Tumor-Suppressor Role of Notch3 in Medullary Thyroid Carcinoma Revealed by Genetic and Pharmacological Induction

This article is featured in Highlights of This Issue, p. 315

Medullary thyroid cancer (MTC) is a neuroendocrine tumor derived from the calcitonin-producing thyroid C cells and accounts for 3% to 5% of all thyroid cancer cases (1–3). Although MTC is relatively uncommon, it disproportionally accounts for more than 14% of thyroid cancer–related deaths. Complete surgical resection at a relatively early stage of the disease remains the only potential cure for MTC (3). Despite initial aggressive surgery, more than 50% of patients with MTC will have persistent disease, manifested as elevated postoperative calcitonin levels (4). Unlike liver metastases from other solid tumors that tend to develop as isolated and potentially resectable lesions, MTC liver metastases are almost always small and widely distributed throughout the liver, precluding curative surgical resection (5). Unfortunately, there is no curative therapy for patients with MTC liver metastases and/or widely metastatic disease (6, 7). Although newer compounds have shown some activity in clinical trials, their affect on long-term survival has not been demonstrated (8, 9). Moreover, there are no effective options to treat many of the debilitating symptoms associated with incurable MTC such as airway obstruction, flushing, abdominal pain, and diarrhea. Although we and others have previously shown that resection of selected MTC tumor foci can provide temporary palliation (5), tumor progression inevitably leads to recurrence of the disabling symptoms. Therefore, understanding the molecular pathways involved in MTC progression is critical to develop effective treatments to improve local/regional control of MTC after surgery, as well as to treat distant metastatic disease.

The biochemical diagnosis of MTC is established by elevated levels of hormones secreted by C cells that include calcitonin, chromogranin A (CgA), and synaptophysin (SYP; refs. 2, 10). In addition, the basic helix–loop–helix transcription factor, achaete-scute complex-like 1 (ASCL1) is also highly expressed in MTC cells (11, 12). Previous research has shown that ASCL1 is critical for the development of C cells and promoting MTC tumor growth (12, 13). Moreover, this transcription factor supports the growth and survival of embryologic precursors by inhibiting apoptosis (14). It has been shown that ASCL1 is regulated by the Notch pathway on the transcriptional level (15, 16) as well as by direct proteasomal degradation (17). It is apparent that MTC cell growth and neuroendocrine tumor marker expression are governed by the same signaling pathway (13, 18). The Notch signaling network is implicated in diverse functions during development, ranging from cell differentiation, cell proliferation, cell survival, and apoptosis (19, 20). Notch mammalian proteins consist of four structurally related receptors Notch1-4 that interact with one of the ligands encoded by the Delta/Jagged gene families (DLL1, DLL3, DLL4, JAG1, and JAG2; ref. 21). Ligand–receptor binding triggers pathway activation by inducing two proteolytic cleavages mediated by metalloprotease and γ-secretase, which release the Notch intracellular domain (NICD) to the nucleus. NICD forms a transcriptional activation complex with DNA-binding transcription factor CSL (also called CBF-1, RBP-jk, and Lag-1), the coactivator Mastermind-like (MAML), and other proteins (22), which subsequently induce expression of Hairy/Enhancer of Split (HES) and Hairy-related (HEY) transcriptional repressors (23). This canonical signaling cascade NICD-CBF1-HES/HEY, in turn, directly antagonizes expression of ASCL1.

In this article, we demonstrate for the first time that Notch3 pathway activation contributes to the alteration of malignant phenotype in thyroid carcinomas of neuroendocrine origin. Our results show that Notch3 protein is not expressed in human MTC tumor samples and cells. To elucidate the role of Notch3 expression, we created a gain-of-function model by generating MTC cells with a doxycycline-inducible Notch3 intracellular domain (NICD3). We further validated Notch3 antitumor properties in MTC cell lines by using novel class I histone deacetylase (HDAC) inhibitor, AB3, which specifically induces Notch3. In our previous studies, compounds closely related to AB3 were identified as potent inhibitors for HDAC2 and HDAC3 (24). We show that forced Notch3 expression in MTC cell lines does not promote and indeed inhibits tumor cell proliferation by triggering apoptotic events in a dose-dependent fashion. Importantly, Notch3 induction leads to the functional activation of Notch3 mediator CBF1, followed by changes in transcriptional levels of HES and HEY genes. Moreover, Notch3 activation causes a reduction in NET markers: ASCL1, CgA, SYP, and calcitonin, indicating that this pathway is conserved in MTC. We also verify that Notch3 activity is required to reduce the growth rate of MTC-TT xenografts. Taken together, this study documents the tumor-suppressing role of Notch3 in MTC in both in vitro and in vivo models, providing the rationale for targeting Notch3 with small-molecule compounds to treat patients with MTC and other tumors in which this pathway is not active.

Human MTC cell line, TT, was kindly provided by Dr. Barry D. Nelkin (John Hopkins University, Baltimore, MD) in 2011 and MZ-CRC-1 cell line was kindly provided by Dr. Gilbert Cote (MD Anderson Cancer Center, Houston, TX) in 2012. The control cell lines MIA-PaCa-2 and OVCAR-3 were obtained from the American Type Culture Collection (ATCC) in 2010 and 2009, respectively. Nontumorigenic human thyroid epithelial cell lines, HTori-3 and Nthy-ori 3-1, were purchased from Sigma-Aldrich (partnership with the European Collection of Cell Cultures—ECACC) in 2011. The identity of cell lines was confirmed by short-tandem repeat (STR) profile testing and the genotype of the cell lines is available in the ATCC STR database and ECACC. TT cells were maintained in RPMI-1640 medium (Life Technologies) supplemented with 16% fetal bovine serum (Sigma) and MZ-CRC-1 cells were maintained in DMEM/F-12 medium (Life Technologies) supplemented with 10% fetal bovine serum (Sigma). Both media were supplemented with 100 IU/mL penicillin (Invitrogen) and 100 μg/mL streptomycin (Invitrogen) in a humidified atmosphere of 5% CO₂ in air at 37°C (25). Doxycycline-inducible cell lines, TT-TRE NICD3, and TT-TRE (vector alone), were maintained in similar media to TT cells, except with tetracycline-free fetal bovine serum (Clontech), 75 μg/mL G418 (HyClone), and 50 μg/mL hygromycin (Invitrogen).

Human MTC tumor samples were obtained from Dr. Jeffrey Moley (Washington University, St. Louis, MO) and other control tumor samples were obtained from the University of Wisconsin Comprehensive Cancer Center Translational Science BioBank with known specimen pathology statuses. All tumor samples were snap-frozen in liquid nitrogen and stored in −80°C. Tumor cell lysates were prepared for Western blot analysis as described below.

The HDAC-Glo I/II Assay Kit (G6420) was provided by Promega Corporation. Human recombinant C-ter-GST-HDAC1 (H83-30G) and C-ter-HIS-HDAC8 (H90-30H) were purchased from SignalChem. Human recombinant C-ter-HIS-HDAC2 (50002) and N-ter-GST-HDAC6 (50006) were purchased from BPS Bioscience and human recombinant HDAC3/NCOR1 complex (BML-SE515) and C-ter-HIS-HDAC10 (BML-SE559) were purchased from Enzo Life Sciences. The HDAC-Glo I/II assay was used as previously described (26) to determine IC₅₀ values. Briefly, a 15-point 3-fold serial dilution of compound AB3 was performed at a 100× concentration in 100% DMSO in a master 96-well plate. A 5-μL aliquot of this master 100×/100% DMSO titration series was added to 245 μL of HDAC-Glo I/II assay buffer to generate a 2X concentrated, 2% DMSO master intermediate titration series of compound AB3 in a 96-well plate. From this master intermediate titration series, 5 μL replicates (n = 4) were transferred to a white, low-volume, round-bottom, nonbinding surface 384-well assay plate (Corning 3673). An equal volume (5 μL) addition of the appropriate 2× concentrated human recombinant HDAC enzyme was then added in HDAC-Glo I/II assay buffer. The 10 μL of human recombinant HDAC enzyme/compound AB3 inhibitor mixes were allowed to preincubate for 20 to 30 minutes at room temperature. Following this preincubation, an equal volume (10 μL) addition of HDAC-Glo I/II final detection reagent was added for a 20 μL final assay volume per well. After a 20-minute incubation at room temperature to allow the reactions to reach steady-state, luminescence was measured on a BMG CLARIOstar (BMG LABTECH).

The plasmid containing Notch3 ICD in pcDNA 3.3 TOPO TA (Life Technologies) was obtained from Dr. Catia Giovannini (Center for Applied Biomedical Research and Departments of Internal Medicine Gastroenterology, University of Bologna, Bologna, Italy). The Notch3 ICD 2.042-kb fragment was subcloned into the pRevTRE vector (Clontech) at the ClaI/BamHI sites. To create inducible TT-TRE NICD3 and TT-TRE cell lines, TT cells were transfected with regulatory plasmid pReVTet-On (Clontech) and selected in medium containing 75 μg/mL G418 (HyClone). The resulting G418-resistant, TT-Tet-on clones were transfected via Lipofectamine 2000 (Invitrogen) either with pRevTRE-Notch3 or pRevTRE plasmid to create TT-TRE NICD3 and TT-TRE cell lines, respectively. Transfected cells were selected in 50 μg/mL hygromycin (Invitrogen). Resistant TT-TRE NICD3 and TT-TRE clones were treated with doxycycline and screened for the presence of Notch3 protein by Western blot analysis.

Cells were plated in quadruplicate on 24-well plates at a density consistent with exponential growth and incubated overnight. The following day TT-TRE and TT-TRE NICD3 cells were treated with doxycycline (0, 0.2, 0.5, and 1 μg/mL)) and incubated for up to 6 days with subsequent treatments on days 2 and 4. TT and MZ-CRC-1 cells were treated with AB3 (0, 0.25, 0.5, 1, 1.5, and 2 μmol/L) and were incubated for up to 8 days with subsequent treatments on days 2, 4, and 6. Viable cell numbers were determined by a 3-(4,5-Dimethylthiazole-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay (Sigma) after 2, 4, and 6 days of treatment. The assay was performed by aspirating the treatment media and adding 250 μL of serum-free media containing 0.5 mg/mL of MTT (Sigma) to each well and incubating at 37°C for 4 hours. After 4 hours, 750 μL DMSO (Fisher Scientific) was added to each well and mixed. The wells were analyzed at 540 nm using a spectrophotometer (μQuant; Bio-Tek Instruments).

TT-TRE and TT-TRE NICD3 cells, after 96-hour doxycycline treatment (0, 0.2, 0.5, or 1 μg/mL), and TT and MZ-CRC-1 cells, after 48-hour AB3 treatment (0, 0.25, 0.5, 1, 1.5, or 2 μmol/L; 1 × 10⁶), were harvested by trypsinization, and processed as previously described (25), followed by DNA content analysis using a FACScalibur (BD Biosciences) and data collection using ModFit (Verity SoftwareHouse). For sub-G₁ population analysis, the floating cells were recovered from the supernatant. To quantify apoptosis, cells (2 × 10⁵) were treated with media or 0.2, 0.5, or 1 μg/mL doxycycline for 96 hours and harvested using nonenzymatic cell dissociation solution, Cellstripper (Cellgro). Both floating and adherent cells were collected, stained with PE Annexin V and 7-AAD fluorescein solutions (BD Pharmingen) for 15 minutes at room temperature in the dark, and flow cytometry was performed as described previously (27). Data were analyzed using FlowJo V5.0 (TreeStar, Inc.).

Protein lysates were harvested from cells according to previously described protocol (28). Denatured cellular extracts were resolved by SDS-PAGE (Invitrogen), transferred onto nitrocellulose membranes (Bio-Rad Laboratories), blocked in milk (1× PBS, 5% dry skim milk, and 0.05% Tween-20) or bovine serum albumin (1× PBS and 0.05% BSA; Sigma) for 1 hour, and primary antibodies were applied for overnight incubation. The primary and secondary antibodies and their dilutions are provided in Supplementary Tables S1 and S2. Following secondary antibody incubation, proteins were visualized as previously described (18, 25, 27).

Tumor tissue (2 mm³) was pulverized in the CryoPrep tissue homogenizer (Covaris) and the tissue powder was used for protein lysates preparation as described previously (29). Briefly, the tissue powder was dissolved in 500 μL of lysis buffer containing 50 mmol/L Tris, pH 7.5, 150 mmol/L NaCl, 1% Igepal CA-630, 0.1% sodium dodecyl sulfate, 0.1 μmol/L phenylmethylsulfonyl fluoride, 5 mmol/L ethylene diaminetetraacetic acid, and 12 μL/mL Protease Inhibitor Cocktail (Sigma); incubated on ice for 45 minutes; and centrifuged at 13,000 rpm for 30 minutes at 4°C. The supernatants were collected, and protein concentration was determined by the bicinchoninic acid protein assay kit (Pierce). Western blot analysis for Notch3, ASCL1, and CgA expression was performed as described above.

The mRNA expression levels of the extracellular region of Notch3 and Notch3 target genes HES1, HES2, HES5, HES6, HEY1, and HEY2 were measured by quantitative real-time PCR (qRT-PCR). RNA was isolated and transcribed into cDNA as published before (25). qRT-PCR reaction was performed using CFX Thermal Cycler and SsoFast EvaGreen labeling system (Bio-Rad) at conditions described earlier (18) on three biologic replicates. The primer sequences are provided in Supplementary Table S3. The cycle numbers obtained at the log-linear phase of the reactions for target genes were normalized to housekeeping gene s27 from the same sample measured concurrently. Finally, the expression ratios were calculated using the comparative cycle threshold (ΔC_t) method and presented as average ± standard error of the mean (SEM).

Notch3 functional activity was measured, by the degree of CBF1-binding, using a luciferase construct containing four CBF1-binding sites (4xCBF1-Luc). TT-TRE and TT-TRE NICD3 cell lines or TT and MZ-CRC-1 cell lines were transiently transfected with CBF1-luciferase reporter construct and then treated with 0 to 1 μg/mL doses of doxycycline or 0 to 2 μmol/L doses of AB3 for 48 hours. To normalize for transfection efficiency, 0.5 μg of cytomegalovirus β-galactosidase (CMV-β-gal) was cotransfected as described in details (13).

Small interfering RNA (siRNA) against Notch3 or nonspecific siRNA (Santa Cruz Biotechnology) were transfected into TT cells using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions. After 24-hour incubation, cells were treated with DMSO or AB3 (1.5 μmol/L) for additional 48 hours followed by mRNA isolation for qRT-PCR analyses. For the MTT cell proliferation analyses (as described above), TT cells were incubated for 24, 48, and 72 hours after AB3 treatment and viable cells were calculated at each time point. In addition, to confirm that increased proliferation is due to Notch3 silencing, qRT-PCR was performed to detect Notch3 expression at each time point.

Four-week-old male athymic nude mice were obtained from Charles Rivers Laboratories. They were allowed to acclimate 1 week in the animal facility to reduce stress after arrival. Mice were maintained under specific pathogen-free conditions. TT-TRE NICD3 and TT-TRE cell lines xenograft tumors were established by implanting 10⁷ cells in 200 μL of Hanks Balanced Salt Solution (Mediatech) subcutaneously into the left flank of 5-week-old mice. Two weeks after inoculation, mice with palpable tumors were randomized into two control groups (TT-TRE NICD3; n = 5 and TT-TRE; n = 6) receiving a standard diet of irradiated feed and two treated groups (TT-TRE NICD3; n = 5 and TT-TRE; n = 6) receiving a doxycycline diet of irradiated feed containing 625 mg of doxycycline per 1 kg of feed (Harlan-Teklad; ref. 30). Treatments continued for 36 days. Tumor volumes were measured by external caliper every 4 days and then were calculated by the modified ellipsoidal formula: tumor volume = ½ (length × width²). At the end of the experiment, mice were sacrificed and the tumors were dissected from the surrounding tissues and flash-frozen in liquid nitrogen for storage in −80°C. Postmortem examination of the lungs, liver, kidneys, and spleen were performed to confirm that there was no evidence of metastases or tumor growth outside of the inoculation site. All experimental procedures were done in compliance with our animal care protocol approved by the University of Wisconsin-Madison (Madison, WI) Animal Care and Use Committee.

Data are expressed as mean ± SE. Statistical analyses were done by two-way ANOVA followed by a protected Fisher LSD for post hoc comparisons or by the Student t test to compare two groups. All analyses were conducted using SAS 9.3 (SAS Institute) software. A P < 0.05 was considered statistically significant in two-tailed statistical tests.

We examined Notch3 expression in human cancer cells lines and tumor samples. Using Western blot analysis, we demonstrated that human MTC TT and MZ-CRC-1 cell lines lacked the full length and NICD3 while nontumorigenic human thyroid epithelial cell lines HTori-3 and Nthy-ori 3-1 and pancreatic cancer MIA-PaCa-2 and ovarian cancer OVCAR-3 cell lines had high Notch3 levels (Fig. 1A). Similarly, human MTC tumors from patients with sporadic and inherited MTC also lacked both portions of Notch3 while human ovarian and colorectal cancer specimens had detectable levels of full length and cleaved Notch3 (Fig. 1B). These results suggest that lack of NICD3 in MTC cell lines and MTC cancer specimens is due to impaired expression of full-length Notch3. The neuroendocrine origin of the MTC cell lines and tumor samples was verified by high levels of ASCL1 and CgA.

To determine the effect of restoration of Notch3 in human MTC cells, we stably transfected MTC TT cells with a doxycycline-inducible NICD3 (active Notch3) construct creating TT-TRE NICD3 cells. As shown in Fig. 1C, in the absence of doxycycline, there is no NICD3 protein. However, increasing doses of doxycycline in TT-TRE NICD3 led to a dose-dependent corresponding increase in NICD3 protein. Transfection with the empty vector construct (TT-TRE) lacked NICD3 at baseline and also with doxycycline treatment. To show that the expressed NICD3 protein was functional, TT-TRE and TT-TRE NICD3 cells were transiently transfected with a CBF-1 luciferase reporter construct. CBF-1 binding is critical for Notch signaling. As shown in Fig. 1D, while doxycycline treatment in TT-TRE cells slightly reduced luciferase activity, doxycycline-mediated induction of NICD3 in TT-TRE NICD3 cells resulted in a marked dose-dependent increase in luciferase activity, indicating increasing CBF-1 binding and functional NICD3.

We then confirmed whether the other components of Notch3 signaling were intact in human MTC cells. As previously mentioned, functional downstream targets of Notch3 include HES and HEY proteins. NICD3 has been shown to upregulate HES1, HES5, HEY1, and HEY2 while inhibiting HES2 and HES6 (31, 32). As predicted, doxycycline induction of NICD3 in TT-TRE NICD3 cells resulted in increases in HES1, HES5, HEY1, and HEY2 mRNA and decreases in HES2 and HES6 (Fig. 1E). These changes were not observed in the TT-TRE cells. Thus, these data demonstrate that while Notch3 is absent in human MTCs, restoration of functional Notch3 protein (NICD3) leads to induction of the Notch3 signaling pathway.

Notch3 is known to downregulate ASCL1, which is overexpressed in a variety of neuroendocrine cancers (33). Treatment of TT-TRE NICD3 cells with doxycycline resulted in a dose-dependent decrease in ASCL1 protein (Fig. 2A). We also examined the effect of Notch3 on other neuroendocrine tumor markers such as CgA, calcitonin, and SYP. As shown in Fig. 2A, overexpression of NICD3 led to marked reductions in these neuroendocrine markers as well. To determine whether the downregulation of these neuroendocrine peptides/hormones was dependent on the continuing presence of NICD3, we performed withdrawal experiments. After 2 days of doxycycline treatment in TT-TRE NICD3 cells, the cells were placed in doxycycline-free media. As shown in Fig. 2B, 2 days after doxycycline withdrawal, the NICD3 disappeared. The absence of NICD3 eventually resulted in a return of ASCL1 and CgA to baseline levels after 6 additional days. Therefore, the neuroendocrine phenotype of MTC is tightly controlled by Notch3.

To determine the effect of Notch3 on MTC cellular proliferation, the MTT assays were performed on TT-TRE NICD3 (Fig. 2C) and TT-TRE (Fig. 2D) cells with varying dosages of doxycycline. Increasing levels of NICD3, produced by doxycycline treatment of TT-TRE NICD3 cells, results in a dose-dependent inhibition of MTC cellular proliferation (Fig. 2C and E). In fact, the highest levels of doxycycline-induced NICD3 expression led to complete cessation of tumor cell growth. Doxycycline had no effect on TT-TRE cell proliferation (Fig. 2D).

To delineate the mechanism of Notch3-mediated MTC growth inhibition, we performed the flow cytometry experiments. As shown in Figs. 3A and B, conditional induction of NICD3 in TT-TRE NICD3 cells resulted in dose-dependent accumulation of sub-G₁ DNA that was not observed in TT-TRE control cells. The percentage of preapoptoic versus apopotoic MTC cells was quantified using PE Annexin V/7AAD staining (Fig. 3C and D). High levels of apopotosis were observed with NICD3 induction in MTC cells. In addition, apoptosis was verified by Western blot analysis as shown in Fig. 3E. Induction of NICD3 by doxycycline treatment in TT-TRE NICD3 cells led to increases in cleaved PARP and p27 with corresponding reductions in Mcl-1, p21, and cyclin D1.

We have shown that Notch3 tightly regulates MTC phenotype and proliferation. In this system, high levels of active Notch3 (NICD3) can be generated. We then investigated whether the rapid increase in NICD3 levels could also be due to auto-induction of endogenous Notch3 protein. We therefore measured Notch3 mRNA levels after NICD3 induction in TT-TRE NICD3 cells using primers in the Notch3 extracellular region (outside NICD3, described in Supplementary Fig. S1A). As shown in Supplementary Fig. S1B, mRNA levels of the Notch3 extracellular domain were similar in TT-TRE cells treated with doxycycline and vehicle, and in TT-TRE NICD3 cells treated with vehicle. However, TT-TRE NICD3 cells treated with doxycycline had high levels of Notch3 mRNA extracellular region. This induction of endogenous Notch3 mRNA results in high levels of the Notch3 full-length receptor protein (Supplementary Fig. S1C). Thus, Notch3 appears to have the capacity to autoregulate its own expression.

Using a MTC xenograft model, we studied the effect of NICD3 induction in vivo. MTC xenografts were established in nude mice using TT-TRE and TT-TRE NICD3 cells. MTC tumors from TT-TRE cells treated with vehicle and doxycycline had robust proliferation (Fig. 4A). Similarly, tumors from TT-TRE NICD3 cells treated with vehicle also continued to grow in this model. However, in sharp contrast, NICD3 induction in TT-TRE NICD3 cells with doxycycline treatment markedly inhibited MTC tumor proliferation (Fig. 4A). As expected and similar to the in vitro conditions, induction of NICD3 in MTC tumors resulted in downregulation of the neuroendocrine marker ASCL1 and an increase in apoptosis as measured by cleaved PARP (Fig. 4B).

Having demonstrated that induction of Notch3 in MTC cells inhibits cancer cell proliferation both in vitro and in vivo, we were interested in the concept of activating Notch3 as a therapeutic target. Furthermore, because Notch3 also suppresses neuroendocrine peptide/hormone levels, this strategy could be used to palliate patients with endocrinopathies that are commonly associated with this disease. Thus, we sought to identify a small molecule that induces Notch3 in MTC cells. We had previously developed a high-throughput screen for Notch-activating compounds (34). Using this system, we identified a synthetic compound, AB3, which appeared to have this property. AB3 is a HDAC inhibitor (Fig. 5A). It was prepared by condensing commercially available para-methoxybenzaldehyde with a known bifunctional reagent, which has a six methylene carbon tether between a hydrazide and a hydroxamic acid (24). As shown in Fig. 5B, exposure of TT cells to AB3 led to a dose-dependent induction in Notch3 mRNA. This induction was specific to Notch3, as AB3 did not affect mRNA levels of Notch1 or Notch2. To determine whether AB3 induces both full length of Notch3 and NICD3 in MZ-CRC-1 and TT cell lines, qRT-PCR was used with primers that recognize the extracellular region of Notch3, NECD3 (Supplementary Fig. S2A and S2B), or primers specific for the intracellular domain, NICD3 (Supplementary Fig. S2C and S2D). We revealed that AB3 induces both the full-length Notch3 and NICD3 to significantly higher levels than control-treated cells. This increase in Notch3 led to an induction of functional NICD3 in both MTC TT and MZ-CRC-1 cells as demonstrated by elevated CBF-1–binding activity (Fig. 5C and D). To further validate whether AB3 acts as a functional Notch3-activating compound in MTC cell lines, we performed Western blot analysis to detect the NICD3 protein (Fig. 5E and F).

Similar to NICD3 induction in our doxycycline-inducible model, treatment of MTC cells by AB3 also led to downregulation of neuroendocrine tumor markers ASCL1 and CgA (Fig. 5G and H). We then investigated the effects of AB3 on MTC cellular proliferation. AB3 suppressed growth in both MTC TT (Fig. 6A) and MZ-CRC-1 (Fig. 6B) cell lines in a dose-dependent manner. As with genetic induction of NICD3, AB3 treatment also caused apoptosis in TT and MZ-CRC-1 cells as demonstrated by cleaved PARP, reduction of XIAP, survivin, p21, and overexpression of p27 through Western blot analysis (Fig. 6C), as well as by sub-G₁ DNA accumulation with flow cytometry experiments (Fig. 6D–F).

To assess the inhibitory activity and selectivity of AB3 to specific HDAC isoenzymes, we used the HDAC-Glo I/II luminescent assay that measures the relative activity of HDAC class I and IIb enzymes from purified enzymes, cells, or cell extracts. Using this assay, first we determined the optimal concentration of purified human recombinant HDAC's to use per reaction. The optimal concentrations of the listed HDAC isoenzymes (Fig. 7A) were in the lower end of the linear portion of the enzyme titration assay and concentrations were also chosen so all HDAC isoforms gave similar luminescent signal (RLU) output (enzyme titration not shown). Next, we determined the IC₅₀ of AB3 against the indicated purified human recombinant HDAC isoforms and showed that AB3 appears to be HDAC1, -2, and -3 selective with less inhibitory potency against the HDAC6, -8, and -10 isoforms (Fig. 7A and Table 1). Furthermore, using the HDAC-Glo I/II luminescent assay, we demonstrated the inhibition of HDAC activity in the MZ-CRC-1 and TT cell lines using AB3 as the HDAC inhibitor. This experiment revealed that AB3 reaches its maximal inhibitory activity in a range of 1 to 10 μmol/L in both MTC cell lines (Fig. 7B and C). Interestingly, the dose of 2 μmol/L of AB3 exerted the highest induction of Notch3 expression in MZ-CRC-1 and TT cell lines.

To confirm that the phenotypic effect of AB3 (i.e., inhibition of ASCL1 and CgA protein expression and reduction of MTC cell proliferation) is a result of activation of Notch3 signaling, siRNA targeting Notch3 were used to silence the gene expression. MTC TT cells were transiently transfected with Notch3 or nonspecific siRNA and 24 hours later treated with vehicle or AB3. Using qRT-PCR analysis, we demonstrated that Notch3-targeted siRNA were capable of significantly reducing AB3-activated Notch3 (Fig. 8A) and downstream HES1 message levels (Fig. 8B) following AB3 treatment. Moreover, blockade of AB3-mediated Notch 3 activation reversed the AB3-induced changes in ASCL1 and CgA expression (Fig. 8C). Finally, the abrogation of AB3-mediated Notch3 induction by Notch3 siRNA resulted in increased proliferation of TT cells (Fig. 8D). These data suggest that phenotypic regulation with AB3 treatment was primarily mediated by Notch3 signaling.

Although intracellular Notch signal transduction seems to be very simple, in human cancer cells the Notch pathway exerts pleiotropic functions. So far, Notch1 is the best characterized isoform with dual action potency which, depending upon the cellular context, may act as a tumor suppressor or as an oncogene. The first evidence for the oncogenic role of Notch1 came from human T-cell neoplasia (35). Later, it was shown that Notch1 was upregulated in various solid tumors, including breast cancer (36), medulloblastoma (37), colorectal cancer (38), and non–small cell lung carcinoma (NSCLC; ref. 39). The contribution of Notch1 to tumorigenesis in these cancers is mainly through maintaining cells in the proliferation stage by inhibiting apoptosis. Conversely, Notch1 signaling is very minimal or absent in neuroendocrine cancers such as small cell lung cancer (SCLC), pancreatic carcinoid, and MTC, where reinstitution of Notch1 by either pharmacologic or genetic approach inhibits the tumorigenic process (13, 40, 41). In addition, in the skin, Notch1 loss of function results in spontaneous basal cell carcinomas (42). These apparent, but paradoxical, functions clearly indicate that the role of Notch1 signaling is dependent on its cellular context.

Although the dual role of Notch1 in human cancers has been extensively studied, there are far less data regarding Notch3. Studies to date have mainly described the role of Notch as an oncogene. Notch3 signaling has been implicated as an oncogene in ovarian serous carcinomas where amplification of the Notch3 genomic locus is critical for cellular survival and growth. In addition, translocation of the Notch3 gene occurs in a subset of NSCLC (43), and constitutively expressed Notch3 induces neoplastic transformation in the breast, brain, colon, and hematopoietic tissues (44–48). Moreover, using an RNAi approach, Notch3, but not Notch1, was found to be critical in maintaining cellular proliferation of ErbB2-negative breast cancers (46).

In contrast, there are little data regarding the tumor-suppressor role of Notch3 in human malignancies. Demehri and colleagues (42) showed that mice lacking Notch3 in the skin developed skin-barrier defects, epidermal hyperplasia, and skin tumors. More recently, it has been shown that upregulation of Notch3 is sufficient to activate senescence and inhibit proliferation in breast, melanoma, and glioblastoma cell lines (49). Thus, similar to Notch1, Notch3 may also have a dual role in cancer as an oncogene or tumor suppressor depending on the cellular context. To date, the role of Notch3 in endocrine malignancies, such as thyroid cancer and neuroendocrine cancers, has not been explored.

In this article, we demonstrate for the first time that Notch3 acts as a tumor suppressor in human MTC. Notch3 protein is not expressed in human MTC tumor samples and cell lines but detectable in nontumorigenic human thyroid epithelial cell lines. To restore normal thyroid phenotype for Notch3 expression and elucidate its role, we created a gain-of-function model by generating MTC cells with a doxycycline-inducible NICD3. We show that forced NICD3 expression in MTC does not promote but indeed inhibits tumor cell proliferation by triggering apoptotic events in a dose-dependent fashion. Importantly, NICD3 induction leads to the functional activation of the Notch3 mediator CBF1, followed by changes in transcriptional levels of HES and HEY genes. Moreover, Notch3 activation causes a reduction of NET markers ASCL1, CgA, calcitonin, and SYP, indicating that this pathway is conserved in MTC. We also verify that Notch3 activity is required in vivo to reduce the growth rate of MTC xenografts.

MTC is derived from the neuroendocrine C cells of the thyroid and is one of the more aggressive subtypes of thyroid cancer. Like most thyroid cancers, surgical resection is the predominant treatment modality and can be curative in selected patients. However, unlike well-differentiated thyroid cancer that can be treated with surgery and/or radioactive iodine even when metastases develop, patients with MTC have limited options. For decades, there were no approved drugs to treat patients with metastatic MTC. Recently, the FDA has approved two compounds for metastatic MTC (9). Although both of these drugs were shown in clinical trials to prolong recurrence-free survival, their effect on long-term survival is unknown.

Currently, the approved therapies for MTC have targeted the inhibition of RET signaling. This concept is based on the important fact that mutations in the RET proto-oncogene have been shown to cause inherited MTC in multiple endocrine neoplasia (MEN) type 2A, MEN2B, and familial MTC (9, 10). In fact, genetic testing for RET mutations in children from families with inherited MTC, and age-appropriate prophylactic removal of the thyroid gland, has been shown to be curative in these patients (50). However, in patients with inherited MTC diagnosed later in life after the MTC has metastasized, surgery is not curative, and RET inhibitors are the mainstay of therapy. In addition, the majority of patients with sporadic MTC also have somatic, activating mutations in RET (9, 10, 50). Therefore, patients with metastatic inherited and sporadic MTC are currently treated with targeted therapies against RET. However, although tyrosine kinase inhibitors that block RET function are very effective in the laboratory, they do not appear to have any effect on overall patient survival in several clinical trials. Therefore, there is a clear need for alternative options that target other pathways important for MTC progression. Although most research in MTC has focused on RET, emerging data suggest that other pathways, such as CDK5 and Notch, may play an essential role in MTC development and progression (13, 18, 40, 51).

In this article, we identify a small molecule that can induce Notch3 signaling in MTC cells. To our knowledge, this is the first description of a compound that specifically activates Notch3 signaling. We demonstrate that AB3 induces Notch3 signaling, inhibits MTC proliferation, and suppresses neuroendocrine tumor markers. These effects are achieved at low micromolar drug doses. AB3 would be a candidate for in vivo models and preliminary clinical studies identifying safe and appropriate dosing parameters and pharmacokinetics as an initial phase to further clinical investigation. It is possible that AB3 could be used in conjunction with tyrosine kinase inhibitors, and this combination is currently under investigation in our laboratory.

In summary, this study documents the tumor-suppressing role of Notch3 in MTC in both in vitro and in vivo models, providing the rationale for targeting Notch3 with small-molecule compounds to treat patients with MTC and other tumors in which this pathway is not active.

No potential conflicts of interest were disclosed.

Conception and design: R. Jaskula-Sztul, W. Tang, H. Chen

Development of methodology: R. Jaskula-Sztul, K.R. Kupcho, W. Tang, H. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Jaskula-Sztul, J. Eide, S. Tesfazghi, A. Dammalapati, A.D. Harrison, C. Scheinebeck, K.R. Kupcho, A.K. Hundal

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Jaskula-Sztul, J. Eide, K.R. Kupcho

Writing, review, and/or revision of the manuscript: R. Jaskula-Sztul, J. Eide, S. Tesfazghi, A. Dammalapati, A.D. Harrison, X.-M. Yu, K.R. Kupcho, A.K. Hundal, W. Tang, H. Chen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Tesfazghi, A. Dammalapati, A.D. Harrison, X.-M. Yu, M.B. Robers, A.K. Hundal, H. Chen

Study supervision: H. Chen

Other (chemical synthesis): G. Winston-McPherson

The authors thank Dr. Muthusamy Kunnimalaiyaan for his critical suggestions during the study, Yash Somnay for the composition, and Heather Hardin for the valuable suggestions for culturing nontumorigenic thyroid epithelial cell lines.

This work was supported by grants NIH (R01 CA121115 to H. Chen), American Cancer Society (MEN2 Thyroid Cancer Professorship 120319-RPM-11-080-01-TBG to H. Chen and Research Scholar Award RSGM TBE-121413 to H. Chen), and Layton F. Rikkers, MD, Chair in Surgical Leadership Professorship (H. Chen).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.