Discovery of dihydrooxazolo[2,3-a]isoquinoliniums as highly specific inhibitors of hCE2

Human carboxylesterase 2 (hCE2) is one of the most abundant esterases distributed in human small intestine and colon, which participates in the hydrolysis of a variety of ester-bearing drugs and thereby affects the efficacy of these drugs. Herein, a new compound (23o) with a novel skeleton of dihydrooxazolo[2,3-a]isoquinolinium has been discovered with strong inhibition on hCE2 (IC50 = 1.19 μM, Ki = 0.84 μM) and more than 83.89 fold selectivity over hCE1 (IC50 > 100 μM). Furthermore, 23o can inhibit hCE2 activity in living HepG2 cells with the IC50 value of 2.29 μM, indicating that this compound has remarkable cell-membrane permeability and is capable for inhibiting intracellular hCE2. The SAR (structure–activity relationship) analysis and molecular docking results demonstrate that the novel skeleton of oxazolinium is essential for hCEs inhibitory activity and the benzyloxy moiety mainly contributes to the selectivity of hCE2 over hCE1.


Introduction
Mammalian carboxylesterases (CEs), important members of the serine hydrolase superfamily widely distributed in the lumen of endoplasmic reticulum in various tissues, are responsible for the hydrolysis of a wide range of endogenous and xenobiotic substrates containing ester, amides, thioesters and carbamates. [1][2][3] In human body, hCE1 and hCE2 are the main carboxylesterases, both of which play crucial roles in endo-and xenobiotic metabolism. As one of the most abundant esterases distributed in human small intestine and colon, hCE2 participates in hydrolysis of the ester-bearing drugs (such as irinotecan, prasugrel, capecitabine, utamide) and thereby affects the efficacy of these drugs. [4][5][6][7] For instance, CPT-11 (irinotecan), an anticancer prodrug, exhibits strong anti-colorectal cancer activity by releasing the effective substance SN-38. However, excessive accumulation of SN-38 in the intestinal mucosa leads to delayed-onset diarrhoea even death. [8][9][10] To improve the potential clinical risk of these drugs, some highly specic hCE2 inhibitors have been used in clinical to reduce the local exposure of SN-38 in the intestinal mucosa, thereby ameliorating the intestinal toxicity of 12 Over the past decade, a wide variety of hCE2 inhibitors have been reported, including the natural triterpenoids, 13,14 avonoids, 13-15 1,2-diones 16,17 and etc. Although many compounds with strong hCE2 inhibitory activities have already been developed, the potent and specic inhibitors targeting intracellular hCE2 are still rarely reported. DCZ0358 (Fig. 1) is a novel dihydrooxazolo[2,3-a]isoquinolinium discovered in the synthesis of berberine analogues. [18][19][20] Preliminary screening indicated that DCZ0358 could effectively inhibit the catalytic activity of both hCE1 (IC 50 ¼ 4.04 mM) and hCE2 (IC 50 ¼ 16.03 mM), while its hydrolyzate 23b showed a signicant reduction of the inhibitory activity (hCE1 IC 50 ¼ 36.80 mM; hCE2 IC 50 ¼ 41.75 mM), which demonstrated that the oxazolinium moiety of DCZ0358 is essential for the CE S inhibitory activity (Fig. 1). In the synthesis of derivatives of DCZ0358, we have found that in addition to compound 23d (Fig. 2), other compounds with modication of the substituents on the A and D rings cause structural instability of the quaternary ammonium salt. Moreover, the bioactivity and selectivity of 23d were improved (for hCE2 IC 50 ¼ 6.889 mM with >14.52-fold selectivity over hCE1). These results encouraged us to make further investigation of the structure-inhibition relationships of these berberine analogues as CEs inhibitors.
The previously reported synthetic route of DCZ0358 is inconvenient to prepare more derivatives because of the harsh reaction conditions (Scheme 1). 21 Therefore, we designed a new synthetic route using compound 12 as the key intermediate (Scheme 2). Among the obtained new analogues, 23o showed the highest selectivity and the best inhibitory activity (hCE1 IC 50 > 100 mM; hCE2 IC 50 ¼ 1.192 mM, K i ¼ 0.84 mM). It was also found that 23o could inhibit hCE2 activity in living HepG2 cells with the IC 50 value of 2.29 mM, suggesting that the compound has remarkable cell-membrane permeability and is capable for inhibiting intracellular hCE2. Further molecular docking results showed that the methoxyl group at the benzyloxy ring of 23o could tightly bind to the catalytic amino acid Ser-228 via Hbonding, which may account for the high selectivity of 23o on hCE2 over hCE1.

Synthetic procedures
Previously, we reported the synthetic route of DCZ0358 (Scheme 1). 21 However, the application of n-butyl lithium reagent and low temperature condition (À78 C) restricted the synthesis of derivatives. Therefore, developing a feasible route is important for the further medicinal chemistry research. Based on the retrosynthetic analysis (Scheme 1), compound 9 could be synthesized via Suzuki coupling reaction from 10 and 11. Compound 10 could be smoothly prepared from the key intermediate 12.
Thus, we developed another route taking commercially available 3,4-dimethoxybenzaldehyde 1 as the starting material (Scheme 4). Compound 1 reacted with DMF and formic acid to afford tertiary amine 2 in 75% yield. 25 Then we added chloroformate to the mixture of 2 and n-butyl lithium under À78 C to produce 16 in 80% yield. 26,27 Next, compound 16 was attracted by electrophilic reagent TMSCN to afford 17 (82% yield). 28 The operation for the hydrolysis of the methyl ester compound 17 to the compound 15 is difficult to be control. Subsequently, both ester and cyano groups were hydrolyzed to carboxyl groups under strong alkaline condition to give 18 (76% yield). Compound 18 was easily dehydrated in the presence of acetyl chloride to obtain compound 19 in 78% yield. 29 However, compound 20 was rather difficult to achieve from compound 18 or compound 19. Aer trying various amines, we found that only ammonium carbonate could react with 19. 30 However, this reaction occurred at a high temperature (280 C) and gave a very low yield (22% yield) of 20. Thus, compound 17 was directly reacted with sodium methoxide to afford compound 21 in 51% yield, followed by demethylation to produce dihydroisoquinoline-1,3-dione 20 with a high yield of 93%. 31 In order to convert 20 to the key intermediate 12, we explored many reagents, such as PCl 5 , POCl 3 , SOCl 2 and PhPOCl 2 , it turned out that PhPOCl 2 behaved the best yield with 47%. 31 The key intermediate 12 reacted smoothly with hydroxyacetal under alkaline conditions to give compound 10 with high yield (98%), 32 and then 10 reacted with various arylboronic acids containing a benzyloxy structure to produce 22 in yields ranging from 46% to 98%. 19 Finally 22 were cyclized under acidic conditions to give a series of dihydrooxazolo[2,3-a]isoquinolinium analogues (Scheme 5, compounds 23d-23o in Fig. 2). The present synthetic route is convenient to scale up and benets further pharmaceutical research.

Biological activity assays
We designed and synthesized more than 30 derivatives of DCZ0358. However, the ve-ring quaternary ammonium component of some derivatives was unstable to decompose easily into its hydrolyzate 23b. With 12 stable compounds in hand, we conducted experiments to assay inhibitory activities against both hCE1 and hCE2 using a panel of uorescent probe substrates. 33-36 D-Luciferin methyl ester (DME) was used as a probe substrate, and nevadensin (a specic hCE1 inhibitor) was used as a positive inhibitor control for hCE1. Fluorescein diacetate (FD) was used as a specic probe substrate, and loperamide (LPA) was used as a positive inhibitor control of hCE2. The IC 50 values of all derivatives were evaluated and listed in Table 1. Table 1 showed that the inhibitory effects of these compounds against hCE2 were enhanced signicantly when the methylenedioxy group on A ring was changed into benzyloxy group.  with electron-donating groups on the benzyloxy ring were similar to that of 23e (hCE2 IC 50 11.46 AE 1.76 mM), 23f (hCE2 IC 50 5.73 AE 0.79 mM) and 23h (hCE2 IC 50 3.32 AE 0.87 mM) with electronwithdrawing groups. In terms of the selectivity, it improved apparently according to the values of IC 50 (hCE2)/IC 50 (hCE1) shown in Table 1. For instance, the value of IC 50 (hCE2)/IC 50 (hCE1) of 23o was up to 83 while that of 23a was only 0.25. Thus, 23o have the best selectivity on hCES2 among all these newly synthesized compounds.
Collectively, the structure-activity relationships of these compounds were summarized as follows, (1) the oxazolinium moiety is crucial for the inhibitory activity against hCEs; (2) the benzyloxy group on the A ring mainly contributed to the selectivity of hCE2 over hCE1 (Fig. 3).
The inhibition kinetic of 23o against hCE2-mediated FD hydrolysis has been carefully investigated and the results showed that 23o functioned as a mixed inhibitor against hCE2mediated FD hydrolysis, with the K i value of 0.84 mM (Fig. 4B). Furthermore, in view of that hCE2 is an intracellular enzyme, the inhibition potential of 23o was also investigated. As shown in Fig. 5, 23o could strongly inhibit intracellular hCE2-mediated NCEN hydrolysis and reduce the uorescence intensity in the green channel (for the hydrolytic metabolite of NCEN) in living HepG2 cells via a dose-dependent manner. Meanwhile, the IC 50 value of 23o against intracellular hCE2 was also evaluated as 2.29 mM (Fig. S2B †).

Molecular docking
In order to investigate the interaction mechanism of 23o with hCE2, molecular docking of 23o to the active site of hCE2 was performed. As shown in Fig. 6, there are hydrogen bond between the methoxyl of ring D with Arg-355 (3.16Å), and a Ttype p-p interaction between the ring D with the Arg-355, as well as, hydrogen bond between the oxygen atom of ring B with Phe-307 (3.17Å) in the entrance of the active cavity of hCE2. These interactions facilitate the entry of 23o into the active cavity of hCE2. However, the hydrolysate of 23o cannot enter the active cavity of hCE2, due to its small inlet. In addition, the methoxyl group at the benzyloxy end of 23o could tightly bind to the catalytic amino acid Ser-228 (1.6Å) via strong H-bonding, as well as, with Ala-150 (3.18Å), and there are strong hydrophobic interactions between the benzyloxy group of 23o with the key residues in the active cavity of hCE2. These interactions may account for the high selectivity of 23o on hCE2. The strong Hbond interaction between 23o and Ser-228 indicates that 23o may obstruct hCE2-mediated hydrolysis, possibly because Ser-    228 is an important residue involved in substrate recognition and catalysis of hCE2. These ndings agreed well with the experimental data where 23o exhibited much more potent inhibitory effect on hCE2 but a relatively weaker one on hCE1.

Conclusions
A new compound 23o with a novel skeleton of dihydrooxazolo [2,3-a]isoquinolinium was discovered with good inhibitory activity on hCE2 (IC 50 ¼ 1.19 mM, K i ¼ 0.84 mM) and high selectivity over hCE1 (IC 50 > 100 mM). The SAR (structureactivity relationship) analysis and molecular docking results revealed that the novel oxazolinium moiety is essential for hCE2 inhibitory activity, while the benzyloxy moiety contributes to the selectivity of hCE2 over hCE1. Furthermore, 23o could strongly inhibit intracellular hCE2 in living HepG2 cells, with the IC 50 value of 2.29 mM. These ndings are important for further research and development of hCE2 inhibitors with high speci-city and efficacy.

Chemical synthesis
Materials. All starting materials were obtained from commercial suppliers and used without further purication. The 1 H and 13 C NMR spectra were taken on Bruker Avance-600 or 500 or 400, Varian MERCURY Plus-400 or 300 NMR spectrometer operating at 400 MHz or 300 MHz for 1 H NMR, 125 MHz or 100 MHz for 13 C NMR, using TMS as internal standard and CDCl 3 or methanol-d 4 or DMSO-d 6 as solvent. 13 C NMR spectra were recorded with complete proton decoupling. The ESI-MS or EI-MS was recorded on Finnigan LCQ/DECA or Thermo-DFS, respectively. The HRMS were obtained from Micromass Ultra Q-TOF (ESI) or Thermo-DFS (EI) spectrometer. Flash column chromatography was carried out using silica gel (200-400 mesh). Thin layer chromatography (TLC) was used silica gel F254 uorescent treated silica that were visualised under UV light (254 nm).
Synthetic procedure. Compounds DCZ0358 and 23b have been reported in our previous work. 18,20 Synthesis of 3,9,10-trimethoxy-5-(4-methoxy-3-((4-methoxybenzyl)oxy)phenyl)-2,3dihydrooxazolo[2,3-a]isoquinolin-4-ium (23d). To a solution of 22d (56 mg, 0.1 mmol) in acetone (5 mL) was added hydrochloric acid (1 mL, 2.0 M in diethyl ether), and then the mixture was stirred for 2 h at room temperature. The solution was evaporated in vacuo to obtain the titled compound 23d as yellow solid (46 mg, 83%). 1 161.2, 159.7, 152.9, 149.9, 147.3, 146.2, 136.8,  134.5, 130.9, 130.1, 126.1, 124.9, 124.5, 120.9, 118.8, 116 3,9,10-Trimethoxy-5-(4-methoxy-3-((4-(methylsulfonyl)benzyl) oxy)phenyl)-2,3-dihydrooxazolo[2,3-a]isoquinolin-4-ium (23e). Compound 23e was prepared from compound 22e (58 mg, 0.1 mmol) as a yellow solid (48 mg, 86%). 1    MD, USA) and stored at À80 C until use. DMSO was purchased from sher. Phosphate buffer was prepared using Millipore water and then stored at 4 C until use. All tested compounds were solved by DMSO and stored at 4 C until use. LC grade acetonitrile and DMSO (Tedia, USA) were used throughout. hCE1 inhibition assay. DME was used as a probe substrate for evaluating the inhibitory effects of all DCZ0358 derivatives on hCE1, while nevadensin (a specic hCE1 inhibitor) was used as a positive control. 39 Briey, 100 mL incubation mixture containing 91 mL PBS (pH 6.8), 2 mL inhibitor at different concentrations and 5 mL HLM (1 mg mL À1 , nal concentration), were pre-incubated at 37 C for 10 min. Subsequently, 2 mL DME (3 mM nal concentration, close to the K m value of DME in HLM) was added to initiate the reaction. Aer incubating for 10 min at 37 C in a shaking bath, the reaction was stopped by the addition of LDR (100 mL). The microplate reader (SpectraMax® iD3, Molecular Devices, Austria) was used for luminescence measurements. The gain value of luminescence detection was set at 140 volts, and the integration time was set at 1 s. The chemical structure of DME and its hydrolytic metabolite (Dluciferin), as well as the detection conditions for D-luciferin are depicted in Table S1. † The negative control incubation (DMSO only) was carried out under the same conditions. The residual activity was calculated using the following formula, the residual activity (%) ¼ (the orescence intensity in the presence of inhibitor)/the orescence intensity in negative control Â 100%. The residual activities are show in Fig. S1. † hCE2 inhibition assay. The inhibitory effects of all DCZ0358 derivatives on hCE2 were investigated using uorescein diacetate (FD) as a specic probe substrate, 40 while LPA was used as a positive inhibitor of hCE2 in this study. 41 In brief, 200 mL incubation mixture containing 0.1 M PBS (PH ¼ 7.4), human liver microsomes (2 mg mL À1 , nal concentration) and each inhibitor. Aer 10 min pre-incubation at 37 C, the reaction was initiated by adding FD (5 mM, nal concentration, close to the K m value of FD in HLM). Aer incubating for 30 min at 37 C in a shaking bath, the reaction was stopped by the addition of acetonitrile (200 mL). The chemical structure of FD and its hydrolytic metabolite (uorescein), as well as the detection conditions for uorescein are depicted in Table S1. † The negative control incubation (DMSO only) was also carried out under the same conditions. 42 The residual activity was calculated using the formula mentioned above in hCE1 inhibition assay. The residual activities are shown in Fig. S1. † Cell culture and uorescence imaging analyses. In view of that hCE2 was an intracellular enzyme, the inhibition potential of 23o was investigated in living HepG2 cells. The HepG2 cells were cultured in Modied Eagle's Medium (MEM) with 5% CO 2 and 0.1% antibiotic-antimycoticmix antibiotic at 37 C, supplemented with 10% fetal bovine serum (FBS) and used NCEN as substrate probe to assay the 23o inhibition potential toward hCE2. NCEN, 43 another specic optical probe substrate for hCE2, the structure and hydrolytic site were shown in Fig. S2(A). † For uorescence imaging, HepG2 cells were seeded in 96well plates (8000 cells per well) with complete medium and then incubated for 24 hours. Aerwards, the cells were washed twice with FBS-free culture medium and then preincubated in the medium containing 23o (prepared in FBS-free at various concentrations) for 30 min with 5% CO 2 at 37 C. HepG2 cells were then co-incubated with NCEN (nal concentration, 10 mM) for another 50 min to assess the intracellular hCE2 function, respectively. The living cells were imaged and analyzed using an ImageXpress® Micro Confocal High-Content Imaging system (Molecular Devices, Austria).

Conflicts of interest
There are no conicts to declare.