Pharmacological Disruption of the Notch1 Transcriptional Complex Inhibits Tumor Growth by Selectively Targeting Cancer Stem Cells

Notch transcription is mediated by the formation of a multiprotein complex termed the Notch Ternary Complex (NTC; refs. 1–4). Notch signaling is initiated at the cell membrane through contact between neighboring cells and results in the cleavage and release of the intracellular domain of the Notch receptor (Notch^ICD; ref. 5). There are 4 Notch receptors expressed in humans (Notch1–4; ref. 5). For all Notch paralogs, the Notch^ICD translocates to the nucleus and binds CSL, followed by the recruitment of Mastermind that binds an interface formed by Notch and CSL to form the NTC (4, 6–8). The NTC subsequently recruits additional coactivators through the MAML C-terminal domain and drives transcription of target genes, including HES, HEY, and CCND1 (5–7). Therefore, approaches that disrupt NTC assembly will inhibit Notch^ICD-directed transcription, and because Notch signaling is uniquely stoichiometric, in contrast with other signal-amplifying cascades, and because transcription is extremely sensitive to even low levels of Notch^ICD, disruption of this critical node can significantly inhibit downstream effectors of Notch signaling (9–13).

Notch signaling is essential to drive the neoplastic progression of many cancers, including breast, prostate, esophageal adenocarcinoma (EAC), glioblastoma, T-cell acute lymphoblastic leukemia (T-ALL), and others (14–19). Because of this broad dependence, disruption of Notch-driven transcription has long been an attractive target in oncology. Inhibiting Notch signaling is challenging due to the paucity of druggable targets, as well as the important tissue-specific pleiotropic roles Notch signaling plays in the homeostasis of a number of adult tissues (20–25). Gamma secretase inhibitors (GSI), which prevent the cleavage of Notch receptors and release of Notch^ICD (16), have been extensively tested in the clinic for over a decade, originally against Alzheimer's Disease and then against cancer (26–30). However, GSI's are not selective for individual Notch paralogs, nor for Notch signaling in general as gamma secretase has >75 cleavage targets (31). Consequently, GSI's have been clinically hampered by toxicities driven by the essential role of Notch signaling in the intestines, primarily orchestrated by Notch1 and Notch2, and most GSI's are no longer under clinical development (32–36). Specifically, high-level inhibition of multiple Notch paralogs significantly disrupts the expression of the HES family of transcription factors, which cooperatively control cell fate determination in intestinal crypts, leading to differentiation of crypt precursor cells toward secretory lineages such as goblet cells (20–22, 37). This disruption unbalances the secretory/absorptive homeostasis of the intestines and is responsible for treatment emergent GI toxicity.

Approaches with greater Notch paralog selectivity have been shown to avoid severe GI toxicity in mice (38). Antibodies to the Notch1 and Notch2 receptors individually avoid intestinal toxicity over 12 days, but cotreatment results in goblet cell metaplasia, underlining the need for paralog selectivity. However, Notch1 mAbs failed in phase Ib clinical trials after proving intolerable in patients with colorectal cancer cotreated with trifluridine/tipiracil (39). Thus, there exists a clear unmet need for selective Notch therapeutics that can maintain a clear therapeutic window between targeting individual paralogs required for oncogenic growth while avoiding toxicity in the intestines and other Notch-regulated tissues.

Deregulation of the Notch signaling pathway plays a role in the maintenance and survival of cancer stem cells (CSC) that likely underlie resistance to chemotherapy, making it an exceedingly attractive target for anti-CSC therapies (40). Current approaches to inhibit the Notch signaling pathway include small molecules that target proteins upstream of the NTC. In spite of the critical role played by the NTC in cancer (40, 41), there are no small-molecule inhibitors in the clinic that successfully target the NTC and selectively inhibit stem-like Notch-driven tumors. Therefore, the full range of potential targets in the pathway has not been exploited.

We discovered the first potent, Notch1-selective inhibitor that dramatically attenuates Notch-dependent tumor cell growth in a variety of cancers, termed NADI-351. NADI-351 acts by selectively ablating the tumoral CSC population by inhibiting Notch1-directed transcription.

The OE33 human EAC cell line was obtained from the European Collection of Cell Culture. MDA-MB-231 (human triple-negative breast cancer) was obtained from Dr. Caroline Briegel at the University of Miami, Miller School of Medicine (Miami, FL). Het1A (human immortalized esophageal epithelial cell line), MCF10A (human breast epithelial cells), PC-3 (human prostate cancer), MCF-7, and T47D (ER⁺ luminal breast cancer) were obtained from the ATCC. All cell lines were tested monthly for Mycoplasma contamination and propagated in growth media as specified by the provider (BI, 20–700–20). Cell lines, excluding OE19 were obtained between 2017 and 2019 and authenticated by the ATCC [cell line authentication profiling using short tandem repeat (STR) profiling]. The OE19 cell line was obtained from the Leibniz Institute DSMZ in 2020, authenticated by STR profiling, and monitored monthly for Mycoplasma contamination. Number of passages of cells between collection or thawing and use in the described experiments were between P2 and P10.

Recombinant proteins were expressed using baculovirus expression vectors in SF21 cells and purified as described previously (4). Notch complex assembly experiments were performed as described previously (40). Data analysis was performed using GraphPad Prism software (Version 8.4.3).

Cells (1 × 10⁵) were seeded in cell culture dishes (100 mm) in RPMI-1640 or DMEM medium containing 10% FCS with antibiotics. After 24 hours of incubation, cells were exposed to NADI-351 in a time-dependent manner. At the end of the treatment, total mRNA was isolated, and cDNA was synthesized according to the manufacturer's protocol (4368814; Life Technologies). RT-qPCR analysis was performed using TaqMan probes according to the manufacturer's instructions (4324018; Applied Biosystems). Gene expression was normalized to HPRT or TBP gene. Data analysis was performed using GraphPad Prism software (Version 8.4.3).

293A cells expressing (i) pCMV-Tet-On 3G, and (ii) pCSL-RElement-Luc and one of the following: (i) pLV[Tet]-Puro-TRE3G>Notch1^ICD, or (ii) pLV[Tet]-Puro-TRE3G>Notch2^ICD, or (iii) pLV[Tet]-Puro-TRE3G>Notch3^ICD, respectively, were used. When doxycycline is added, the Tet-ON gene activates expression of hNotch1^ICD, 2^ICD, or 3^ICD, which together with endogenous NTC components binds to CSL responsive elements (pCSL-RElement-Luc) and expresses luciferase. A total of 1 × 10⁴ cells are plated in 100 μL (96-well format). 24 hours later, compounds (10 mmol/L stock in DMSO) are diluted in DMSO to ×200, then added (5 μL into 1 mL) to cell culture media containing 50 ng/mL Dox. This is then added 1:1 to cells. Final DMSO = 0.25%, Dox = 25 ng/mL. After 24 hours, media are discarded, and cells are lysed in passive lysis buffer (Promega). Cells are rocked at room temperature for 15 minutes and then lysate is divided for luciferase assay (Luciferase Assay System, Promega) and cell viability (CellTiter Glo 2.0, Promega). Raw luciferase is normalized to cell viability, and then scaled to DMSO wells. Results are analyzed and IC₅₀'s are determined in GraphPad by nonlinear regression curve fitting (4 parameter) of dose–response curves.

DNA pulldown was performed as previously described (42). Streptavidin agarose beads (Pierce) were incubated with previously annealed 47 mer biotinylated dsDNA containing two high-affinity CSL-binding sites facing forward (2× CSL binding DNA) or 2 mutated CSL-binding sites (43).

MDA-MB-231 cells, OE33 cells, and PC-3 cells were cross-linked with 1% formaldehyde for 10 minutes at 37°C and cross-linking was quenched by adding glycine to a final concentration of 0.125 mol/L. Nuclear pellet was prepared using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermofisher). Nuclear pellet was digested using Micrococcal Nuclease (NEB) to yield chromatin fragments of approximately 300 to 800 bp and lysed in chromatin immunoprecipitation (ChIP) buffer (1% Triton, 2 mmol/L EDTA, 150 mmol/L NaCl, and 20 mmol/L Tris-HCl pH 8 and 0.01% SDS) containing a protease inhibitor cocktail (Roche). Lysates were immunoprecipitated with α-Notch1, α-Notch2 (Bethyl Laboratories), α-Notch3, α-Notch4 and α-Maml1 (CST) and α-IgG (Abcam) antibodies and were reverse cross-linked overnight at 65°C in 200 mmol/L NaCl and proteinase K. DNA was purified using the PCR Purification Kit (Qiagen) and HES1 promoter were amplified by qPCR. Primer sequences:

• Forward 5′: CGTGTCTCCTCCTCCCATT

• Reverse 5′: GGGGGATTCCGCTGTTAT

Nuclear protein analysis was performed as previously described (44) using anti-Notch1^val1744 (Cell Signaling Technology, 4147S), anti-Notch2 (Cell Signaling Technology, D76A6), anti-Notch3 (Cell Signaling Technology, D11B8), anti-Notch4 (Cell Signaling Technology, L5C5), anti-MAML1 (Cell Signaling Technology, D3K7B), anti-MAML2 (Cell Signaling Technology, D41E6), anti-MAML3 (Bethyl Laboratories, A300–6841), and anti-RBPSUH (CSL, Cell Signaling Technology, D104A).

PDX cancer models and xenografts were established as described previously (45). We used the EAC74 PDX from our library of surgically resected patient tumors to evaluate the effect of NADI-351 on tumor growth. Six-week-old nude female mice were purchased from Charles River Laboratories. When the tumor size reached 200 mm³, the mice were treated with daily dose of 20 and 30 mg/kg for intraperitoneal injection and 40 mg/kg for oral gavage, 6 mice per treatment/6 mice per negative control. The Vehicle used for oral gavage: 3.25% NMP, 50% PEG-400, 46.75% normal saline. The Vehicle used for intraperitoneal: DMSO. The stop/end point criteria are based on the days it takes for tumors to reach 2,000 mm³ according to what is approved in our animal protocol. Tumor volume was measured by the formula: volume = (S × S × L)/2. Animal experiments were reviewed and approved by the University of Miami Institutional Animal Care and Use Committee (Miami, FL).

Immunohistochemical analysis of small intestine from C57BL/6 or nude mice injected intraperitoneally with different concentrations of NADI-351 and DZB, daily for 5 or 30 days. Tissue sample preparation and PAS staining IHC were performed as previously described (22). ALDH1A and ki67 stains were performed using anti-ALDH1A (Abcam, ab52492) and anti-ki67 (Leica biosystems, PA0118). ALDH1A scoring: negative (0), weak (1+), moderate (2+), strong (3+). Ki67 scoring: Count the percentage of positive cells in one ×40 field with highest positive.

Cells were cultured at low density under treatment, and then colonies were stained with 0.01% crystal violet and counted. The cells were measured using the Cell Titer-Glo assay (G7572; Promega) for Cell Viability Assays.

To obtain tumor spheres, cells were cultured in DMEM/F12 with 2% B-27 serum-free supplement (17504–044; Invitrogen), 20 ng/mL EGF (PHG0311L; Invitrogen), and 20 ng/mL basic FGF (PHG0266; Invitrogen) for 14 days to select for CSCs and early progenitor cells. Resulting tumor spheres were examined and counted under the microscope.

Cells lines or single cells derived from tumors were stained using the ALDEFLUOR kit (Stem Cell Tech) or a specific monoclonal BB515 anti-human Notch1 antibody (BD Horizon, 564781), following the manufacturer's instructions and were analyzed by flow cytometry, as described previously (46). Singles cells were obtained from tumors following the manufacturer's instructions (MACS, Tumor Dissociation kit, 130–095–929).

Tissue sample preparation and TUNEL staining immunofluorescence were performed following the manufacturer's instructions (Promega, DeadEnd Fluorometric TUNEL System, G3250). TUNEL/DAPI staining was analyzed by Sp5 inverted confocal microscope (Leica Biosystems).

NADI-351 was assayed by Eurofins Discovery Services (CEREP) against the Safety44 panel (38 radioligand binding assays + 6 activity assays) at 10 μmol/L, (n = 2). Results were compiled and analyzed in GraphPad Prism.

NADI-351 (10 μmol/L) was assayed by Reaction Biology Corp. to determine changes in activity of 372 wild-type human kinases, (n = 2), (ATP) = 10 μmol/L. Positive control compounds (staurosporine, or alternate kinase inhibitors, depending on kinase) were used in 10-concentration dose–response assays to verify kinase activity. Results were analyzed in GraphPad Prism.

Modeling of the Notch1/CSL active site and small-molecule docking experiments were performed using MOE (Molecular Operating Environment, 2019.01; Chemical Computing Group ULC, 1010 Sherbooke Street West, Suite #910, Montreal, QC, Canada, H3A 2R7, 2020).

The P value was calculated using χ² in contingency table. Data are presented as mean ± SD and were analyzed by the two-tailed Student t test. A P value of less than 0.05 was considered significant. In all other cases, statistical significance was determined by the Student t test. The P value <0.05 was considered statistically significant.

Previously, we reported the first small-molecule inhibitor of the NTC, IMR-1 (43). This small-molecule binds to a pocket formed by Notch^ICD and CSL and prevents Mastermind from binding and forming a stable NTC. Although this demonstrated proof of principle that a small molecule could disrupt the formation of the NTC, IMR-1 is a pan-Notch inhibitor and lacked the necessary potency to advance. Therefore, a molecular modeling-focused SAR campaign was used to improve potency, Notch paralog specificity and drug-like properties and resulted in the identification of NADI-351 (Supplementary Fig. S1A). Molecular modeling of NADI-351 in the context of the NTC suggests that NADI-351 interacts with both Notch1 and CSL residues (Fig. 1A and B; Supplementary Fig. S1B). Activity of NADI-351 against an NTC AlphaScreen assay revealed that NADI-351 is 15 times more potent than IMR-1 (Fig. 1C). NADI-130, an inactive analog of NADI-351 (Supplementary Fig. S1C), displayed no activity and served as a negative control in the AlphaScreen. An inducible Notch1^ICD-responsive luciferase reporter was employed to further analyze the effect of NADI-351 on Notch transcription in cells and, consistent with the AlphaScreen, there was a significant increase in potency (NADI-351 IC₅₀ = 8.8 vs. > 40 μmol/L for IMR-1; Fig. 1D).

Our previous work demonstrated the dependence of EAC on Notch signaling (14). NADI-351 inhibited OE33 cell growth in single-dose MTT assays (EC₅₀ = 10 μmol/L) and, more potently, in multi-dose colony formation assays (EC₅₀ = 1.3 μmol/L; Fig. 1E). Similar results were observed in other reported Notch-dependent cell lines, including MDA-MB-231 cells (human triple-negative breast cancer) and PC-3 cells (human prostate cancer; Supplementary Fig. S2A and S2B). As previously reported, DAPT (a GSI) and IMR-1 inhibited OE33 colony formation and other Notch-dependent cell lines, with EC₅₀ values between 10 and 15 μmol/L (43), underscoring the higher comparative potency of NADI-351 against Notch-dependent cell growth. No significant effect was observed in Notch-independent cell lines MCF-7 and T47D (ER⁺ Luminal breast cancer) or on normal esophageal cells (Het1A) or breast epithelial cells (MCF10A; Supplementary Fig. S2C–S2F). To demonstrate that NADI-351 can disrupt the DNA-bound NTC in cells, we used a CSL-DNA affinity pulldown assay (DAP). When OE33 cells are treated with NADI-351, we observed a dose-dependent concomitant inhibition of Notch1^ICD and Maml1 binding to CSL on DNA, without a loss of CSL binding. This result demonstrates that NADI-351 inhibits Notch1-driven transcription by disruption of the NTC in cells (Fig. 1F).

Because OE33 cells only express Notch 1–3, we next sought to confirm the selectivity of NTC inhibition by NADI-351. To fully validate the proposed mechanism of action, we used the triple-negative breast cancer (TNBC) cell line MDA-MB-231, which are Notch-dependent and express Notch1–4. MDA-MB-231 cells were treated with the determined single-dose MTT EC₅₀ value (10 μmol/L, Supplementary Fig. S2A) and were harvested at several time-points for NTC DAP assays. NADI-351 selectively inhibited Notch1 and MAML1 binding to DNA-bound CSL, whereas no effect was observed in Notch 2–4 or CSL binding to DNA (Fig. 2A; Supplementary Fig. S3A). No reduction in binding of either MAML2 or MAML3 was observed, and no other NTC proteins displayed a similar kinetic effect over time or magnitude of inhibition as displayed by Notch1 and MAML1 (Fig. 2A). RT-qPCR analysis of MDA-MB-231 cells indicated that NADI-351 treatment significantly decreased transcription of Notch target genes HES1 and HES5 (Fig. 2B). Furthermore, ChIP in MDA-MB-231 cells indicated that NADI-351 selectively inhibits the recruitment of Notch1 (Fig. 2C) and MAML1 (Fig. 2D) to the HES1 promoter. We also observed that only Notch1 was inhibited from binding at the HES1 promoter with NADI-351 treatment, whereas Notch2–4 had no significant reduction in binding (Fig. 2E), thereby confirming DAP results. We then evaluated the selectivity index for Notch1 NTCs using higher concentrations of NADI-351 (1, 2, and 3X the EC₅₀) in NTC DAP assays. This analysis revealed that although cellular complexes comprised of Notch1^ICD and MAML1 were dramatically disrupted (Fig. 2F and Supplementary Fig. S3B and S3C), Notch 2–4 complexes remained largely intact, therefore demonstrating exquisite selectivity for Notch1 NTCs. Similar results were observed in PC-3 and OE33 cell lines (Supplementary Fig. S3D andS3E; Supplementary Fig. S4A–S4F). Consistent with DAP assays, selectivity for Notch1 over Notch2 and Notch3 was observed using inducible Notch1, 2, or 3-driven luciferase reporter assays (Fig. 2G).

To evaluate the effect of NADI-351 on tumor growth, we used several Notch-dependent cell lines and patient-derived xenograft models. MDA-MB-231, PC-3, and OE19 cell line-derived tumors were formed and grown in nude mice to approximately 200 mm³ before treatment. Both oral or intraperitoneal treatment routes resulted in significant inhibition of tumor growth in all xenograft models and had no significant effect on mouse weight and no observable effects on overall appearance (Fig. 3A; Supplementary Fig. S5A–S5I). We further investigated effects on the intestines; other Notch inhibitors (such as GSIs) have demonstrated toxicity (goblet cell metaplasia) due to inhibition of transcription driven by multiple Notch^ICD's (20–22). In contrast with GSI treatment (DBZ, 10 mg/kg), we found no evidence of GI toxicity after 5 daily treatments with NADI-351, even at the highest dose tested (40 mg/kg; Fig. 3B). NADI-351–treated mice displayed intact crypts and normal intestinal architecture with no evidence of goblet cell metaplasia (PAS staining), comparable with the vehicle group (DMSO). To assess the activity of NADI-351 on a more clinically relevant model, we used an EAC patient-derived xenograft model (EAC47 PDX; ref. 14). NADI-351 significantly inhibited tumor growth in the EAC model (Fig. 3C; Supplementary Fig. S5J) with no effect on weight, compared with vehicle (Supplementary Fig. S5K). We examined tumors from NADI-351–treated EAC47 PDX mice and found that apoptosis was increased, whereas the CSC marker ALDH1A and Ki67 were lower in comparison with vehicle-treated controls (Fig. 3D; Supplementary Fig. S6A–S6D). Examination of small intestines from these animals further demonstrated that 30 days of treatment (30 mg/kg) with NADI-351 produced no effect on villi architecture or proliferation and did not drive goblet cell metaplasia (Fig. 3E). To further assess potential off-target interactions, NADI-351 was assayed against a panel of 44 common targets responsible for toxicity in human trials (Supplementary Fig. S7A). No targets were bound or inhibited to a significant degree by NADI-351 (10 μmol/L). Profiling of 372 wild-type human kinases similarly showed no significant off-target activity by NADI-351 (10 μmol/L; Supplementary Fig. S7B). These results indicate a clear therapeutic window for the use of NADI-351 as a Notch inhibitor without the dose-limiting toxicity of previous approaches and hint at the mechanisms underlying in vivo activity.

It is thought that CSC populations drive tumor formation and are responsible for much of the neoplastic phenotype (40, 41). In addition, many studies have attributed progression, metastasis, and resistance to therapy to CSCs (47). Because NADI-351 had significantly increased potency against EAC and TNBC clonogenic growth compared with general proliferation assays and reduced ALDH1A in EAC47 tumors, we reasoned that NADI-351 might exert its effects primarily through ablation of CSC populations. In many tumor cells types, certain cell populations have been identified as CSCs (43). In the TNBC cell line MDA-MB-231, the primary CSC population is marked by CD44⁺/CD24^+/low whereas the CD44⁺/CD24⁻ does not possess CSC properties (48). Sorting (Supplementary Fig. S8A and S8B) and treatment of these populations revealed that NADI-351 more potently inhibited CD44⁺/CD24^+/low populations with an EC₅₀ = 2.3 vs. 10 μmol/L for CD44⁺/CD24⁻ (Supplementary Fig. S8C). Enriched CD44⁺/CD24^+/low cells had higher Notch activity than CD44⁺/CD24⁻ (24) and we found that NADI-351 inhibited the formation of primary secondary spheres derived from enriched MDA-MB-231 CD44⁺/CD24^+/low in a dose-dependent manner unlike CD44⁺/CD24⁻ spheres (Supplementary Fig. S8D and S8E).

EAC cell lines are Notch-dependent and display characteristics of CSC according to their expression of ALDH1A (14), making them a good model to study the effect of NADI-351 on CSCs. We first examined the effect of NADI-351 on colony formation of OE33 cells sorted by ALDH expression (Fig. 4A; Supplementary Fig. S9A and S9B). NADI-351 more potently inhibited ALDH⁺ OE33 colony (EC₅₀ = 1 μmol/L) compared with ALDH⁻ colonies (EC₅₀ = 3.4 μmol/L, Supplementary Fig. S9C). Further RT-qPCR analysis demonstrated that NADI-351 selectively inhibited Notch target gene expression in OE33 ALDH⁺ cells only (Fig. 4B and C). Because cell proliferation is well correlated to the regulation of cell-cycle progression, we evaluated the effect of NADI-351 on the cell cycle of ALDH⁺ OE33 cells. We observed that NADI-351 modulates an S-phase arrest in ALDH⁺ cells after 24 hours of treatment (Supplementary Fig. S9D). We also observed that ALDH⁺ populations experienced a significant, dose-dependent increase in apoptosis following NADI-351 treatment (Supplementary Fig. S9E). Together, these data indicate that cell-cycle arrest and activation of apoptosis are a direct consequence of the inhibition of Notch-dependent transcriptional regulation by NADI-351 in EAC CSC populations.

Consistent with colony formation results in ALDH-sorted OE33 populations, NADI-351 more potently inhibited EAC47 PDX ALDH⁺ colony formation and cell viability compared with ALDH⁻ colonies (Fig. 4D and E; Supplementary Fig. S10A and S10B). NADI-351 selectivity against ALDH-sorted EAC47 subpopulations was further examined in tumor spheres. Intriguingly, ALDH⁻ EAC47 cells did not form tumor spheres, suggesting that CSC activity may be crucial for tumor sphere formation. In contrast, ALDH⁺ EAC47 cells readily formed tumor spheres that were potently inhibited by NADI-351 (EC₅₀ = 110 nmol/L, Fig. 4F; Supplementary Fig. S10C). RT-qPCR analysis of ALDH⁺ EAC47 cells revealed NADI-351 treatment significantly decreased Notch target genes and the anti-apoptosis regulator Bcl-2 using sub-micromolar concentrations (Fig. 4G).

To determine whether CSC ablation occurs in vivo in response to NADI-351 treatment, mice bearing EAC47 PDX tumors were treated with vehicle (DMSO) or 30 mg/kg NADI-351 for 14 days, after which, tumors were excised and isolated tumor cells were analyzed for ALDH expression by FACS analysis. Indeed, in contrast with vehicle, NADI-351 ablated ALDH-expressing cells in vivo in EAC47 PDX tumors (Fig. 4H). Together, these results confirm NADI-351 inhibits Notch transcription and target CSCs in Notch-driven EAC tumors.

As Notch plays a critical role in CSC populations (14), we sought to further determine the specificity of NADI-351 inhibition in cells sorted on the basis on Notch1 expression. OE33 cells were sorted by Notch1 expression (Notch1⁺ and Notch1⁻; Fig. 5A and B), using a specific monoclonal anti-Notch1 antibody. We observed that treatment with NADI-351 significantly inhibited colony formation in the Notch1⁺ population in contrast with Notch1⁻ colonies, which were not inhibited (Fig. 5C; Supplementary Fig. S10D). Furthermore, NADI-351 selectively inhibited the transcription of Notch target genes in a dose-dependent manner only in Notch1⁺ cells and caused no significant changes in the transcription of target genes in Notch1⁻ cells (Fig. 5D). These results demonstrate that Notch1 is crucial to the mechanism and activity of NADI-351. To confirm that NADI-351 targets Notch1 in tumors, we treated nude mice bearing EAC47 PDX tumors for 14 days with NADI-351, after which, tumors were harvested and cells were analyzed for Notch1 expression by FACS analysis. Vehicle-treated tumors have high numbers of Notch1-expressing cells, but in contrast, NADI-351 treatment dramatically reduces Notch1-expressing cells in tumors (Fig. 5E). Further RT-qPCR analysis of these ex vivo tumor cells demonstrated that NADI-351 treatment inhibited Notch target gene expression and induced a pro-apoptotic transcriptional program (Fig. 5F) consisted with results obtained from ALDH⁺ EAC47 PDX spheres (Fig. 4G) These results confirm NADI-351's selective anti-Notch1 activity in tumors and further underscore the Notch-dependent, anti-CSC mechanism through which NADI-351 acts in vivo.

Specific inhibition of Notch signaling has been of interest in oncology drug development for more than 2 decades. To date, however, all approaches have largely failed in the clinic, most commonly due to dose-limiting GI toxicity caused by non-selective Notch inhibition, which has precluded attainment of a therapeutic dose (34–36). Even more selective approaches, including Notch1 mAbs, have failed due to limited efficacy or toxicity when combined with standard-of-care therapies, perhaps due to chronic (weeks-long) Notch inhibition by long systemic exposure inherent to mAb modalities (49).

Herein, we describe the first Notch1-selective small-molecule inhibitor of the NTC, NADI-351. NADI-351 is an orally available and potent inhibitor of Notch 1-mediated transcription that dramatically attenuates tumor cell growth. NADI-351 acts by selectively driving cell-cycle arrest and apoptosis of the CSC population by inhibiting Notch-directed transcription. Unlike other attempts to target Notch signaling in cancer, NADI-351 does not induce goblet cell metaplasia, the dose-limiting toxicity of pan-Notch inhibitors. Therefore, with its potent activity and lack of apparent toxicity, NADI-351 provides the necessary selectivity and therapeutic window required for a cancer therapeutic targeting the Notch pathway. Moreover, NADI-351 is the first compound with demonstrable selectivity in the inhibition of the CSC population of tumors.

Notch signaling is mediated by the formation of specific transcriptional activation complexes. These complexes are formed from Notch- and MAML-specific NTC's bound to DNA via CSL (3, 4). Although it is known that the 4 Notch proteins can form NTC with various MAML proteins (MAML1–3), it is not known what the contextual or functional differences are in Notch transcriptional signaling (50, 51). Many cells, including tumor cells, express multiple Notch proteins. But the specific contributions of Notch1–4 NTC in normal or pathological physiology are not well understood. For the first time, we have been able to specifically inhibit Notch1^ICD/MAML1 transcriptional complexes in tumor cells, which has provided a clearer understanding of the functional consequences of selective Notch inhibition. These data further underscore the importance of selectively targeting individual Notch paralogs and that drugging the Notch1 NTC specifically is essential for high efficacy with low toxicity.

Finally, our extensive characterization demonstrated that NADI-351 activity is derived through downstream-targeted ablation of CSCs, likely as a consequence of Notch1-specific inhibition. Developmental pathways, specifically Notch, Hedgehog and Wnt, have long been desirable targets due to their role in early development and influence in governing stem cell biology (40). Various studies have implicated a role for these pathways in CSC, which has generated hope that targeted therapies could lead to more durable responses and improved patient outcomes. However, well-controlled studies on CSC-specific effects have been limited using currently approved inhibitors and acquired drug resistance indicates that these agents do not meaningfully impact CSCs clinically; therefore, there has not been a clear validation of this hypothesis (40, 41, 44, 52). We demonstrate that NADI-351 selectively ablates multiple independent Notch-dependent CSC populations from breast and EAC cell lines and from both in vivo and ex vivo EAC47 PDX cells and tumor spheres by selectively inhibiting Notch-mediated transcription in the Notch positive CSC, resulting in cell-cycle arrest and induced apoptosis, thereby attenuating viability and tumor growth. Our studies provide compelling evidence that selective inhibition of Notch1 transcriptional complexes can profoundly arrest tumor growth while avoiding GI toxicity through specific depletion of the CSC population and suggest that NADI-351 has promise in translation to the clinic for Notch-dependent cancers.

W. Guerrant reports grants from National Cancer Institute during the conduct of the study. D.J. Robbins is a founder of Stemsynergy Therapeutics Inc., which has licensing rights to some of the molecules used in this article. D. Orton reports grants from StemSynergy Therapeutics, Inc. during the conduct of the study, as well as grants from StemSynergy Therapeutics, Inc. outside the submitted work, and a patent for StemSynergy Therapeutics, Inc. pending; and reports employment with StemSynergy Therapeutics, Inc. A.J. Capobianco reports other support from StemSynergy Therapeutics Inc. during the conduct of the study, as well as a patent for PCT/US2020/017685 and it is tiled IMPROVED INHIBITORS OF THE NOTCH TRANSCRIPTIONAL ACTIVATION COMPLEX AND METHODS FOR USE OF THE SAME. pending and licensed to StemSynergy. No disclosures were reported by the other authors.

A. Alvarez-Trotta: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. W. Guerrant: Conceptualization, data curation, visualization, methodology, writing–original draft, writing–review and editing. L. Astudillo: Investigation, methodology. M. Lahiry: Methodology. G. Diluvio: Methodology. E. Shersher: Methodology. H. Kaneku: Validation, methodology. D.J. Robbins: Conceptualization, investigation, visualization. D. Orton: Conceptualization, validation, methodology. A.J. Capobianco: conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, writing–original draft, project administration, writing–review and editing.

This work was supported by the Bankhead Coley Cancer Research from Florida Department of Health (6BC02 to A.J. Capobianco, Lead optimization and preclinical evaluation of small-molecule inhibitors of Notch transcriptional activation). This project was also generously supported by funding from the Dewitt Daughtry Family Department of Surgery and the Sylvester Comprehensive Cancer Center (to A.J. Capobianco). The authors thank Dr. Caroline Briegel (The DeWitt Daughtry Family Department of Surgery, Miller School of Medicine, University of Miami) for the MDA-MB-231 cell line. They thank Teresa De Tony (Diabetes Research Institute, Miller School of Medicine, University of Miami) for the assistance with the confocal microscope and images. The authors want to thank Dr. Daniel Bilbao (Director of Cancer Modeling Shared Resource from the Sylvester Comprehensive Cancer Center) and members of his laboratory for technical assistance in animal experiments.

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.