Synthesis of fluorinated donepezil by palladium-catalyzed decarboxylative allylation of α-fluoro-β-keto ester with tri-substituted heterocyclic alkene and the self-disproportionation of its enantiomers

Abstract

The synthesis of fluorinated donepezil, a promising new therapeutic agent for Alzheimer's disease, was achieved by the palladium-catalyzed decarboxylative allylation (DcA) of an allylic ester having a tri-substituted heterocyclic alkene system as a key step. This strategy is very different from a patented method for the title compound which was successfully extended to the first catalytic asymmetric synthesis of fluorinated donepezil. Fluorinated donepezil showed a noticeable magnitude in the self-disproportionation of enantiomers.

---

Donepezil¹ is a reversible inhibitor of the enzyme acetylcholinesterase and has been used worldwide as one of the most important drugs for the treatment of Alzheimer-type diseases (Fig. 1a). Donepezil was originally developed by Eisai Ltd and has continually maintained an important position in the drug market for Alzheimer-like diseases. The side effects of donepezil have also been reported.² In this context, we are interested in the chiral versions of donepezil. Donepezil has one chiral center, thus there are (R) and (S)-enantiomers. However, it has been launched as a racemate. According to the reports by Eisai Ltd, there are not many differences in the biological activities of the (R) and (S)-enantiomers of donepezil (IC₅₀ of 5.3 nM for the (R)-enantiomer, IC₅₀ of 7.7 nM for the (S)-enantiomer).³ The enzyme-binding mode of donepezil was thus analyzed on the basis of the crystal structure⁴ of the enzyme using docking programs and these results are in agreement with the data that show that both enantiomers are equally active,^3b while only the (R)-enantiomer of donepezil was crystalized by binding within the active-site gorge. A clear difference in the plasma concentration of two donepezil enantiomers in patients with Alzheimer's disease was also reported,⁵ and the mean plasma (S)-donepezil peak concentrations were 2.6-fold higher than for (R)-donepezil. Further studies are expected to investigate the binding,⁶ stereospecific metabolism and biological activity of enantiomerically pure donepezil. However, the problem is that donepezil enantiomers are racemized⁷ via the same enol form of donepezil induced by hydrogen-bonding,⁸ making an enantiospecific biological study of donepezil meaningless. Incidentally, we developed fluorinated thalidomide as a non-racemized isostere of thalidomide and its biological activity was examined in 1999.⁹ Accordingly, we designed fluorinated donepezil 1a as the first non-racemized isostere of donepezil (Fig. 1b).¹⁰

Fig. 1 (a) Donepezil, (b) fluorinated donepezil 1.

The idea of non-racemized fluorinated donepezil was first realized and patented by one of us (NS) and others in 2000, more than 15 years ago.¹⁰ As expected,¹¹ fluorinated donepezil 1a was found to be more potent, i.e., IC₅₀ of 1.3 nM against acetylcholinesterase, than the original donepezil (IC₅₀ = 5.9 nM).¹⁰ The OH-analogue 1b shows an even more potent IC₅₀ of 0.32 nM.¹² Later, its derivative 1c was optically resolved by HPLC with a chiral column using CHIRALPAK AD as the stationary phase to afford (+)- and (−)-isomers of 1c. Interestingly, IC₅₀ values of (+)-1c and (−)-1c for rat brain acetylcholinesterase were very different for each isomer, 0.20 nM and 12 nM, respectively.¹³ These results indicate that the (+)-isomer 1 is 60 times more potent than the (−)-isomer 1. This spurred us to achieve the enantioselective synthesis of fluorinated donepezil 1 for further medicinal experiments. Although our original patented method for the synthesis of 1 is the direct generation of lithium enolate of donepezil by LDA followed by electrophilic fluorination using N-fluorobis(phenylsulfonyl)amine (NFSI) at a low reaction temperature,¹⁰ the method is applied with difficulty to the enantioselective reaction. Therefore, an alternative synthetic method of 1 that is applicable for catalytic asymmetric synthesis has been awaited over the past decade.¹³ We disclose herein an efficient method for the catalytic synthesis of fluorinated donepezil 1a from dimethoxy indanone carboxylate 2 with heterocyclic allylic alcohol 3, and its application to asymmetric synthesis (Scheme 1). The key for the synthesis of 1a is a palladium-catalyzed decarboxylative allylation (DcA) of the tri-substituted heterocyclic allylic ester of α-fluoro-1-indanone-2-carboxylate 5. The ferrocene ligand 1,1′-bis(diphenylphosphino)ferrocene (dppf) was found to be suitable for this DcA reaction in high yield, and the asymmetric version of this reaction was investigated using chiral ligands. Finally the fine-scale combination of Pd₂dba₃ (5 mol%), (R,R_p)-Ph-Taniaphos (L5, 12.5 mol%), and H₂O (0.0018 vol%) in cyclopentyl methyl ether (CPME) (0.002 M) was found to be most suitable for the asymmetric DcA reaction of allylic α-fluoro-β-keto ester 5 to furnish the target (−)-6 in 83% with 48% ee. The antipode, (+)-6 was also accessed using (S,S_p)-Ph-Taniaphos (L7) in 71% with 45% ee. The hydrogenolysis of the end-alkane moiety in (−)-6 with Pt–C in THF gave fluorinated donepezil (−)-1a in 64% yield with 50% ee (Scheme 1).

Scheme 1 Novel method for fluorinated donepezil 1a and its asymmetric synthesis by palladium-catalyzed DcA reaction.

A key for the DcA reaction of α-fluorinated β-keto ester 5 to α-fluorinated ketone 6 is screening out suitable catalysts/ligands for previously unknown DcA reactions of allylic ester having a tri-substituted heterocyclic alkene moiety. A large number of palladium-catalyzed DcA reactions have been reported, but reactions using tri-substituted cyclic alkene systems are surprisingly rare and have remained a challenge.^14,15 Our investigation of 5 to 6 started with the following reaction conditions: 5 mol% tris(dibenzylideneacetone)dipalladium [Pd₂(dba)₃] and 1,2-bis(diphenylphosphino)ethane (dppe) in CPME. However, the reaction did not proceed (Table 1, entry 1). Further optimization of the reaction conditions using a variety of ligands, including mono- and bidentate phosphines, also failed (entries 2–4). A change in the ligands to 1,4-bis(diphenylphosphino)butane or 1,5-bis(diphenylphosphino)pentane having a longer alkyl chain delightfully gave product 6, but low yields of 25–35% (entries 5, 6). We found that other bidentate ligands such as dpephos or xantphos were also effective in this transformation (entries 7 and 8). The use of dppf in particular gave a product with a high yield of 77% and a shorter reaction time of 5 h, although 1,1′-bis(dicyclohexylphosphino)ferrocene (dcpf), 1,1′-bis(diisopropylphosphino)ferrocene (dipf) and 1,1′-bis(di-tert-butylphosphino)ferrocene (dtpf) gave lower yield or even no reaction (entries 9–12). The catalytic conditions are also effective for the DcA reaction of non-fluorinated substrate 4 instead of 5 (entries 13 and 14). While common conditions using dppe did not give the product (entry 13), the use of dppf ensured the transformation of 4 to the corresponding non-fluorinated analogue 7 in 76% yield (entry 14), representing a new synthetic route for donepezil (Scheme ESI-1†).

Table 1 Optimization of reaction conditions from 5 to 6 ^a

Entry	Ligand	Time (h)	Yield^b (%)
a Reaction condition: 5 (0.05 mmol), Pd2dba3 (5 mol%), ligand (12.5 mol%) in CPME (2.5 mL) at 40 °C, unless otherwise noted.b Yields were determined by ^19F NMR.c 25 mol% of ligand was used.d Isolated yield.e The reaction was carried out using non-fluorinated substrate 4 instead of 5.f Non-fluorinated compound of 6, H instead of F, was obtained.
1	dppe	42	N.R.
2^c	nBu3P	24	N.R.
3^c	CyJohnphos	40	N.R.
4	dppp	43	N.R.
5	dppb	43	25
6	dpppent	48	35
7	dpephos	40	50
8	xantphos	24	35
9	dppf	5	77^d
10	dcpf	43	13
11	dipf	43	N.R.
12	dtpf	43	N.R.
13^e	dppe	43	N.R.
14^e	dppf	5	76^d^,^f

---

Encouraged by this transformation, we next investigated the asymmetric DcA reaction of 5 in the presence of a variety of chiral bidentate phosphine ligands (Table 2). The reaction did not proceed in the presence of (R)-BINAP or (R)-SEGPHOS as a chiral ligand (entries 1 and 2). (S)-tBu-PHOX led to 35% yield of 6 with only 4% ee (entry 3). A well-explored (S,S)-DACH-phenyl-Trost ligand (L1) was also attempted in the reaction, but the production of 6 was not observed (entry 4). Based on our initial study for a racemic system, we next considered several ferrocene ligands. Contrary to our expectation, L2 with an oxazoline moiety did not give the DcA product (entry 5). After screening three ligands (entries 6–8), we found that L5 was suitable for this transformation (86% yield, 42% ee, entry 8), whereas L6, having a cyclohexyl moiety, provided 6 with poor results (12%, 20% ee, entry 9). The best result of 48% ee (83% yield) was obtained when the reaction was carried out in the presence of L5 under highly diluted conditions (0.002 M of CPME, entry 10), although it required a longer reaction time (entries 8 vs. 10). During optimization, we noticed that moisture was important for higher enantioselectivity. When the reaction was examined under the best conditions (entry 8), but in freshly dried CPME, enantioselectivity decreased to 32% (entry 11), and it was recovered to 43% ee in the presence of 5 equivalents of water (entry 12). The amount of water also strongly affected the ee values in the asymmetric DcA reaction (entries 12–15). Other additives also affected the enantioselectivity (entries 16 and 17). The best amount of water was found to be 0.0018 vol% (entry 12). Screening solvents revealed that CPME was the most appropriate solvent for this reaction (entries 18–21). The reaction temperature was finely optimized (entries 22–25) and the best result was obtained under these conditions at 30 °C (85% yield, 46% ee, entry 26). The 6 obtained in these conditions is the (−)-isomer of 6. When the reaction was carried out under the best conditions but using the ligand, L7 (the antipode of L5), the (+)-isomer 6 was obtained in 71% yield with 45% ee (entries 27 and 28). The ees achieved here seem to be modest, and to our knowledge, this is the first example of an asymmetric palladium-catalyzed decarboxylative allylation of α-fluoro-β-keto ester with tri-substituted heterocyclic alkene. Although the role of water to improve enantioselectivity is not clear, it could be explained as follows: 5 is converted into a cationic allylpalladium complex TS-II armed with chiral ligands via decarboxylation of TS-I. Since water did not decrease the yield of 6 (i.e., no formation of protonated by-product¹⁶), the ion pair complexes TS-I and TS-II are tight and in close contact with the Pd(II) metal center. Thus water presumably coordinates ketone and/or the benzylamine portion of TS-I via hydrogen bonding as outlined in Scheme 2. This coordination should affect the stabilization of the desirable conformation resulting in improved enantioselectivity (Scheme 2).

Table 2 Screening conditions for asymmetric DcA reaction from 5 to 6 ^a

Entry	Ligand	Additive	Time (h)	Yield^b (%)	ee (%)
a Reaction conditions: 5 (0.05 mmol), Pd2dba3 (5 mol%), ligand (12.5 mol%), additive (5.0 equiv.) in CPME (2.5 mL) at 28 °C, unless otherwise noted.b Isolated yield.c 0.0018 vol%.d 0.00036 vol%.e 0.00072 vol%.f 25 mL of CPME was used.
1	(R)-BINAP	—	72	N.R.	—
2	(R)-SEGPHOS	—	72	N.R.	—
3	(S)-tBu-PHOX	—	85	35	(−)-4
4	L1	—	65	N.R.	—
5	L2	—	110	N.R.	—
6	L3	—	36	77	(+)-28
7	L4	—	12	89	(+)-19
8	L5	—	34	86	(−)-42
9	L6	—	12	12	(−)-20
10^f	L5	—	48	83	(−)-48
11	L5	—	37	87	(−)-32
12	L5	Water^c	36	81	(−)-43
13	L5	Water^d	36	78	(−)-41
14	L5	Water^e	36	77	(−)-42
15	L5	MS4 Å	34	79	(−)-34
16	L5	MeOH	34	80	(−)-43
17	L5	HFIP	34	66	(−)-23
18	L5 (toluene)	—	74	77	(−)-35
19	L5 (THF)	—	24	79	(−)-21
20	L5 (dioxane)	—	39	54	(−)-39
21	L5 (CH2Cl2)	—	63	76	(+)-6
22	L5 (25 °C)	Water^c	72	78	(−)-45
23	L5 (30 °C)	Water^c	36	85	(−)-44
24	L5 (35 °C)	Water^c	24	88	(−)-42
25	L5 (40 °C)	Water^c	6	89	(−)-41
26^f	L5 (30 °C)	Water^c	36	85	(−)-46
27	L7	—	37	87	(+)-42
28^f	L7 (30 °C)	Water^c	40	71	(+)-45

---

Scheme 2 A plausible transition state model with the effect of water.

Completion of the synthesis of fluorinated donepezil 1a was finally investigated, namely, the hydrogenolysis of (−)-6 with 46% ee in the presence of Pt–C in THF gave target (−)-1a in 64% yield with 50% ee (Scheme 1).

The 4% increasing in ee of 1a from 6 could indicate the potential of SDE¹⁷ via achiral chromatography. We thus decided to examine the SDE of 1a via gravity-driven achiral column chromatography (Fig. 2). 1a, whose ee was enriched to 44% in (−)-enantiomer, was loaded onto the column as a solution in hexane/EtOAc = 2/8. The fractions were collected (3.0 mL each), and the ee of 1a eluted in the fractions was determined by HPLC using CHIRALPAK IA. The result is shown in Fig. 2. Excitingly, 1a showed a noticeable magnitude of the SDE phenomenon, and the first eluted fractions were enantiomerically enriched with 72% ee while later eluted fractions were gradually enantiomerically depleted, reaching 29% ee. The magnitude of SDE in 1a is particularly unique since 1a does not have any hydrogen donor in the structure which should be indispensable to form homochiral/heterochiral dynamic molecular associations.^9c A plausible mechanism of SDE for 1a is shown in Fig. 3. The α-fluorine of 1a would effectively stabilize the hydrophilic hydrate of 1a over the silica-gel. The 1a-hydrate should exist as a mixture of hydrate (SS)/(RR)-homodimer and hydrate (SR)-heterodimer.^9c SDE should be induced by the difference in stability and polarity of the monomer (S)/(R)-1a-hydrate, the 1a-hydrate homodimer and the 1a-hydrate heterodimer.

Fig. 2 Chromatographic plots as enantiomeric excess (%) vs. elution fraction obtained for 1a in hexane/EtOAc = 2/8 as the eluent.

Fig. 3 A plausible mechanism of SDE for 1a on the silica-gel.

In conclusion, a novel method for the synthesis of fluorinated donepezil is disclosed by the palladium-catalyzed decarboxylative allylation (DcA) of the heterocyclic allylic ester of α-fluoro-1-indanone-2-carboxylate as a key step. The ferrocene ligand dppf is suitable for this DcA reaction in high yield, and the method is applicable for the asymmetric DcA reaction using a catalyst of Pd₂dba₃ and (R,R_p)-Ph-Taniaphos, providing (−)-6. The (+)-isomer of 6 was also obtained using (S,S_p)-Ph-Taniaphos. Water plays an important role in this asymmetric DcA reaction and the precursor of fluorinated donepezil was obtained with up to 46% ee. Finally, fluorinated donepezil (−)-1 was obtained in 50% ee after hydrogenolysis. This is the first example of the catalytic asymmetric generation of fluorinated donepezil. Since the (+)-enantiomer of fluorinated donepezil is 60 times more potent than its (−)-enantiomer, our method will activate further clinical research of fluorinated donepezil as a promising new therapeutic agent for Alzheimer's disease in the next generation, while the clinical use of enantiomerically pure donepezil should become meaningless due to its racemization in the human body. The observed noticeable magnitude of the SDE phenomenon of fluorinated donepezil suggests different mechanisms of SDE, and a further study is under way to compare with donepezil.

Acknowledgements

These studies were financially supported in part by the Platform for Drug Discovery, Informatics, and Structural Life Science from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) Japan, the Japan Agency for Medical Research and Development (AMED), JSPS KAKENHI Grant Number JP 16H01142 in Middle Molecular Strategy, and the Advanced Catalytic Transformation (ACT-C) from the Japan Science and Technology (JST) Agency.

Notes and references

1. (a) H. Sugimoto, Y. Iimura, Y. Yamanishi and K. Yamatsu, J. Med. Chem., 1995, 38, 4821; (b) H. Sugimoto, H. Ogura, Y. Arai, Y. Iimura and Y. Yamanishi, Jpn. J. Pharmacol., 2002, 89, 7.
2. (a) N. R. Dunn, G. L. Pearce and S. A. W. Shakir, J. Psychopharmacol., 2000, 14, 406; (b) V. P. Prasher, Int. J. Geriatr. Psychiatr., 2004, 19, 509; (c) A. Tanaka, S. Koga and Y. Hiramatsu, Intern. Med., 2009, 48, 1219.
3. (a) Y. Iimura, T. Kajima, N. Araki and H. Sugimoto, JP04021670, 1992; (b) Y. Yamamoto, Y. Ishihara and I. D. Kuntz, J. Med. Chem., 1994, 37, 3141; (c) M. G. Cardozo, Y. Iimura, H. Sugimoto, Y. Yamanishi and A. J. Hopfinger, J. Med. Chem., 1992, 35, 584.
4. (a) G. Kryger, I. Silman and J. L. Sussman, J. Physiol., 1998, 92, 191; (b) G. Kryger, I. Silman and J. L. Sussman, Structure, 1999, 7, 297.
5. W. Lili, G. Cheng, Z. Zhiyong, Y. Qi, L. Yan, L. Dan, Z. Xueli and Z. Yuan, Chirality, 2013, 25, 498.
6. Z. Fu, X. Li, Y. Miao and K. M. Merz Jr, J. Chem. Theory Comput., 2013, 9, 1686.
7. K. Matsui, Y. Oda, H. Ohe, S. Tanaka and N. Asakawa, J. Chromatogr. A, 1995, 694, 209.
8. (a) H. Sugimoto, J. Synth. Org. Chem., Jpn., 1998, 56, 320; (b) R. Gotti, C. Bertucci, V. Andrisano, R. Pomponio and V. Cavrini, Anal. Bioanal. Chem., 2003, 377, 875.
9. (a) Y. Takeuchi, T. Shiragami, K. Kimura, E. Suzuki and N. Shibata, Org. Lett., 1999, 1, 1571; (b) T. Yamamoto, Y. Suzuki, E. Ito, E. Tokunaga and N. Shibata, Org. Lett., 2011, 13, 470; (c) M. Maeno, E. Tokunaga, T. Yamamoto, T. Suzuki, Y. Ogino, E. Ito, M. Shiro, T. Asahi and N. Shibata, Chem. Sci., 2015, 6, 1043.
10. Y. Takeuchi, N. Shibata, E. Suzuki, Y. Iimura, T. Ozasa, Y. Yamanishi and H. Sugimoto, JP2000319257A, 2000.
11. It is very possible for fluorinated compounds to have improved biological activity, see: (a) S. Purser, P. R. Moore, S. Swallow and V. Gouverneur, Chem. Soc. Rev., 2008, 37, 320; (b) E. P. Gillis, K. J. Eastman, M. D. Hill, D. J. Donnelly and N. A. Meanwell, J. Med. Chem., 2015, 58, 8315; (c) Y. Zhou, J. Wang, Z. Gu, S. Wang, W. Zhu, J. L. Aceña, V. A. Soloshonok, K. Izawa and H. Liu, Chem. Rev., 2016, 116, 422.
12. Y. Iimura, T. Ozasa, Y. Yamanishi, H. Sugimoto, Y. Takeuchi, N. Shibata and E. Suzuki, JP2000319258A, 2000.
13. Y. Iimura, T. Ozasa and N. Suzuki, JP2002105052A, 2002.
14. For selected example on DcA reaction using tri-substituted alkene, see: (a) T. Tsuda, Y. Chujo, S. Nishi, K. Tawara and T. Saegusa, J. Am. Chem. Soc., 1980, 102, 6381; (b) J. Tsuji, T. Yamada, I. Minami, M. Yuhara, M. Nisar and I. Shimizu, J. Org. Chem., 1987, 52, 2988; (c) I. Shimizu, H. Ishii and A. Tasaka, Chem. Lett., 1989, 18, 1127; (d) K. Chattopadhyay, E. Fenster, A. J. Grenning and J. A. Tunge, Beilstein J. Org. Chem., 2012, 8, 1200.
15. For selected example on DcA reaction using a fluorine-containing substituent, see: (a) I. Shimizu and H. Ishii, Tetrahedron, 1994, 50, 487; (b) J. T. Mohr, D. C. Behenna, A. M. Harned and B. M. Stoltz, Angew. Chem., 2005, 117, 7084 (Angew. Chem., Int. Ed., 2005, 44, 6924); (c) M. Nakamura, A. Hajra, K. Endo and E. Nakamura, Angew. Chem., 2005, 117, 7414 (Angew. Chem., Int. Ed., 2005, 44, 7248); (d) É. Bélanger, K. Cantin, O. Messe, M. Tremblay and J.-F. Paquin, J. Am. Chem. Soc., 2007, 129, 1034; (e) É. Bélanger, C. Houzé, N. Guimond, K. Cantin and J.-F. Paquin, Chem. Commun., 2008, 3251; (f) M.-H. Yang, D. L. Orsi and R. A. Altman, Angew. Chem., 2015, 127, 2391 (Angew. Chem., Int. Ed., 2015, 54, 2361).
16. J. D. Weaver, A. Recio III, A. J. Grenning and J. A. Tunge, Chem. Rev., 2011, 111, 1846.
17. V. A. Soloshonok, C. Roussel, O. Kitagawa and A. E. Sorochinsky, Chem. Soc. Rev., 2012, 41, 4180.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra21253k

---