Enantioselective Synthesis of (+)-Coerulescine by a Phase-Transfer Catalytic Allylation of Diphenylmethyl tert-Butyl α-(2-Nitrophenyl)Malonate

A 7-step enantioselective synthetic method for preparing (S)(+)-coerulescine is reported through the use of diphenylmethyl tert-butyl α-(2-nitrophenyl)malonate (16% overall yield, >99% ee). Allylation is the key step under phase-transfer catalytic conditions (86% ee). This synthetic method can be used as a practical route for the synthesis of various derivatives of (S)(+)-coerulescine for analyzing its structure–activity relationships against its biological activities.


INTRODUCTION
A spirooxindole (Marti and Carreira, 2003) is a key structure of coerulescine, horsfiline, elacomine, spirotryprostatin B, and strychnofoline that has various biological activities (Galliford and Scheidt, 2007). Among the spirooxindole alkaloids, the relatively simple structure of both coerulescine and horsfiline has become attractive to many medicinal chemists as a new pharmaceutical skeleton (Figure 1). In particular, coerulescine can derivatize an aromatic moiety in oxindole via halogenation (De et al., 2015). (-)-Coerulescine was first isolated in 1998 from Pharalis coerulescens by Colegate et al. (Anderton et al., 1998). The construction of the quaternary stereogenic center of (-)-coerulescine has been quite challenging, and until now, only four synthetic methods have been reported for the synthesis of enantiomerically enriched coerulescine.
As part of our program to develop a new antitumor agent, we chose to derivatize coerulescine as a spirooxindole scaffold. The systematic investigation of the biological activity of coerulescine has not yet been performed due to the limited supply from natural isolation. In this paper, we report our efforts toward an efficient method for the synthesis of unnatural (+)-coerulescine via an enantioselective PTC (Jew and Park, 2009;Shirakawa and Maruoka, 2013) α-allylation of malonate, which is our precedent research for our studies on the structure-activity relationship and chirality-activity relationship.

RESULTS AND DISCUSSION
As shown in the retrosynthetic analysis (Scheme 2), we employed two pathways [N-methylpyrrolidine (A) and oxindole (B)] for the construction of the spirooxindole skeleton from 4. The corresponding methoxy-substituted N-methylpyrrolidine was a key intermediate for (-)-horsfiline in our previous report (Hong et al., 2013). We also attempted to convert the oxindole intermediate B to coerulescine by intramolecular lactamization.
For the first step, the enantioselective allylation of 1 was performed under the reported PTC conditions (Hong et al., 2011). The PTC allylation of 1 was performed with allyl bromide (10.0 equiv.), 50% KOH (aq., 5.0 equiv.), and (R, R)-3 at 0 • C in toluene to produce the allylated product (S)-4 (83%, 82% ee). The optimization of the PTC conditions was achieved by varying the following parameters: solvent, base, temperature, and the amount of catalyst 3 ( Table 1). Both enantioselectivity and chemical yield were dependent on the solvent and base conditions at 0 • C (entries 1-4). The highest enantioselectivity and chemical yield were observed with 50% KOH base in toluene (entry 1). In the case of temperature, lower reaction temperatures showed higher enantioselectivities (entry 1, entries 5-6). However, poor SCHEME 1 | Enantioselective PTC α-alkylation of malonate. SCHEME 2 | Retrosynthetic analysis.
conversion was observed at −40 • C and resulted in a low chemical yield (entry 6). The optimal amount of catalyst 3 was 5 mol% (entry 5, entries 7-8). After all the tests, the following optimal allylation conditions were selected: 50% KOH aq. base and (S, S)-3 (5 mol%) in toluene at −20 • C (entry 7; 87% yield, 86% ee). The enantioselectivity of the allylation of 1 in this study was slightly lower than that of α-phenyl case (Hong et al., 2011). We speculate that such a low enantioselectivity is caused by the steric effect of the ortho-nitro group, which may form an unfavorable binding conformation with the PTC catalyst 3.
Ozonolysis of 4, followed by a reduction using sodium borohydride, afforded lactone 6 (95%) via aldehyde 5 (93%) (Scheme 3). Oxindole 7 was prepared by hydrogenation of 6 in the presence of Raney Ni/H 2 and the following mesylation (8) under trimethylamine basic conditions (Scheme 4). However, the treatment of methylamine (33 wt % in absolute EtOH) in various solvents such as DMF, CH 3 CN, CH 2 Cl 2 , and DMSO at room temperature did not give the expected amine product 9 or the further cyclized N-methylpyrrolidine analog 12. Only the starting material 8 was recovered. The increase of temperature led to so many side products. By another route, we tried to prepare N-methylbutyrolactam 12, which could be selectively reduced to produce coerulescine (Scheme 5) (Trost and Brennan, 2006). The ring opening of lactone 6 with N-methylamine afforded N-methylamide 10. The mesylation of 10 followed by an intramolecular lactamization successfully provided 11. However, the catalytic hydrogenation under Pd/C, Pd(OH) 2 , PtO 2 , and Raney Ni under H 2 (1 atm) gave no oxindole 12. We speculate that the unsuccessful results were due to the carbonyl group in lactam moiety. Finally, we adapted our previous synthetic route of horsfiline, as shown in Scheme 6. Without purifying 6, which could be prepared from 5, the treatment of additional sodium borohydride with cerium (III) trichloride heptahydrate and tetrahydrofuran as a cosolvent at 0 • C selectively reduced 6 to the corresponding diols 13 (52% from 5) in situ (Martin et al., 2010). Diol 13 was purified as a single stereoisomer (>99% ee) with a 45% yield by recrystallizing (86% ee) using ethylacetate and hexane (1:5). Dimesylation of 13 (94%) followed by N-alkylation followed by intramolecular N-alkylation in the presence of excess methylamine successfully produced N-methylpyrrolidines 15 (99%). The reduction of the nitro group on 15 by a catalytic hydrogenation under Pd/C and atmospheric H 2 afforded amine 16. Finally, the prepared amine 16 was directly cyclized to (+)coerulescine {observed [α] 20 D = 3.08 (c 1, MeOH); literature [α] 20 D = 1.0 (c 2.4, MeOH) )} by stirring with silica gel (SiO 2 ) in CH 2 Cl 2 with no racemization (90% from 15, >99% ee).

CONCLUSION
In summary, as a precedent study for systematic research of biological activities, enantioselective synthetic routes of (+)coerulescine were investigated. (+)-Coerulescine was prepared SCHEME 3 | Enantioselective PTC α-allylation and conversion to lactone 6. through seven steps from diphenylmethyl tert-butyl α-(2nitrophenyl)malonate via an enantioselective PTC allylation as the key step (16% overall yield, >99% ee). The large scalable synthetic method of coerulescine enables a systematic investigation of its antitumor activity. Further preparation of derivatives of coerulescine is now under investigation for structure-activity relationship studies.

General Information
All reagents purchased from commercial sources were used without further purification. The phase-transfer catalyst, (R,R)-3,4,5-trifluorophenyl-NAS bromide (3), was purchased from commercial sources. TLC analyses were performed using pre-coated TLC plate (silica gel 60 GF254, 0.25 mm). Flash column chromatography was performed on flash silica gel 230-400 mesh size. The values of enantiomeric excess (ee) of chiral products were determined by HPLC using 4.6 × 250 mm DAICEL Chiralpak AD-H and Chiralpak AS-H. Infrared analyses (neat) were performed by FT-IR. 1 H-NMR spectra were recorded at 400 MHz with reference to CHCl 3 (δ 7.24). 13 C-NMR spectra were obtained by 100 MHz spectrometer relative to the central CDCl 3 (δ 77.0) resonance. Coupling constants (J) in 1 H-NMR are in hertz. Low-resolution mass spectra and highresolution mass spectra (HRMS) were measured on positive-ion SCHEME 4 | Route via oxindole 8. FAB spectrometer. Melting points were measured on melting point apparatus and were uncorrected. Optical rotations were measured on a polarimeter and calibrated with pure solvent as blank.