNA (Santa Cruz, sc-44226, sc-40135, sc-40986, and sc-37007) were transfected into BON GI carcinoid cells using Lipofectamine 2000 (Invitrogen).
Total RNA was isolated using silica-gel membrane-based spin-column technology (RNeasy Mini-/Micro-Kit, Qiagen, Valencia, CA). cDNA was generated from 2µg of total RNA by iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA). Sequences for Notch1-specific primers were as follows: sense, 5’-GTCAACGCCGTAGATGACCT-3’; antisense, 5’-TTGTTAGCCCCGTTCTTCAG -3’. Sequences for Notch2-primers were: sense, 5’-TGTGACATAGCAGCCTCCAG-3’; antisense, 5’-CAGGGGGCACTGACAGTAAT-3’. GAPDH- primers (sense, 5’-ACCTGCCAAATATGATGAC-3’; antisense, 5’-ACCTGGTGCTCAGTGTAG-3’). Reverse transcription was performed in a thermal controller (MJ Research, Watertown, MA). The qPCR reactions were performed on the BioRad iCycler, with 3 minutes 95°C, 35 cycles 30 seconds at 95°C, 25 seconds at 60°C and 30 seconds at 72°C followed by 1 minute at 95°C and 1 minute at 55°C. Results were normalized to GAPDH. Expression was then plotted as average ± standard error of the mean (SEM).
NCI-H727 cells (1 × 106) were suspended in Hanks’ balanced salt solution (Invitrogen) and injected s.c. into the left flanks of 20 nude athymic NU/NU mice (Charles River Laboratories, Wilmington, MA), which were subsequently divided into two groups of 10 animals each. Animal protocols were approved by the University of Wisconsin-Madison Research Animals Resources Center. Ten days following inoculation, mice were given 5mg/kg of RESV (Biomol) in 1× PBS (Invitrogen) or solvent alone in 1× PBS by i.p. injections daily for 15 days. The growth rate of s.c. tumors was monitored every 5 days. Tumor size in the animals was measured every 5 days with a Vernier caliper in two dimensions (length and width) measuring to the nearest millimeter, and the tumor volumes were calculated using the previously validated formula: volume=length × width2 × 0.52.
Statistical analyses were performed using analysis of variance testing (SPSS software version 10.0, SPSS; Chicago, IL). P <0.05 was considered statistically significant.
The BON human GI carcinoid cell line was stably transfected with pCBF1-Luc containing the CBF1 responsive elements cloned into the pGL4.20. A luciferase assay was used to measure relative activity of multiple stable clones with the known Notch activating compounds, valproic acid (VPA) and suberoyl bis-hydroxamic acid (SBHA) 23, 24. One of the clones which demonstrated low basal luminescence and high induction activity (named BON-CBF1-Luc) was used for all subsequent experiments (Figure 1A). Having established that the BON-CBF1-Luc cells were sensitive to Notch activating compounds, we investigated the dose-dependent effects of these same compounds. Both compounds stimulated luciferase activity in a dose-dependent manner (Figure 1B).
Pilot screens were performed in duplicate, testing the entire KBA01 and the National Cancer Institute Diversity, Challenge, Mechanistic, and Natural Products collections. Compounds that showed over 150% activation were selected for retesting, 112 in total. 27 of 7,264 tested compounds produced luminescence ≥ 300% of the negative control in both replicates, and were defined as initial hits. Results of the two duplicate screens were in agreement. The 27 compounds were then evaluated in a secondary HTS at two concentrations (1µM and 10µM) (Figure 1C). The strongest hit from was RESV. Interestingly, the triacetylated derivative of RESV—triacetyl resveratrol (Figure 1C, “compound C”)—exhibited similar luminescence on the secondary screen.
The Notch signaling pathway is a significant regulator of NE differentiation and hormone production in GI 21 and pulmonary 25, 26 carcinoids. It has been previously shown that Notch signaling is absent in neuroendocrine tumors, and that Notch overexpression with an inducible Notch construct causes inhibition of cell growth and hormone production 19–21, 27, 28. Based on the results of our HTS screen for Notch-activating compounds, we wanted to validate that RESV induced Notch signaling in carcinoid cells.
To determine if RESV activated the Notch signaling pathway, a luciferase reporter assay incorporating the CBF1 binding site was used. RESV treatment of GI (BON) carcinoid cells resulted in a nearly threefold induction of luciferase activity (Figure 2A). Notch activation is dependent upon cleavage of the Notch receptor by γ-secretase to the active Notch intracellular domain (NICD). Importantly, the γ-secretase inhibitor DAPT blocked the effect of RESV on Notch-CBF1 binding, indicating that the increase in CBF1 luciferase activity was a result of Notch induction.
To further confirm that Notch-CBF-1 binding observed which RESV resulting in downstream signaling pathway alterations associated with Notch activation, Western blot analysis of RESV-treated GI and pulmonary carcinoid cancer cells was performed. We and others have previously shown that Notch activation in carcinoid cells results in suppression of ASCL1, a basic helix-loop-helix transcription factor that is regulated transcriptionally by Notch 20, 21. As shown, immunoblot analysis of RESV-treated BON and NCI-H727 carcinoid cancer cells (Figure 2B) demonstrated a dose-dependent decrease in ASCL1 protein. Exposure of both BON and NCI-H727 cells to 100µM RESV for 48h suppressed ASCL1 to nearly undetectable levels, consistent with the developmental role of Notch to extinguish ASCL1 expression.
Carcinoid cancer cells (BON and NCI-H727) were treated with 0–125µM of RESV for up to 6 days and analyzed by MTT. Treatment of BON (Figure 3A) and NCI-H727 (Figure 3B) carcinoid cells with RESV resulted in a dose-dependent and time-dependent inhibition of cell proliferation. Significant growth inhibition was seen after 4 days of exposure to 25µM of RESV in both BON and NCI-H727 cell lines. After 6 days, growth of BON and NCI-H727 cells exposed to 50µM of RESV was inhibited greater than 50%.
Based on the observed growth inhibition associated with RESV-treated BON and NCI-H727 cells, we next assessed the levels of cell cycle regulators associated with cell cycle arrest. Previously, RESV has been shown to induce cell cycle arrest in a variety of cancer cell lines by modulating expression of p21Waf1/Cip1 and phosphorylated cdc2 (Tyr15) 29–31. RESV treatment (0–50µM for 48h) induced protein expression of p21Waf1/Cip1 and CDK2, and cyclin B1 accumulation was observed (Figure 4A). Furthermore, levels of phosphorylated cdc2 (Tyr15) increased following RESV treatment without any change in total cdc2 level (Figure 4), suggesting G1/S phase arrest. Overall, these results suggest the possible involvement of cdc2, CDK2, p21Waf1/Cip1, and cyclin B1 in RESV-induced cell cycle arrest in BON and NCI-H727 cells. It should be noted that immunoblot analysis for cleaved poly ADP-ribose polymerase (PARP) and caspases 3/9 revealed minimal to no induction of apoptosis in RESV-treated carcinoid cells (data not shown).
After establishing that RESV induces expression of specific cell cycle inhibitors, we were interested in examining the effects of RESV on cell cycle phasic progression. BON and NCI-H727 cells were treated with either DMSO alone or RESV (25, 50, and 100µM). After 48 h of treatment, cells were labeled with propidium iodide (PI) and analyzed by DNA flow cytometry. A representative histogram for the BON cells is shown in figure 4B, and the data obtained with carcinoid cell lines are summarized in Figure 4C. In both BON and NCI-H727 cells, RESV treatment resulted in a dose-dependent, statistically significant increase in the percentage of cells in the S phase of the cell cycle, accompanied by a decrease of cells in the G1 and G2/M phases (Figures 4B–C). There were no prominent Sub-G1 apoptotic peaks detected in either cell line at 48 h.
Carcinoid tumors typically contain dense neurosecretory granules which contain hormones and neuropeptides that biologically characterize the NE phenotype of carcinoid disease. CgA is an acidic glycoprotein that is cosecreted with hormones such as serotonin by carcinoid tumors, and it is thought to be the precursor to several functional peptides including vasostatin 32 and pancreastatin 33, 34. All NE tumors produce CgA, making this protein a useful marker for this class of malignancies both in vitro and in vivo. Activation of the Notch pathway in NE cancer cells has been shown to inhibit bioactive peptide/hormone production. We therefore examined whether RESV could inhibit the expression of CgA in BON and NCI-H727 cells. After 48 h of treatment of carcinoid cancer cells with RESV, Western blot analyses (Figure 5A) demonstrated a significant dose-dependent reduction in CgA expression, indicating a change in the NE phenotype of these tumor cells.
We next sought to evaluate the effect of RESV treatment on 5-HT production in BON cells by ELISA analysis. Previous studies have demonstrated that 5-HT is an autocrine growth factor for BON cells 35 and is implicated in the pathogenesis of the carcinoid syndrome. As shown in Figure 5B, treatment of BON cells with increasing doses of RESV resulted in a progressive reduction of 5-HT. The relative 5-HT reductions were 11% at 50µM RESV and 59% at 100µM RESV after 48 h of drug treatment (Figure 5B). Importantly, this dramatic decrease in 5-HT concentration was detected in whole cell lysates of RESV-treated BON cells rather than the supernatant of RESV-treated cells (data not shown), suggesting that RESV inhibits the cellular production rather than the secretion of 5-HT.
In vitro, RESV affects the initiation, promotion, and progression of carcinogenesis by modulating signal transduction pathways that regulate cellular differentiation, growth, apoptosis, angiogenesis, and metastasis 36. Besides Notch signaling, RESV is known to influence important mitochondrial enzymes like complex I 37 and the F0F1-ATPase 38, 39, nuclear factor-κB 40, 41, CYP1A1 42, Sirtuin enzymes (Sirt1) 43, VEGFR 44, and hypoxia-inducible factor-1α (HIF-1α) activity and expression 45, 46. Furthermore, we have shown that RESV activates Notch signaling but the exact isoform of Notch associated with the observed effects on carcinoid cells is unclear. To confirm that the phenotypic effects of RESV—i.e., inhibition of CgA, 5-HT, and downregulation of ASCL1 protein expression—are a result of activation of Notch signaling and to determine the dominant Notch isoform, small interfering RNA (siRNA) targeting Notch1 and Notch2 were used to knock down gene expression. BON carcinoid cells were transiently transfected with siRNA against Notch1 or Notch2, nonspecific siRNA, or vehicle (Lipofectamine) alone, and mRNA and protein expression were analyzed by quantitative RT-PCR and immunoblotting, respectively. Notch-specific mRNA analysis by RT-PCR indicated that Notch1 and Notch2 siRNA transfection resulted in more than a 60% and 80% decrease in Notch1 and Notch2 mRNA basal expression, respectively (Figure 6A). In the absence of siRNA, RESV treatment led to lower levels of ASCL1 and CgA (Figures 6B, lanes 1–2). Similar results were obtained in the presence of nonspecific siRNA (Figures 6B, lanes 3–4). Blockade of RESV-mediated Notch1 induction did not reverse the RESV-mediated changes in ASCL1 and CgA expression (Figures 6B, lanes 5–6). Importantly, the abrogation of Notch2 induction with siRNA reversed the RESV-mediated changes in ASCL1 and CgA expression. Similar results were noted with 5-HT production as evaluated by serotonin ELISA (Figure 6C). These data suggest the phenotypic regulation seen with RESV treatment is mediated primarily by Notch2 signaling, with minimal contribution from Notch1 activation.
After confirming that RESV suppresses carcinoid cell growth in vitro, we were interested in testing its efficacy in vivo. After subcutaneous flank injection of 1 × 106 NCI-H727 cells, measurable tumors developed in the injection site in all mice given the transplants (in the middle of the left posterior quadrant of the abdomen). Visible and measurable tumors were observed beginning on day 10 after transplantation. Tumors grew as solid masses in the subcutaneous tissue of the abdominal wall. Histological examination demonstrated tumor cells exhibiting a monotonous morphology, with delicate intervening fibrovascular stroma (data not shown). The pulmonary carcinoid cells contained scant, pink granular cytoplasm with minimal variation in cell and nuclear size, characteristic of carcinoid cellular morphology (data not shown). When left untreated, tumors reached a volume of approximately 350 mm3 by day 25 (Figure 7). Impressively, 15 daily courses of 5mg/kg RESV significantly delayed tumor growth without demonstrating any significant morbidity or mortality to the mice in the treatment group. In fact, the growth of the tumors in the treatment group was almost static, demonstrating a nearly 65% reduction in tumor volume when compared to control mice after 15 days of treatment. These data provide evidence that systemically administered RESV is able to effectively repress carcinoid tumor growth in vivo.
Patients with metastatic carcinoid tumors, rare neuroendocrine (NE) neoplasms characterized by the secretion of amines and polypeptide hormones like serotonin (5-HT) and histamine, often have debilitating symptoms of the carcinoid syndrome, including diarrhea, bronchospasm, flushing, and thickening of the right-sided heart valves 47–50. Widespread metastases render conventional surgical treatment palliative at best and therefore there exists a great need to develop novel targeted therapeutic strategies both to reduce tumor burden and control symptoms of the carcinoid syndrome.
Recently, Notch signaling has been shown to play an essential role in the neuroendocrine (NE) differentiation of cells in the lung and GI tract 6, 51. In the developing fetal lung epithelium, Ito et al demonstrated strong immunohistochemical staining for Notch1, Notch2, Notch3, and the immediate Notch effector HES-1 in non-NE pulmonary cells, whereas the mammalian homologue of achaete-scute complex-like 1 (ASCL1) was found only in clusters of cells belonging to the pulmonary NE cells (PNECs) 25. Thus, the overall effect of Notch activation is to limit the number of cells that can differentiate into enteroendocrine cells.
In the present study, we identified multiple small molecule activators of the Notch pathway by HTS and validated the strongest hit, RESV, in a series of studies. Treatment of BON GI carcinoid cells and NCI-H727 bronchopulmonary cells with RESV resulted in marked induction of the Notch signaling pathway, and profound growth inhibition associated with S-phase cell cycle arrest. Western blot analysis revealed upregulation of p21WAF1/Cip1 and phosphorylated cdc2 (tyr15). Active Notch signaling led to a dramatic reduction in ASCL1 expression, as well as marked inhibition of CgA and serotonin. Transfection of Notch antisenseRNAs into GI carcinoid cells blocked the effects of RESV on Notch expression, ACSL1 suppression, and the NE markers CgA and serotonin. This study is the first to report the results of a large-scale HTS to identify activators of the Notch signaling pathway and validates the strongest hit from this screen.
The effects of RESV on other cancer cells have been extensively described. Broadly, in in vitro models, RESV inhibited the growth of tumor cell lines derived from various human cancers. This effect has been associated with the ability of RESV to arrest cell cycle progression, to promote cell differentiation and to induce programmed cell death by caspase-independent or caspase-dependent apoptosis or by autophagocytosis. Specifically, in human colorectal cancer cells, RESV has been shown to induce cell death by a novel pathway involving lysosomal cathepsin D 52. Similarly, using microarray gene expression profiling and high-throughput immunoblotting (PowerBlot) methodologies, Whyte et al. showed the arrest of lung cancer cells in the G1 phase of the cell cycle and demonstrated this growth arrest is mediated via the transforming growth factor pathway, particularly through the Smad proteins53. While these and other studies intimate that RESV possesses cancer chemopreventive properties in vivo, several phase I pharmacokinetic dose-escalation studies and our own data support the hypothesis that RESV is safe and results in minimal adverse effects on normal cells. In a phase I study of oral RESV (single doses of 0.5, 1, 2.5, or 5 g), Boocock et al. showed that RESV was well tolerated with follow-up of over 40 healthy volunteers failing to reveal any serious adverse reaction either clinically or by biochemical and hematologic analyses54. Similarly, our mouse xenograft studies with i.p injections failed to demonstrate any adverse effects in either the control or RESV-treated mice over the 2-week treatment period.
The mechanism by which RESV regulates ASCL1 expression is a subject of ongoing investigation, as the dietary polyphenol is known to influence diverse biologic pathways. Several studies have demonstrated that an early target of RESV is the sirtuin (Sirt) class of nicotinamide adenine dinucleotide (NAD)-dependent deacetylases 55. It has been shown that under oxidizing conditions in neuronal precursor cells, induced Sirt1 and Hes1 form a complex that binds to and deacetylates histones at the mASH1 promoter, while recruiting co-repressors such as TLE1 56. Together, these events cause downregulation of ASCL1 expression and block neuronal differentation; furthermore, the influence of the redox state on cell-fate decisions was eliminated by removal of Sirt1 activity by siRNA or through the use of Sirt1 inhibitors. Interestingly, RESV-induced suppression of ASCL1 in human carcinoid cells appears to be independent of Sirt1 activity, as blockade of RESV-mediated Sirt1 induction did not reverse the RESV-mediated changes in ASCL1 protein expression (data not shown).
RESV-induced regulation of ASCL1 expression, and subsequently the NE phenotype in human carcinoid tumor cells, more certainly appears to be regulated by Notch signaling. In medullary thyroid cancer (MTC), a NE tumor derived from the calcitonin-producing C-cells of the thyroid gland, Notch1-mediated silencing of ASCL1 gene transcription results in decreased expression of CgA and calcitonin 20. Similarly, stable expression of Notch1 in BON cells has been shown to result in decreased expression of CgA, 5-HT, SYP, and NSE 19. In the present study, we clearly demonstrate that RESV induces functional Notch signaling, the first time that this has been shown in human carcinoid cells. More importantly, it is the relative contribution of Notch2 rather than Notch1 induction which seems to be responsible for RESV-induced regulation of ASCL1 protein expression, as demonstrated by Notch1 and Notch2 siRNA knockdown studies. Based on these findings it is suggested that if Notch1 and Notch2 ICDs exert the same downstream effects; therefore, they may be used interchangeably in scenarios where the goal is to exogenously activate the Notch signaling pathway for the purpose of controlling cell fate.
Utilization of cell based HTS systems should be done with caution and represent a limitation of these data. The current cell based HTS is based on the NCI-60 cell line screen system, an anticancer drug screen developed in the late 1980s as an in vitro drug-discovery tool intended to supplant the use of transplantable animal tumors in anticancer drug screening57. However, such cell line screen systems remain in their relative infancy and are not mature enough to utilize as primary therapeutic indicators. In cell based, high-throughput molecular-targeted screens, compounds can potentially function in various ways to activate cell based reporters. Molecular action in any of several 'upstream' pathways could result in activity. Therefore, it is essential to identify the detailed mechanism of action of selected compounds, often a very challenging laboratory task. In the present study, RESV was identified as a potential Notch activator based on the preliminary HTS screen; furthermore, extensive laboratory testing confirmed that RESV does, in fact, activate functional Notch signaling and that this activation is responsible for the phenotypic changes seen in vitro.
In summary, we described the use of a novel, laboratory-derived cell based HTS for the identification of Notch activating compounds. We analyzed 7,264 compounds and confirmed that the top HTS hit, RESV, activates the Notch signaling pathway in human carcinoid cells in vitro. Importantly, this dietary polyphenol induces functionally active Notch which represses ASCL1 protein expression and directly inhibits the NE phenotype characterized by the production of bioactive hormones and peptides like CgA and 5-HT. Similarly, RESV-induced Notch activation is associated with S phase cell cycle arrest in vitro and growth inhibition. As RESV is a nontoxic, naturally occurring compound with an established safety profile, it is an attractive target for further in vivo studies utilizing a murine model of metastatic carcinoid cancer. Based on the data presented here, RESV may additionally be an attractive target for the development of a limited phase II clinical trial in patients with unresectable carcinoid disease. Thus, this HTS method could be used to screen for other modulators of the Notch signaling pathway.
Association for Academic Surgery Award and NIH Training Grant T32CA090217 (SP). Research Scholars Grant from the American Cancer Society, NIH Grants RO1CA121115 and RO1CA109053 (HC). University of Wisconsin Medical School grant (MK).
We wish to thank Dr. Thomas Warner, Pongthep Pisarnturakit, Abram Vaccaro, and Joel Adler for their technical assistance.