Design, synthesis and evaluation of 2, 6, 8-substituted Imidazopyridine derivatives as potent PI3Kα inhibitors

Abstract Inhibition of PI3K pathway has become a desirable strategy for cancer treatment. In this work, a series of 2, 6, 8-substituted Imidazo[1,2-a]pyridine derivatives were designed and screened for their activities against PI3Kα and a panel of PI3Kα-addicted cancer cells. Among them, compound 35 was identified as a PI3Kα inhibitor with nanomolar potency as well as acceptable antiproliferative activity. Flow cytometry analysis confirmed 35 induced cell cycle arrest and apoptosis in T47D cells. In addition, it also showed desirable in vitro ADME properties. The design, synthesis, and SAR exploration of 35 are described within.


Introduction
Malignant tumours have always been one of the major diseases threatening human health. At present, the commonly used treatment methods for cancer are inseparable from the support of chemotherapy drugs 1 . Therefore, it is of great significance to find anti-tumour drugs with high efficiency and low side effects 2 .
Phosphatidylinositol-3-kinases (PI3Ks) are a family of lipid kinases that are responsible for catalysing the phosphorylation of inositol phospholipids to produce PIP3 (phosphatidylinositol triphosphate). This phosphorylation process results in recruiting cytosolic signalling enzymes such as Akt to the plasma membrane, triggering cell growth, proliferation, differentiation, and motility [3][4][5] . Aberrant activation of PI3K as well as its downstream effectors including Akt and mTOR has been linked to numerous forms of cancer including lymphatic tumours, breast, lung, brain, ovarian, melanoma, and prostate cancers [6][7][8] . Moreover, the negative regulator PTEN 9 which dephosphorylates PIP3 to PIP2, is often inactivated in many cancer types, leading to elevated levels of PIP3 and increased tumour survival 3 . With distinct sequence homology and substrate preferences, the eight known PI3K family members are divided into classes I, II, and III. The most extensively studied Class I PI3Ks are further subdivided into Class IA (a, b, and d) and Class IB (PI3Kc), based upon the types of regulatory subunits and the catalytic domains to which they bind in the active heterodimeric forms. The former Class IA PI3Ks are activated through tyrosine kinase signalling, whereas the sole PI3Kc is mostly activated through GPCRs 10 .
While there is growing evidence that small molecule inhibition of PI3Ks as an attractive strategy for oncology indication 11,12 . Several small molecule inhibitors have been approved by FDA for use in patients or under active clinical development, including pan-PI3K inhibitors Copanlisib (1) 13,14 , Buparlisib (2) 15 , and GDC-0941 (3) 16 , a-specific PI3K inhibitor Alpelisib (4) 17,18 , PI3Kd-selective inhibitor Idelalisib (5) 19 , and so on ( Figure 1). The common pharmacophoric features of reported PI3Ka inhibitors are summarised as follows: First, the morpholine or heterocycle ring forms an important hydrogen bond with Val851 in the hinge region. Another hydrogen bond can be found between aromatic ring or lipophilic side chain with Tyr836, Asp810 and/or Lys802 at the affinity pocket. Furthermore, some derivatives extend to the solvent exposed region to form additional interaction with the surrounding amino acids as well as improve the druggability of designed compounds. Among them, PIK-75 (6) is an Imidazo[1, 2-a]pyridine derivative that shows good selectivity for PI3Ka over the other related Class I PI3Ks as well as desirable activity in a human cancer xenograft model 20 . However, the SAR explorations on PIK-75 are focussed on either replacing the Imidazo[1,2-a]pyridine ring with other heterocycles or modifying the sulfonohydrazone side chains 21 , modifications on C-2 or C-8 position have barely been reported. Considering the sulfonohydrazide group as an alert structure, we were interested in removing this group, and introducing amide and substituted aryl group to the 2-, and 8-position of Imidazo[1,2-a]pyridine ring to improve the solubility as well as provide hydrogen binding group that was essential for PI3Ka activity ( Figure 2).
In this manuscript, we communicate the discovery and optimisation of a series of Imidazo[1,2-a]pyridine derivatives as PI3Ka selective inhibitors and their potential application in the treatment of cancers. The efforts leading to the discovery of the PI3Ka-specific inhibitor 35 are described herein.

Results and discussion
Chemistry As depicted in Scheme 1, a simple five-step synthetic sequence was used for the preparation of compounds described herein. Firstly, 2-aminopyridine derivatives and N-Bromosuccinimide (NBS) undergo electrophilic aromatic substitution reaction in the presence of DMF to obtain 2-amino-3-bromo-pyridine derivatives (11a-b), 11a-b and commercially available 3-bromopyridin-2amine were then cyclized with ethyl 3-bromopyruvate to give the ethyl 8-bromo-Imidazo[1,2-a]pyridine-2-carboxylate derivatives (12a-c). Intermediates 12a-c were further hydrolysed by NaOH to offer the corresponding carboxylic acids (13a-c). Subsequently, the key intermediates 14a-h were synthesised through amidation reaction of 13a-c with various amines and coupling reagent HBTU in DMF. Finally, compounds 14a-h undergo Suzuki-Miyaura reaction with the corresponding boronic acids or esters (9a-c) to obtain the corresponding target products 15-46. The chemical structures of the target compounds were confirmed with HRMS, 1 H NMR, and 13 C NMR.

PI3Ka inhibition and SARs
To screen the inhibitory activity of the target compounds against PI3Ka, the Kinase-Glo TM assay with ATP at 25 lM was utilised with PIK-75 as positive control 22 . We chose to explore the SARs at three positions on the Imidazo[1,2-a]pyridine core. To start with, a set of 8-position modifications were prepared while keeping the morpholinyl amide substituent on 2-position fixed, since it allowed quick structure À activity profiling via coupling reactions of bromide with aryl borates or boronic acids. The results are summarised in Table 1. Fluoro substituted phenyl was first incorporated, through scanning the 2, 3, 4-position of phenyl, it was clear that the ortho-position was not well tolerated (15, 16 vs 17), with only 14.1% inhibitory rate at 10 lM. Moreover, meta-fluoro substituted analog 16 showed a slightly enhanced PI3Ka activity compared to para-fluoro substituted analog 15. A similar trend can be observed for compound 23 vs 22. Replacement of fluorophenyl with 3-pyridinyl 20, 6-methoxy-3-pyridinyl 21 or 4-pyridinyl 30 did not increase the activity (20, 21 vs 16; 30 vs 29). We then turned our attention back to the phenyl substituents. Changing the fluoro  group of compound 15 to amide 18, sulphonamide 19, or urea 33 led to an increase in potency, with the inhibitory rate range from 44.5 to 58.8, a similar trend can be found both at the pyrrolidinyl amide (26-28) and N-methylpiperidinyl series (29-33).
Unable to significantly improve the potency on the phenyl group at the 8-position of the Imidazo[1,2-a]pyridine template, we examined the binding mode of PIK-75 to PI3Ka, the removing sulfonohydrazide group formed an important H-bond with PI3Ka. To compensate for this loss, we speculated that the pyridinephenylsulfonamide on the 8-position would project into the affinity binding pocket of PI3Ka, making H-bond with Lys802 which is beneficial for potency. Thus, compounds 34-36 were prepared. However, compound 34 offered no advantage over the amide or sulphonamide analogs (26-28). Gratifying, a substantial increase in potency was observed when 4-fluoro or 2, 4-difluoro group were further incorporated (35, 36 vs 34). Having established the pyridinesulfonamide on the 8-position in 35 as the optimal substituent, different amines on the 2-position of the Imidazo[1,2a]pyridine core were also designed and explored (40-44), since they may have influence on potency as well as physicochemical properties. The results revealed that all the compounds exhibit moderate to excellent inhibitory rate, ranging from 53.4 to 90.1, with the rank: pyrrolidine > 2-morpholinoethan-1-amine > 3-morpholinopropan-1-amine > morpholine > 4-methylpiperidine > 1methylpiperazine, which was further demonstrated by compounds 36-39. Finally, the impact of 6-position on activity was also explored, neither the chloro nor the hydrogen substituent showed positive effect on the potency compared to methyl substituent (45 vs 39; 46 vs 43). Overall, The SAR studies depicted above (see also Figure 3) led to the identification 35, which was the most potent compound that we had obtained.
In vitro antiproliferative activity Following the direction, Compounds 35, 36, and 40 with inhibitory rate higher than 85% at 10 lM were progressed into full IC 50 value determination with Alpelisib as positive control. As shown in Table 2, we detected these three compounds exhibit submicromolar activity for PI3Ka, with IC 50 of 0.15, 1.12, 0.50 lM, respectively. Given the desirable enzyme activity profile, these compounds were further assayed for their antiproliferative activities against five independent human tumour cell line collections harbouring PIK3CA mutation 23 , including human ovarian cancer cell line SKOV-3, human breast cancer T47D and MCF-7, human lung cancer H1975 and H460. As shown in Table 2, all compounds exhibited antiproliferative effects at micromolar concentrations. In general, the cellular activity of tested derivatives corrected with their effects on kinase activity. Interestingly, these PI3K inhibitors were more sensitive with human breast cancer than non-small cell lung cancer and ovarian cancer. Notably, compound 35 potently inhibited the proliferation of human breast cancer lines T47D and MCF-7 with IC 50 of 7.9 and 9.4 lM, respectively. Hence, compound 35 was selected for further evaluation.

Flow cytometry
Considering compound 35 was most sensitive in T47D cells, we decided to get some mechanistic insight into the mode of action   in T47D cell line. As such, the annexin V/7-AAD staining and cell cycle phase distribution have been carried out. In line with its activity, compounds 35 induced a S cell cycle arrest in T47D cells. Furthermore, T47D cells were also treated with different concentrations of 35, and the percentage of apoptotic cells was measured. Exposure to 50 lM of 35 resulted in an induction of early (1.51%) and late (21.8%) apoptosis of T47D cells ( Figure 4).

In vitro ADME properties of compound 35
To assess its druggability, In vitro ADME properties of compound 35 were also determined. As depicted in Table 3, compound 35 showed acceptable stability both in human liver microsomes (HLM) and mouse liver microsomes (RLM), with T 1/2 of 45.1 and 31.9 min, respectively. To screen its membrane permeability, Parallel artificial membrane permeability assay (PAMPA) was conducted and the result revealed that compound 35 was highly permeable, with permeability rates (P e ) higher than 10 nm/s. Besides, to avoid any drug-drug interaction, compound 35 was further screened for its cytochrome P450 inhibition activity, and 35 showed little or minimal inhibitions for 5 major cytochrome P450 (CYP1A2, 2D6, 2C9, 2C19, and 3A4) enzymes.

Molecular modelling and dynamic simulation
To better understand the activity of 35 against PI3Ka, the binding mode between the 35 and PI3Ka was then studied, and the structure of PI3Ka (PDB ID code: 4JPS) 18 was selected as the docking model. Compound 35 adopted a similar binding mode as previously reported PI3Ka inhibitors 24,25 . As shown in Figure  5(A), the methoxy in 2-position of pyridine and the oxygen of sulphonamide in compound 35 formed two H-bond donor-acceptor interactions with Lys802, and the pyridine nitrogen atom of 35 is part of a hydrogen bond network involving the conserved water molecule with the side chains of residues Tyr836 and Asp810. In addition, compound 35 also made hydrophobic interactions within Ile848, Val851, Thr856, and Asp933, respectively. The overlay of 35 with PIK-75 revealed the pyrrolidine amide moiety formed weaker interactions with PI3Ka than that of PIK-75 ( Figure 5(B)). However, the binding affinity of compound 35 with PI3Ka was partly compensated by forming strong hydrogen bonds with the backbone of Tyr836 and Asp810 than that of a bromine atom in the Imidazopyridine ring of PIK-75. To further access the binding stability of compound 35/PI3Ka adducts, a 90 ns long molecular dynamics simulation was carried out. The RMSD values of protein backbone atoms relative to the initial structure were calculated to examine the protein stability over the course of the simulation period. As shown in Figure 5(C), the RMSD did not fluctuate significantly, and the value of RMSD converged to 3.0 Å at 90 ns, which revealed that compound 35 could stably bind to PI3Ka. In general, the docking result confirm the rationality of our design strategy.

Conclusion
In

General procedure for preparation of intermediate 12a-c
Commercially available 2-aminopyridine or 11a-b (181.8 mmol), ethyl bromopyruvate (44.3 g, 227.2 mmol) were successively added to ethanol (300 mL), the mixture was heated to 80 C for 4 h. Upon completion, the reaction solution was cooled to room temperature, concentrated under reduced pressure to obtain a brownish red solid, which was then triturated with acetone to give 12a-c.  General procedure for preparation of intermediate 13a-c To a solution of corresponding carboxylate 12a-c (37.2 mmol) in ethanol (50 mL) was added NaOH solution (15 mL , 1 N), the reaction was heated to 80 C for 3 h. After cooling, the reaction solution was concentrated under reduced pressure to remove the solvent and the pH was adjusted to 5 with diluted hydrochloric acid (15 mL , 1 N), a large amount of solid was precipitated. The precipitant was filtered, washed with water, and dried under vacuum to afford the products 13a-c.

General procedure for preparation of intermediate 14a-h
To a solution of corresponding carboxyl acid 13a-c (1.2 mmol) in DMF was subsequently added morpholine (0.31 g, 3.5 mmol), triethylamine (0.17 g, 1.8 mmol) and HBTU (0.67 g, 1.8 mmol). The reaction was stirred at room temperature overnight. Upon completion, the mixture was then poured into ice-water, extracted with EtOAc (15 mL Â 3). The combined extracts were then washed with brine, dried over Na 2 SO 4 , and concentrated. The crude products were then subjected to flash chromatography to give pure products 14a-h.   4 (0.076 g, 0.093 mmol) was stirred at 90 C for 12 h under nitrogen atmosphere. After cooling to room temperature, water (30 mL) was added followed by extraction with EtOAc (10 mL Â 3). The combined organic layers were dried over MgSO 4 and concentrated under reduced pressure. The crude products were purified using silica gel column chromatography to give pure compounds.           enzyme (PI3Ka from Invitrogen), the PIP2 (Life Technologies) substrate, and ATP (25 lM, Sigma) were diluted in kinase buffer to the indicated concentrations. The assay plate was covered and incubated at room temperature for 1 h. Then, the Kinase-Glo reagent (Promega) was added to the PI3Ka plate to stop the reaction, mixed briefly with centrifuge, shaken slowly on the shaker and then equilibrate for 120 min. Finally, add Kinase detection reagent to each well, shake 1 min, equilibrate for 30 min before reading on a plate reader for luminescence. The data were collected on Envision and presented in Excel. IC 50 values were calculated from the inhibition curves.
Cell proliferation assays All target compounds were evaluated for antiproliferative potency against SKOV-3, T47D,NCI-H1975, NCI-H460,and MCF-7 tumour cell lines using a CellTiter-Glo V R Luminescent Cell Viability Assay. The human tumour cell lines used were obtained from the ATCC or Shanghai ZhongQiaoXinZhou Biotech. All the mediums and FBS were Gibco. T47D, H1975 and NCI-H460 tumour cells were cultured in RPMI1640 medium supplemented with 10% FBS. SKOV-3 was cultured in McCoy's 5 A medium supplemented with 10%FBS. The day before treatment with compounds, cells were seeded at a density of 2000 or 5000 in each well of a 96-well plate. The tumour cells were then treated with 3-fold serial diluted compound or DMSO control in the incubator at 37 C and 5% CO 2 for 3 days, prior to the addition of CellTiter-Glo reagents (Promega) and reading of luminescence using a PerkinElmer Envision plate reader. Data were analysed using GraphPad Prism 7.0.

Cell cycle and apoptosis analysis
For cell cycle analysis, T47D cell (1 Â 10 6 ) were seeded in 6-well plates overnight and treated with different concentrations of compound 35 and DMSO control on the next day. After 24 h, the cells were collected by EDTA-free trypsinization, centrifuged at 1500 rpm for 5 min, and fixed in 75% ethanol at À20 C for 1 h. The cells were then centrifuged at 2000 rpm for 5 min, resuspended in 500 mL PI/Rnase Staining Buffer, incubated for 30 min at room temperature, and analysed using flow cytometry. The data were analysed using Modfit software. For cell apoptosis analysis, T47D cell were seeded in 6 well-plates and treated with DMSO or compound 35 for 24 h. The treated cells were collected and washed with cold PBS for three times, resuspended in 185 mL binding buffer containing 5 lL Annexin V-FITC Apoptosis Detection Kit and 10 lL 7-AAD Viability Staining Solution for 15 min at room temperature in the dark. Apoptotic cells were analysed by flow cytometer.

Liver microsomal stability assay
The metabolic stability of compound 35 was determined in human or SD rat liver microsomes, with and without the NADPH regenerating system. Briefly, compound 35 was incubated with microsomes (human microsome, CORNING, lot No. 38296; rat microsome, Xenotech, Lot No. 2110178) (0.5 mg protein/mL) at 1 lM at 37 C in potassium phosphate buffer (100 mM at pH 7.4 with 10 mM MgCl 2 ). The reactions were initiated by adding prewarmed cofactors (1 mmol NADPH). After incubation for different times (0, 5, 10, 20, 30, and 60 min) at 37 C, cold acetonitrile containing 200 ng/mL tolbutamide and 200 ng/mL labetalol as internal standards was added to precipitate the protein. Then, the samples were centrifuged, and the supernatants were transferred into HPLC water, mixed by plate shaker for 10 min prior to LC À MS/ MS analysis.
Parallel artificial membrane permeability assay (PAMPA) 10.0 lM donor solution (5% DMSO) was prepared by diluting of working solution with PBS. The donor solution was added to each well of the donor plate, whose PVDF membrane was precoated with 5 mL of 1% lecithin/dodecane mixture. Duplicates were prepared.Then, PBS was added to each well of the PTFE acceptor plate. The donor plate and acceptor plate were combined and incubated for 4 h at room temperature with shaking at 300 rpm. Acceptor samples and donor samples were prepared and analysed by LC-MS/MS.

Cyp 450 inhibition assay
Cytochrome P450 inhibition was evaluated in human liver microsomes (0.253 mg/mL, Lot No. 38292) using five specific probe substrates (CYP1A2, 10 lM phenacetin; CYP2D6, 5 lM dextromethorphan; and CYP3A4, 2 lM midazolam; 2C9, 5 lM Diclofenac; 2C19, 30 lM S-mephenytoin) and positive control in the presence of multiple concentrations of the test compound. After pre-warm at 37 C for 10 min, the reaction was initiated by the addition of of NADPH. The mixture was incubated at 37 C for 10 min, terminated by the addition of cold stop solution (200 ng/mL tolbutamide and 200 ng/mL labetalol in acetonitrile). The samples were centrifuged, and the supernatants were analysed by LC À MS/MS.

Molecular docking and dynamics simulation
The protein coordinate (PDB: 4JPS), downloaded from the Protein Data Bank (http://www.rcsb.org/pdb/), was chosen as templates to compare the docking mode among compound 35 bound to PI3Ka. Molecular docking calculations were conducted using the Dock6 protocol in Yinfo Cloud Platform (http://cloud.yinfotek.com/ ). Briefly, the structure of compound was built with energy minimisation in MMFF94 force field, and PI3Ka was assigned hydrogen atoms and partial charges in Amber ff14SB force field and partial charges in Chimaera. The binding pocket of the crystal ligand was assumed to be analogous to that of 35 in PI3Ks. The box centre and the dimensions were thus set. The DOCK 6.7 program was utilised to conduct semiflexible docking with 10 000 different orientations generated. Then, the Grid-based score was calculated for each pose. The image files were generated by Pymol. To investigate the combined stability of inhibitors to PI3K kinase, the MD simulations were conducted on the PI3K kinase in complex with compound 35 by Amber 16 software. The restrained electrostatic potential (RESP) calculated by the Gaussian16 package was used to fit the charges for the inhibitors. The ff14SB force field and the general AMBER force field (gaff2) were used for PI3K kinase and the inhibitors, respectively. Build a solvent octahedral box with a boundary 15 Å away from the protein and use the TIP3P water model to fill the entire octahedral box. 7 Na þ ions were added to neutralise the system. The MD simulation protocol comprised the following steps 1 : The limiting potentials of proteins, ligands, and counterions were all restricted by the force constant of 200 kcal/ (mol Å), and the energy of the solvent water molecules was minimised to make the water molecules reach a relaxed state. (2) The energy of the system is further minimised. Protein, ligand and ions were subjected to the limiting potential with a force constant of 300 kcal/(molÅ). (3) The restriction potential of the protein backbone was restricted by the force constant of 20 kcal/(mol Å). (4) Then the system was minimised without any restriction. The cpptraj module in AMBER16 was employed for the RMSD calculations.