Small-molecule antagonists of Gli function

Specific Aims (verbatim from original R03 application)

Specific Aim 1. Screen the NIH Molecular Libraries-Small Molecule Repository (MLSMR) collection for Gli antagonists using the Sufu-KO-LIGHT cell assay described herein.

Specific Aim 2. Perform secondary assays to validate chemical hits (probes).

• Subaim 2a: To validate Hh pathway-selective probes using cell lines responsive to Hh ligands or other growth factors.
• Subaim 2b: To investigate how these compounds modulate Gli function, phosphorylation, proteolysis, and trafficking in Hh-responsive cells.
• Subaim 2c: To test validated compounds on cell-based models of Hh pathway-dependent cancers.

Background and Significance

As cancer treatments have shifted toward targeted therapies, the role of developmental signaling pathways in tumorigenesis has garnered special attention. The molecular links between ontogeny, stem cell biology, and oncogenesis is exemplified by the Hh pathway, which plays a critical role in the patterning of certain embryonic tissues and contributes to their neoplastic transformation later in life. For example, Hh signaling regulates cerebellar patterning by promoting the proliferation of neuronal precursor cells, and constitutive Hh target gene expression can lead to medulloblastoma, the most common pediatric brain tumor[1].

Hh signaling is normally initiated by the binding of Hh ligands—Sonic, Indian, and Desert Hh (Shh, Ihh, and Dhh) in mammals—to the twelve-pass transmembrane protein Patched1 (Ptch1), which is localized to a microtubule-based protrusion in the plasma membrane called the primary cilium (Figure 1)[2]. Ligand binding induces Ptch1 exit from the cilium, leading to Smo accumulation and activation within this antenna-like organelle. Activated Smo then shifts the balance between repressor and activator forms of the Gli transcription factors (Gli1, Gli2, and Gli3). While Gli2 and Gli3 are normally proteolytically processed into N-terminal repressors (Gli2/3R) in a primary cilium-, phosphorylation-, and proteosome-dependent manner, Smo promotes the stabilization of full-length Gli proteins, their conversion into transcriptional activators (Gli2/3A), and their accumulation at the distal tip of the cilium. How Smo regulates Gli function remains unclear, but this process appears to involve the scaffolding protein Suppressor of Fused (Sufu), which can directly inhibit the Gli proteins and facilitate their proteolysis. The activator form of Gli2, and to a lesser extent that of Gli3, subsequently drives the expression of Hh target genes, including cell cycle regulators, oncogenes, Ptch1, and the constitutively active transcription factor Gli1. Oncogenic Hh pathway activation can be initiated at multiple points within the signaling mechanism described previously. Mouse models suggest that prostate cancer can require Hh ligand-dependent paracrine signaling for its proliferation[3], whereas small cell lung cancer is believed to involve autocrine pathway activation[4]. Ligand-independent Hh target gene expression is operative in other cases. Nearly all basal cell carcinomas contain inactivating mutations in Ptch1 or activating mutations in Smo[5–7], and medulloblastomas can be caused by mutations in these transmembrane proteins[8–10] or the downstream regulator Sufu[11]. These observations have prompted the development of Hh pathway inhibitors by several academic and industrial laboratories, resulting in a large ensemble of Smo antagonists[12]. As a GPCR-like signaling protein, Smo is perhaps the most “druggable” target within the Hh pathway, and Smo inhibitors have demonstrated efficacy in murine tumor models[13–19] and human clinical trials[20,21]. For example, Genentech reported last year that the Smo antagonist GDC-0449 induced significant tumor regressions in patients with either metastatic basal cell carcinoma or medulloblastoma[20,21]. Yet the excitement generated by these findings is tempered by the emergence of GDC-0449-resistant tumors, which either contain point mutations in Smo that prevent drug binding or sustain Hh target gene expression in a Smo-independent manner[21]. Tumors expressing GDC-0449-resistant Smo were also insensitive to cyclopamine, a structurally distinct Smo inhibitor[22], suggesting that the development of additional Smo antagonists will not be an adequate solution to this problem. Moreover, certain cancers such as Ewing’s sarcoma and KRAS-induced pancreatic adenocarcinoma can proliferate in response to Gli activation through noncanonical mechanisms[23–27], and Smo inhibitors will be ineffective against these diseases.

Fig. 1

The Hh pathway. Functional interactions between Hh signaling components and representative small-molecule inhibitors of the pathway are shown. Positive and negative regulatory proteins are depicted in green and red. respectively.

Chemical inhibitors that act downstream of Smo therefore constitute an important therapeutic strategy for the treatment of Hh pathway-dependent cancers. Unfortunately discovering such small molecules is not a trivial endeavor. Due to the high susceptibility of Smo to small-molecule modulation, high-throughput screens using Hh ligand-stimulated cells are overwhelmingly dominated by Smo antagonists[28]. Our laboratory also recently screened 120,000 compounds for their ability to inhibit Hh target gene expression induced by saturating amounts of a chemical Smo agonist, yielding four antagonists of Gli function (HPI-1 through 4)[29]. Yet even these conditions failed to exclude hundreds of Smo inhibitors, requiring extensive, time-consuming secondary assays to identify and segregate these upstream-acting antagonists. Alternative approaches by other laboratories have utilized cells stably transfected for Gli1 or Gli2 overexpression[30–32]; however, these assay conditions survey only a limited range of “druggable” Gli regulatory mechanisms, as overexpressed Gli transcription factors escape several of the biochemical and cellular processes that control Gli function. The research plan described in this application overcomes these limitations by employing a cell-based reporter that lacks Sufu and exhibits constitutive Hh target gene expression in response to endogenous Gli activators. We emphasize that previous studies all screened using Gli-overexpressing cells, which favors hit compounds that target a subset of Gli regulatory mechanisms (since overexpressed Gli escapes certain regulatory processes. In contrast, the Sufu-KO-LIGHT cells used for this project rely on endogenous Gli for their activity. The project screen is therefore configured to identify antagonists that would have been missed in earlier studies.

Prior Art

The Hedgehog (Hh) signaling pathway is a key regulator of embryonic development and stem cell self-renewal, and its inappropriate activation can promote the onset and/or progression of several human cancers[1]. In particular, Hh pathway dysregulation has been linked to tumors of the brain, skin, pancreas, breast, ovaries, and blood, and small molecules targeting this pathway are now being pursued as anti-cancer therapies[12,33]. However, the vast majority of these compounds inhibit Smoothened (Smo), a G protein-coupled receptor-like transducer of the Hh signal, and some antagonists have already been tested in human clinical trials[20,21,33]. Small-molecule screens that have focused on downstream signaling events have utilized Gli-overexpressing cells, which favors hit compounds that target a subset of Gli regulatory mechanisms (since overexpressed Gli escapes certain regulatory processes. In contrast, the Sufu-KO-LIGHT cells used for this project rely on endogenous Gli for their activity. The project screen is therefore configured to identify antagonists that would have been missed in earlier studies.

During our validation and optimization of the Sufu-KO-LIGHT assay, as expected, GDC-0449 was unable to reduce firefly luciferase levels in these cells even though it inhibits Smo activity with nanomolar potency[17], whereas HPI-1 blocks reporter expression in a dose-dependent manner (IC50 = 0.90 μM; Hill coefficient ~1) - see Figure 2

Figure 2

Inhibition of Gli-Sufu-KO LIGHT reporter by HPI-1 and GDC-0449.

A SciFinder search for Gli antagonists found several of the citations in this introduction describing Smo inhibitors[12,29–32]. Moreover, while several citations describe Gli antagonists[34–38], these antagonists have been found using cell assays with overexpressed Gli, which suffers from escape of certain Gli regulatory processes, as noted previously. Also the follow-on work to establish the Gli pathway antagonism in these publications is incomplete and inconclusive.

2.1. Assays

Table 2

Summary of Assays and AIDs.

2.2. Probe Chemical Characterization

b. Probe chemical structure including stereochemistry if known

Probe ML340 does not contain any stereocenters (see Figure 3)

Figure 3

Structure of ML340.

d. If available from a vendor, please provide details

The probe is not commercially available. We synthesized this compound and deposited 34 mg with MLSMR (Bio-Focus DPI), (see probe Submission Table 4)

Table 4

Probe and Analog Submissions to MLSMR (BioFocus DPI) for Gli-Sufu Antagonist ML340.

e. Solubility and Stability of probe in PBS at room temperature

The solubility of ML340 was investigated in aqueous buffers at room temperature (see Figure 4). As noted in the Summary of in vitro ADME/T properties (see Table 7), ML340 has a poor solubility of 0.67 μg/mL in 137 mM NaCl, 2.7 mM KCl, 10 mM sodium phosphate dibasic, 2 mM potassium phosphate monobasic, pH 7.4 (PBS) at room temperature (23°C). The Chemical Stability was determined in 1xPBS, pH7.4/1xPBS and 50% ACN with 92.5% of the parent compound remaining after 48hrs. ML340 appears very chemically stable when in solution (1:1 PBS:acetonitrile), and the apparent instability is rather a reflection of its poor aqueous solubility.

Figure 4

Stability of ML340 in PBS at ambient temperature.

Table 7

Summary of in vitro ADME Properties of Gli-Sufu Antagonists Probe CID6538918.

f. Calculated and known probe properties: are shown in Table 3

Table 3

Calculated Properties for ML340 (CID 6020222).

2.3. Probe Preparation

Preparation of N-(6-chloropyrimidin-4-yl)benzamide [2]. (cmpds are numbered as in Scheme 1)

Scheme 1

Synthesis of ML340, conditions. a. benzoyl chloride, DIPEA, DMAP, DCM, reflux, overnight (78%); b. 4-methylpiperidine, 80°C, 2 hours (60%).

2

6-chloropyrimidin-4-amine 1 (2.1 g, 16 mmol) and benzoyl chloride (2.3 g, 16 mmol) was dissolved in 100 ml of dichloromethane. The resulting mixture was added 3 ml of DIPEA and catalytic amount of DMAP, then refluxed at 50 °C overnight. When the reaction was determined to be complete by HPLC, the reaction mixture was cooled to room temperature and concentrated under reduced pressure. The resulting oil was chromatographed on silica gel and eluted with ethyl acetate and hexane to yield 3.0 g of product 2 (78 % yield). ). ¹H NMR (400 MHz, CDCl3) δ(ppm) 7.52 (m, 2H), 7.62 (m, 1H), 7.90 (m, 2H), 8.43 (s, 1H), 8.55 (s, 1H), 8.96 (s, 1H). ¹³C NMR (400 MHz, CDCl3) δ(ppm) 110.0, 127.4, 129.1, 132.9, 133.2, 158.0, 158.4, 162.6, 166.2. MS (EI) m/z 234 (M+1).

N-(6-chloropyrimidin-4-yl)benzamide 2 (1.0 g, 4.3 mmol) and 4-methylpiperidine (0.85 g, 8.6 mmol) were heated at 80 °C for 2 hours. When the reaction was determined to be complete by HPLC, the reaction mixture was cooled to room temperature and concentrated under reduced pressure. The resulting oil was triturated with methanol and filtered to yield ML340 as a white solid 0.76 g (60 % yield). ¹H NMR (400 MHz, CDCl3) δ(ppm) 0.98 (m, 3H), 1.19 (m, 1H), 1.74 (m, 4H), 2.92 (m, 2H), 4.45 (m, 2H), 7.52 (m, 2H), 7.63 (m, 1H), 7.66 (m, 1H), 7.90 (m, 2H), 8.29 (s, 1H), 8.52 (s, 1H). ¹³C NMR (400 MHz, CDCl3) δ(ppm) 21.8, 31.1, 33.8, 44.7, 90.0, 127.2, 128.9, 132.5, 133.9, 156.6, 157.6, 162.9, 166.3. MS (EI) m/z 297 (M+1). The characterization by ¹H NMR and HPLC/MS and the experimental details can be found in the Supplemental Appendices in Section 6.2.

3.1. Summary of Screening Results

The flow chart in the Figure 5 summarizes the hit identification, confirmation and validation for the Gli antagonist (Sufu-KO-LIGHT) screen. In summary, the Molecular Libraries Small Molecules Repository (MLSMR) collection comprising ~360,000 compounds was screened at 5 μM compound concentration in a final volume of 6 μL per well (AID 588413). The average Z′ for the screen was 0.75, the signal to background was 85, the signal to noise was 240, and the signal window was 5.9 as in Table 5. These statistics indicate that the assay and screen behaved in a manner typical for a well-run cell based high throughput screen. With a cutoff of ≥55 % of inhibitory activity compared to the DMSO controls, 2,515 primary hits were identified, resulting in a 0.7 % hit rate, which is reasonable.

Figure 5

Screening hit triage.

Table 5

Summary of Gli-antagonist (Sufu-KO-LIGHT) HTS Performance.

Fresh stock solutions “cherry picks” of these 2,515 hits were requested from the MLSMR (Biofocus DPI) and 2,464 were received (98.0 %). These were then retested in the Sufu-KO-LIGHT confirmatory assay (AID 602428) in triplicate at 5 μM resulting in 1056 confirmed (42.9 % confirmation) actives, based on at 1 of 3 retests being ≥45 % inhibition threshold. Subsequently, to triage these confirmed hits were tested in duplicate against a cytotoxicity counterscreen (ATPLite) and 550 hits (52.1 %) passed this. These were screened for selectivity against for antagonism of Wnt3a signaling pathway resulting in 345 compounds (62.7 %). This testing funnel, detailed previously and shown in Figure 5, excluded a total of 2119 compounds. Hit validation in 10-point dose response titration studies, using these 3 assays were performed on the remaining 345 compounds (Gli:AID 602464, cytotox-AID 651569, Wnt3a-AID 651570). These compounds were subject to further computational filters. Data mining of the PubChem and BCCG databases was performed to filter out hits that promiscuously inhibit other HTS assays. Our cheminformaticians applied cheminformatic filtering to eliminate known PAINS[39,40] (Pan Assay Interference Compounds). Additionally, our experienced project chemists visually inspected the chemical structures of the filtered hits and eliminated any compounds they judged were a priori undesirable, intractable or were previously problematic in other screens and projects based on these experimental results and informatic analysis, dry powder compounds were selected for follow up studies and additional, commercially available analogues were ordered.

3.2. Dose Response Curves for Probe

Figure 6Dose response curve for ML340 in the cell-based Gli antagonist (Sufu-KO-LIGHT) assay

3.3. Scaffold/Moiety Chemical Liabilities

This scaffold contains no reactive moieties or functional groups known to form covalent bonds, and appears stable in aqueous buffer, though is rather insoluble (Figure 3). The amide bond linking the two hemispheres of the probe molecule may provide a facile site for metabolism of the compound in vivo. These predictions are verified by the values obtained for the in vitro ADME/T panel (Table 7. in Section 3.6) showing poor aqueous solubility, moderate lability in plasma and poor stability in liver microsomes.

3.4. SAR Analysis

From the original hit series we found several promising scaffolds. We focused our efforts on the pyrimidine series represented by CID 1540596 (see Entry 1, Table 6, 1^st line, CP = MLSMR cherry pick). In the initial screen, CID 1540596 was one of the most potent compounds screened to date, IC₅₀ = 295 nM, when retested from freshly supplied 10 mM DMSO stocks from the NIH MLSMR. Subsequently, we purchased this from a commercial supplier and found the powder to be ~ 5–10 fold more potent at 56.5 and 22.9 nM (see Entry 1, Table 6, 2^nd line, P = purchased). In addition, the Assay Provider’s lab (Prof. Chen, Stanford) tested this compound independently and obtained an IC₅₀ ~ 35 nM. We independently synthesized this compound during the course of rounds of analog-by-catalog (ABC) and medicinal chemistry synthesis and SAR. The resynthesized batch of CID 1540596 had an IC₅₀ of 24.2 nM. Overall, we synthesized 57 compounds in the pyrimidine series, to ultimately select ML340 (CID 65389180) as our probe with the consistently lowest IC₅₀ (33.6 ± 4.3 nM) and selectivity against Wnt3a. The following narrative and Table 6 detail the SAR explorations.

Table 6

SAR explorations of ML340.

Our initial SAR strategy for this series of molecules involves investigating the effects of changing the piperidine ring (R₁) of CID 1540596 to other ring systems including pyrrolidine (Entry 3), morpholino (Entry 33), thiomorpholine (Entry 29), piperazine (Entry 40), and N-methylpiperazine (Entry 31), all of these analogs are less potent than CID 1540596. When we replaced the piperidine ring with an azepane (Entry 28) or 4-methylpiperidine (Entry 39) the activity of the series was comparable to CID 1540596 IC₅₀ = 61.2 nM and 33.6 nM respectively. Substitution of various alkyl amines (Entries 4, 16, 17, 32, 34, 35, 37, 38) for the piperidine produced compounds with reduced potency versus CID 1540596.

When we retained the piperidine ring (R₁) of CID 1540596 and changed the R₂ group we found that substitution of electron donating groups or electron-withdrawing group on the pendant phenyl ring the potency of the compounds increased (Entries 2, 7, 10, 20, 43–50, 52). Substitution of the phenyl group with a thiophene ring (Entries 54 and 55), furan (Entry 51), naphthalene ring (Entries 56 and 57) or the 2-benzo-thiophene group (Entry 55) resulted in compounds with less activity than CID 1540596.

We synthesized several compounds where a methyl group was substituted for a hydrogen atom on the pyrimidine ring at the C-2 position. For example, Entries 1 and 14, the compound with the methyl group was substantially less potent, IC₅₀ = 24.2 nM versus IC₅₀ > 11.7 μM. The Probe Compound ML340, we selected is Entry 39 as it was the most potent compound synthesized, IC₅₀ = 33.6 nM and displayed 64.9 fold selectivity for inhibiting Gli transcription versus the Wnt signaling assay.

3.5. Cellular Activity

All of the functional assay describe that support this project are cell-based and therefore this probe does demonstrate efficacy in antagonizing the Gli function in live cells, while not exhibiting general cytotoxity, and is not antagonistic to the Wnt3a pathway. We do note however, the probe and some of the close analogs exhibit some strong direct activation (rather than inhibition) of Wnt3a signaling at concentrations > 1000 nM. Similarly, the Hh reporter activity begins to increase at these elevated compound levels. We hypothesize that this reflects a non-specific off-target effect manifested at higher compound doses.

3.6. Profiling Assays

In vitro Pharmacology Profiles conducted by SBCCG Exploratory Pharmacology group of Probe CID6538918 [ML340] (See Table 7).

ML340 has poor solubility in aqueous media 0.84, 0.62 μg/mL, 0.72 μg/mL at pH 5.0, 6.2 and 7.4 respectively.

The PAMPA (Parallel Artificial Membrane Permeability Assay) assay is used as an in vitro model of passive, transcellular permeability. An artificial membrane immobilized on a filter is placed between a donor and acceptor compartment. At the start of the test, drug is introduced in the donor compartment. Following the permeation period, the concentration of drug in the donor and acceptor compartments is measured using UV spectroscopy. Consistent with the predicted LogP (see Table 3), ML340 is very permeable at pH 5.0 and highly permeable at pH 6.2 and 7.4 in this assay.

Plasma Protein Binding is a measure of a drug’s efficiency to bind to the proteins within blood plasma. The less bound a drug is, the more efficiently it can traverse cell membranes or diffuse. Highly plasma protein bound drugs are confined to the vascular space, thereby having a relatively low volume of distribution. In contrast, drugs that remain largely unbound in plasma are generally available for distribution to other organs and tissues. ML340 shows high binding to plasma proteins in both mouse and human plasma.

Stability in PBS and in 1:1 PBS Acetonitrile. As ML340 is very insoluble in PBS, and ML340 is apparently fairly stable in PBS and completely stable in 1:1 PBS:Acetonitrile with 57.21 % and 92.54 % of parent ML340 remaining after 48 hrs. of incubation at ambient temperature.

Plasma Stability is a measure of the stability of small molecules and peptides in plasma and is an important parameter, which strongly can influence the in vivo efficacy of a test compound. Drug candidates are exposed in plasma to enzymatic processes (proteinases, esterases), and they can undergo intramolecular rearrangement or bind irreversibly (covalently) to proteins. ML340 shows very good stability in human (85.04 % remaining at 3 hrs) and moderate stability (39.54 % remaining at 3 hrs) in mouse plasma.

The microsomal stability assay is commonly used to rank compounds according to their metabolic stability. This assay addresses the pharmacologic question of how long the parent compound will remain circulating in plasma within the body. ML340 shows poor stability in both human and mouse liver microsomes, potentially limiting the utility of this probe in in vivo rodent models and in a human therapeutic context. This is one area of needed improvement for future PK optimization approaches.

ML340 exhibits no toxicity (LC₅₀ > 50 μM) towards immortalized Fa2-N4 human hepatocytes.

4.1. Comparison to existing art and how the new probe is an improvement

In this report we have identified the most potent Gli-Sufu Antagonists, in the literature to date. ML340 (CID 6538918) IC₅₀ = 33.6 nM. ML340 is more potent than any known compound in the literature and it is chemically stable and will serve as a unique biochemical probe.

Nearly all known Hh pathway inhibitors target the transmembrane protein Smo, and therefore probes that act on downstream signaling proteins are needed. ML340 is a small molecule that is epistatic to the Gli antagonist Sufu, providing a compound for studying the mechanisms that regulate Gli transcription factor function and structural leads for new Hh pathway-targeting anti-cancer therapies. This compound will be particularly valuable to the developmental biology and cancer biology communities, as the biochemical and cellular processes that control Gli activity state remain elusive and Hh pathway-dependent tumors that acquire resistance to Smo antagonists have been observed in the clinic.

4.2. Mechanism of Action Studies

Mechanism of action studies are planned as future studies (see Section 4.3).

4.3. Planned Future Studies

Having identified ML340 as a potent antagonist of Gli function, we will investigate its mechanism of action through a variety of cell-based assays. We will first confirm its epistatic relationship with other Hh pathway components by evaluating its efficacy against Hh target gene expression induced by Shh, the small-molecule Smo agonist SAG, Gli1 overexpression, or Gli2 overexpression. If ML340 can inhibit exogenous Gli1 and/or Gli2 function, we will assess the degree to which it can discern between these two transcription factor isoforms. This will be achieved by expressing the individual Gli genes in mutant murine fibroblasts lacking the other isoform.

The effects of ML340 on Gli state will then be determined. For example, Gli3 can be electrophoretically resolved into its N-terminal repressor, full-length, and phosphorylated full-length activator states, and we will investigate whether the compound alters Gli3 state in the absence or presence of Shh stimulation. Similarly, Gli2 trafficking in response to Shh treatment can be followed by immunocytochemistry, and we will determine whether ML340 prevents the Shh-induced accumulation of Gli2 in the primary cilium. Other aspects of Gli function can also be interrogated, including its association with Sufu, various post-translational modifications associated with Gli regulation, and its ability to bind DNA.

6.1. Assay Details

6.2. Chemistry: ¹H NMR, ¹³C NMR, and LC-MS spectra of ML340

¹H NMR Spectrum of ML340 (400 MHz, CDCl₃)

¹³C NMR Spectrum of ML340 (400 MHz, CDCl₃)

LC-MS for ML340

(Reverse phase Acquity UPLC HSS T3 C18 column)