Discovery of SIPI6473, a New, Potent, and Orally Bioavailable Multikinase Inhibitor for the Treatment of Non-small Cell Lung Cancer

Abstract A novel series of quinazoline derivatives were designed, synthesized, and evaluated as multikinase inhibitors. Most of these compounds showed antiproliferation activities of several human cancer cell lines and exhibited inhibition efficacy against the estimated glomerular filtration rate (EGFR) in the nanomolar level. Among those compounds, compound B5 (also named SIPI6473) displayed the maximum effect, and thus was chosen for further study. Our data revealed that B5 inhibited the activity of several kinases (such as EGFR, VEGFR2, and PDGFRα) that contributed to the development of non-small cell lung cancer (NSCLC). Besides, an in vivo study also showed that B5 inhibited tumor growth without signs of adverse effects in the A549 xenograft model. In conclusion, B5 may represent a new and promising drug for the treatment of NSCLC.


Introduction
Nitrogen-containing heterocycles, including quinazolinebased compounds, have gained great interest in targeted therapies as antitumor drugs. [1][2][3] As a privileged scaffold, the 4-aminoquinazoline scaffold is an interesting pharmacophore, and widely used for making many specific kinase inhibitors, including tyrosine kinase, in medicinal chemistry for the treatment of lung, breast, colorectal cancers, etc. (►Fig. 1). [4][5][6] Despite the clear clinical benefits, the efficacy of tyrosine kinase inhibitors (TKIs), such as Gefitinib or Erlotinib, in cancer treatment is eventually diminished because of acquired point mutations in the kinase domain of the estimated glomerular filtration rate (EGFR), and the subsequent induction of carcinogenic events, including ALK rearrangements, vascular EGFR (VEGFR) amplifications, platelet-derived growth factor receptor (PDGFR) amplifications, and so on. [7][8][9][10][11][12] To our acknowledge, cancer has multiple pathologies with unique molecular signatures, thus, developing novel multitargeted TKIs targeting different signaling pathways and oncogenic drivers will be more efficient and beneficial than those obtained with single-target approaches. [13][14][15][16] It is unambiguously confirmed that the quinazoline core binds to a narrow hydrophobic pocket in the N-terminal domain of EGFR protein, which is a common pharmacophoric feature of EGFR TKIs containing 4-aminoquinazoline core. The N-1 and N-3 nitrogen atoms of the quinazoline ring interact with amino acid residue NH or a bridging water through H-bonding, which played a crucial part in binding and the subsequent inhibitory capacity. Meanwhile, the 4-anilino group occupied a deep pocket at the back of the ATP-binding site and the C6 and/or C7-positions of quinazoline extended toward the entrance of the binding site of the protein. [17][18][19] Up to now, most approaches have been adopted to enhance the potency and selectivity of these quinazoline derivatives in terms of differing functional groups on the C6 and C7 regions or changing the substituents on the 4-anilino group. [20][21][22][23][24][25][26] However, the exploration of the flexibility limitation of the NH spacer on 4-aminoquinazoline score is rarely reported. In fact, it would be essential to know how the flexibility of the NH spacer affects the TKI's inhibitory activity for further drug design.
In this article, we aim to develop a novel TKI skeleton according to the binding model of Erlotinib with EGFR protein.
We designed and synthesized a novel series of quinazoline derivatives by means of conformational restriction for NH spacer and evaluated their antitumor activities (►Fig. 2).

Results and Discussion
The synthesis of series A and B target compounds is shown in Scheme 1. Coupling of chloride 1 with 4-substituted indolines 2a-2d gave the desired compounds A1-A4. Chloride 1 was coupled with substituted 2-aminobenzyl alcohol 3a-3m to give the key intermediates 4a-4m. Intramolecular cyclization of 4a-4m with phosphoryl chloride afforded the desired compounds B1-B13.
The effects of the target compounds on proliferation of different cancer cells were determined using MTT assay according to a reported study. 23 Taxol and Erlotinib served as control drugs. Cancer cells used in this assay included K562 (human chronic myelogenous leukemia cells), Jurkat (E6-1) (human T lymphocyte leukemia cells), PANC-1 (human pancreatic carcinoma cells), A549 (human non-small cell lung cancer cells), Colo320 (human colorectal adenocarcinoma cells), Hut78 (human cutaneous lymphoma cells), MDA-MB-435s (human breast ductal carcinoma cells), Hep3B (human hepatoma cells), and PC-3 (human prostate cancer cells); and the concentration of the target compounds (A1-A4 and B1-B13) were 10 μmol/L. As shown in ►Table 1, most target compounds exhibited moderate to potent antiproliferative activities against the tested The IC 50 values of the target compounds for EGFR activity were also determined. ►Table 1 shows that incorporation of N-containing indoline substituents to yield compounds A1-A4 resulted in the complete loss of EGFR inhibitory activities (IC 50 > 10 μmol/L). Interestingly, compounds B1-B13, only bearing a N-1 nitrogen as H-bond donor in the quinazoline fragment, maintained or slightly enhanced inhibitory activity against EGFR, indicating that N-3 nitrogen may not be critical in the common binding mode.
Besides, the preliminary structure-activity relationship of B series analogues with substituted groups at the phenyl nucleus was further investigated. Our data showed that the position of the substituents on the phenyl moiety had great effect on EGFR activities. For instance, the introduction of the same substituent at 11-position, in comparison with that at 9/10/12-position, could significantly improve the inhibitory activity against EGFR (compound B5 vs. B1, B3, B9, and B11). Meanwhile, a chlorine substituent on 11-position, in comparison with other substituents, showed stronger EGFR inhibition (compared B5 with B6, B7, and B8). Among them, compound B5 was the most potent compound (IC 50 value was 0.23 nmol/L), in particular, which was over twofold more potent than Erlotinib (IC 50 value was 0.56 nmol/L). Excitingly, compound B5 demonstrated the best inhibitory activity against cell assay for K562 cells, A549 cells, and MDA-MB-435s cells, and the effect of which was significantly superior to that of Erlotinib.
Accordingly, we were prompted to investigate whether this potency was specifically against EGFR. As shown in ►Fig. 3, compound B5 strongly inhibited the kinase activity of FLT3, VEGFR2, and PDGFRα at the test concentration of 1 μmol/L (inhibition ratio > 50%) with IC 50 values being 0.23, 0.44, and 0.20 μmol/L, respectively. Apparently, compound B5 was an anticancer multitargeted protein kinase inhibitor, which remarkably inhibits cell proliferation by targeting multiple kinases.
On the basis of the above results, we conducted acute toxicity test and hERG automated patch clamp (QPatch) assay  to examine the ancillary safety profiles of compound B5 as an anticancer agent. As is apparent from the data in ►Table 2, compound B5 exhibited low toxicity to ICR mice with maximum tolerance dose value over 1,500 mg/kg by oral administration. Meanwhile, compound B5 displayed weak inhibitory activity against hERG (IC 50 ¼ 10.46 μmol/L), similar to that of Erlotinib (IC 50 ¼ 14.26 μmol/L). The pharmacokinetic studies of compound B5 and Erlotinib were performed using female Sprague Dawley rats when dosed intravenously at both 1 mg/kg and orally at 20 mg/kg (►Table 2). The terminal phase half-life of B5 after po (oral) dosing was 2.32 hours. In contrast, po administration of Erlotinib led to a similar half-life (t 1/2 ¼ 2.65 hours). B5 was quickly absorbed after oral dosing, with a T max of 0.44 hour. The oral bioavailability (F) of B5 was 21.08%, which is lower than that of Erlotinib (F ¼ 66.60%).
Given these promising pharmacokinetic parameters, the most promising compound B5 named SIPI6473 was further evaluated in the non-small cell lung cancer A549 xenograft mouse model at daily oral doses of 37.5, 75, and 150 mg/kg for 21 days (►Fig. 4). As shown in ►Fig. 4, SIPI6473 could suppress tumor growth in a dose-dependent manner with tumor growth inhibition (TGI) ratios being 35.93, 44.44, and 66.1%, respectively. Our data showed that the po administration of SIPI6473 at a high dose of 150 mg/kg resulted in a superior antitumor effect when compared with Erlotinib (TGI ¼ 55.41%, 150 mg/kg).

Conclusion
We designed and synthesized a series of novel quinazoline derivatives by conformational restriction. Among them, compound B5 (also named SIPI6473) showed strong proliferation activity in several cellular assays. SIPI6473 not only showed most potent EGFR inhibition activity but also exhibited nanomolar level inhibition against several other NSCLCrelated oncogene kinases, including VEGFR2 and PDGFRα. In addition, in vivo studies showed that SIPI6473 had a good pharmacokinetic profile and displayed significant antitumor activity against A549 xenografts without signs of adverse effects. Hence, this new scaffold of potent protein kinase inhibitor SIPI6473 demonstrated a promising prospect for the discovery and development of new NSCLC agents and drugs. Further evaluation of SIPI6473 is ongoing and will be reported in due time.

Experimental Section
General Synthetic Procedure of A1-A4 The chloride 1 (1 mmol), indoline (1.2 mmol), and K 2 CO 3 (3 mmol) were dissolved in 15 mL DMF and reacted at 130°C for 4 hours. After the reaction was complete, the reaction solution was poured into ice water, extracted with ethyl acetate (50 mL Â 2), combined with organic phase, washed with water, saturated sodium carbonate solution, and saturated salt water (10 mL Â 2), dried with anhydrous sodium sulfate, and concentrated, and by column chromatography (DCM: MeOH ¼ 50: 1-10: 1) a gray white solid A1 was obtained.  Abbreviations: MTD, maximum tolerance dose; PK, pharmacokinetics; po, oral. a hERG patch clamp screen as described in Dubin et al. 27 IC 50 values represent the concentration to inhibit 50% of hERG current (IKr). Numbers represent IC 50 values generated from 3-point concentration response relationships in duplicate. b For pharmacokinetic study, blood was collected from rats at various time points up to 24 hours, and plasma samples were analyzed using an Agilent 1200 HPLC system coupled with an Agilent 6410B triple quadruple mass spectrometer. A solution of 0.05 N HCl in saline was used as the vehicle for both intravenous and oral dosing. c The symbol "-" means that the MTD was not measured.

General Synthetic Procedure of B1-B13
The chloride 1 (1.0 mmol) and 2-aminophenylethanol (1.0 mmol) were dissolved in 10 mL isopropanol, heated to reflux, and reacted for 6 hours. The reaction solution changed from turbid to clear and finally turbid. The white solid was obtained by recrystallization from ethyl acetate and then was dissolved in 5 mL of phosphorus oxychloride and heated to reflux for 3 hours. The heating was stopped, the unreacted phosphorus oxychloride evaporated, and the residue was added to ice water. pH was adjusted to 10 with saturated sodium carbonate solution. The mixture was extracted with ethyl acetate for 3 times, and the combined organic phase was washed with water, saturated salt water, and dried with anhydrous magnesium sulfate. After filtration, the filtrate was taken and concentrated under reduced pressure to obtain crude product. It was then recrystallized using ethanol to obtain B1 as a white solid with a yield of 50.47%. NMR (600 MHz, CDCl 3 þ CD 3 OD LOCK CDCl 3 ) d 7.92 (s, 2H), 6.93 (s, 2H), 5.27 (s, 2H), 4.10 (m, 4H), 3.64-3.60 (m, 4H), 3.25 (s, 3H), 3.23 (s, 3H).