The Synthetic-Technical Development of Oseltamivir Phosphate Tamiflu<Sup>

HOT TOPICS: SANDMEIER PRIZE 2006 93 doi:10.2533/chimia.2007.93 CHIMIA 2007, 61, No. 3

Chimia 61 (2007) 93–99 © Schweizerische Chemische Gesellschaft ISSN 0009–4293

The Synthetic-Technical Development of Oseltamivir Phosphate TamifluTM: A Race against Time

Stefan Abrechta , Muriel Cordon Federspielb , Heinrich Estermannb , Rolf Fischerb , Martin Karpf*a , Hans-Jürgen Mairb , Thomas Oberhauserc , Gösta Rimmlerb , René Trussardia , and Ulrich Zuttera

Recipients of the Sandmeyer Prize 2006 of the Swiss Chemical Society

Abstract: The clinical development of the first orally available neuraminidase inhibitor prodrug oseltamivir phosphate (TamifluTM) proceeded very fast. In order to support this program an unprecedented team effort in chemical process research, development, piloting, production and analytics took place, which allowed the successful launch of TamifluTM in 1999, only two and a half years after it was licensed from Gilead Sciences. This article describes selected aspects of the commercially used synthesis route and a brief summary of alternative syntheses devised by Roche chemists. Keywords: Analytics · Influenza neuraminidase inhibitor · Oseltamivir phosphate · Production · Synthesis · Technical development

1. Introduction H O H O CO H In September 1996 a license agreement be- H O 2 O CO Et tween Gilead Sciences, Foster City, Califor- 2 H O nia and F. Hoffmann-La Roche Ltd, Basel AcHN H N NH was signed for the co-development of the AcHN 2 NH2 •H3 PO4 novel orally available neuraminidase inhibi- NH tor prodrug molecule 1 (Fig. 1) patented by Gilead Sciences in February 1995,[1] today 1 2 known as oseltamivir phosphate, trade name TamifluTM. In April 1999, after only two and Oseltamivir phosphate Zanamivir a half years of development time, a new drug TamifluTM RelenzaTM GS-4104-02 GG-167 application (NDA) for 1 was filed with the RO0640796-002 US Food and Drug Administration (FDA) for the use of 1 for the treatment of influenza Fig. 1. Marketed neuraminidase inhibitors

virus infections. This ambitious development and technical development activities lead- program was in part triggered by the parallel ing to commercial route selection followed development of zanamivir ( 2), trade name Re- by piloting and including all related analyti- lenzaTM by GlaxoSmithKline. This neuramin- cal activities had to be challenged in order idase inhibitor emerging from the research ef- to finally allow for a successful and timely forts of Prof. von Itzstein, Monash University, launch production of several metric tonnes. Australia[2] was already patented in 1990 and An unprecedented and overlapping team is equipotent in vitro to 1 but shows low oral effort regarding decisions on critical mile- *Correspondence: Dr. M. Karpfa bioavailability and was filed with the NDA for stones was required throughout the func- Tel.: + 41 61 688 5299 Fax.: + 41 61 688 1670 the topical treatment of influenza using disc tions involved. E-Mail: [email protected] inhaler technology in October 1998. F. Hoffmann-La Roche Ltd Due to the exceptionally fast clinical Pharmaceuticals Division a Synthesis & Process Research development program for 1 , the standard 2. A Swift Start b Chemical Process Development chemical-technical development plans in- c Technical Project Management cluding trouble-shooting of the discovery In September 1996, chemists from Grenzacherstrasse 124 synthesis, synthesis & process research the Synthesis & Process Research depart- CH-4070 Basel HOT TOPICS: SANDMEIER PRIZE 2006 94 CHIMIA 2007, 61, No. 3 ment of F. Hoffmann-La Roche Ltd were 1) SO2 Cl2 first confronted with the synthetic pathway 2) pyrrolidine H O shown in Schemes 1 and 2 (taken from H O Pd(PPh3 ) 4 cat. CO Et H O CO H O CO Et O 2 2 2 3) H SO extr. refs[3,4]) created and used by Rohloff et 2 4 O H O O al., Gilead Sciences for the scale-up of the 43% transformation of quinic acid ( 3 ) to oselta- OH ~70% OMs cryst. OMs mivir phosphate ( 1 ), published in 1998.[3] 3 4 5 Due to the limited information available (-)-Quinic acid yield and safety ? O on the majority of issues circled in the two selectivity ? HClO cat. Schemes 1 and 2 ( e.g. the availability of world supply ? 4 the starting material 2 on a large scale, the feasibility, safety and scalability of several BH . Me S NaHCO3 /H2 O 3 2 synthetic steps) the immediately required CO Et CO Et O 2 EtOH CO Et O 2 assessment was largely based on the confi- O 2 TMSOTf dence in the skills of the chemists who were 67% O O availability H O oil OMs to be involved. OMs price ? 8 yield from (-)-quinic acid ~20% 7 6 Epoxide RO0640792 3. The Access to the Key Epoxide 8: Where to Start from? Scheme 1. Synthesis of epoxide 8 according to Rohloff et al.[3]

Initially quinic acid ( 3 ) was used as the raw material which is isolated from the availability bark of the cinchona tree as a side prod- stench ? uct during the extraction of cinchona alka- NaN3 ,NH 4 Cl PMe 3 CO Et loids. The cinchona tree is indigenous pre- O CO Et EtOH ,65°C O 2 MeCN ,RT O CO Et dominantly in Central Africa, Zaire. The 2 2 worldwide yearly capacity for quinic acid H O O N H N was estimated to be approx. 80 tonnes. 3 Potential suppliers at the time offered to 819 0 Epoxide provide us with max. 15 tonnes per year RO0640792 safety of translating into approx. 5 tonnes of drug NaN ,NH Cl substance 1 , which was clearly below the selectivity ? azide reagents 3 4 &intermediates ? DMF, 85 °C quantities forecasted. Therefore for large- scale production an alternative raw mate- 1) Lindlar /H 2 Ac2 O O CO Et 2) H 3 PO pyridine rial had to be found. 2 4 O CO2 Et O CO2 Et

AcHN 80% AcHN 35% H N 3.1. Problems when Starting from 2 NH •H3 PO4 N N Quinic Acid 3 2 overall yield from 3 3 The key problematic step in the trans- 1 (-)-quinic acid ~ 6% 12 11 Oseltamivir formation of quinic acid ( 3 ) to the ethyl- phosphate mesylshikimate ( 5 )was the dehydration of [3] hydroxymesylate 4 (Scheme 3). The origi- Scheme 2. Synthesis of oseltamivir phosphate ( 1 )according to Rohloff et al. nal method using SO2 Cl2 /pyridine was not particularly regioselective resulting in a Cl H O 3–4:1 mixture of the regioisomers 5 and 13 O CO Et CO Et O CO Et CO2 Et O 2 O 2 2 together with smaller amounts of the chloro + + derivative 14. In order to separate the de- O O O O sired crystalline isomer 5 without chroma- OMs OMs OMs OMs tography the mixture had to be subjected to 4 5 13 14 a Pd(0)-catalyzed substitution with pyrrolidine, whereby only the allylic mesylate in Gilead [3] 1) SO Cl ,Pyridine, the undesired isomer 13 reacted, followed 2 2 CH2 Cl -25 °C by extractive removal of the corresponding 2) Pyrrolidine pyrrolidine derivative. A much higher regi- cat. Pd(PPh3 ) 4 3) H SO extr. oselectivity of 35–50:1 was achieved using 2 4 3-4 1 1 [5] Regioselectivity: 3-4/1 Vilsmeier’s salt. The desired product 5 (Yield: 43%) was isolated in 60% yield after crystallization from MeOH. Roche [5] Vilsmeier's salt + Cl- N Cl 3.2. Starting from Shikimic Acid (15) (in situ preparation or purchased) Shikimic acid ( 15), the ideal raw ma- DMF or ethyl acetate, 60-80 °C terial for the synthesis of 1 , was com- Regioselectivity: 35-50/1 3.5-5 0.1 1 mercially only available in small research (Yield: 58-65%) quantities by mid 1997. However, the newly Scheme 3. Regioselective elimination: from quinic acid derivative 4 to the corresponding shikimic generated high demand for this compound acid derivative 5 HOT TOPICS: SANDMEIER PRIZE 2006 95 CHIMIA 2007, 61, No. 3 quickly changed the situation. The two 1) Me C(OMe) (2.0 eq.) sources were and still are production by 1) EtOH, SOCl2 2 2 H O COOH (0.5 eq.) H O COOEt TsOH (0.01 eq.) COOEt fermentation using a recombinant E. coli [6] EtOAc, max. 35 O reflux, 3h °C strain and extraction of 15 from star anise. 2) evaporation O H O 2) evaporation H O Therefore the supply of sufficient quantities OH OH 95% OH 97% of shikimic acid ( 15) is secured even in the 15 16 18 face of the dramatically increased demand 1) MsCl (1.3 eq.) 1) 3-pentanone (25 eq.) 2) Et3 N(2.0 eq.) for Tamiflu ( 1 ) caused by the threat of an CF SO H(0.05 eq.) EtOAc, 0-5 °C 97% 3 3 89% 2) extraction 3) filtration influenza pandemic. 3) evaporation 4) evaporation The conversion of shikimic acid ( 15)to 5) cryst.MeOH the ketal 6 is outlined in Scheme 4. For the O COOEt O COOEt efficient production of 6 the longer route via O O the crystalline acetonide 5 is preferred since OH OMs in the one step shorter route via the ketal 17 17 5 all intermediates are oils, which are difficult 1) MsCl (1.3 eq.) 1) 3-pentanone (15 eq.) 2) Et N(2.0 eq.) 93% 98% to purify efficiently on large scale. 3 CF3 SO3 H(0.045 eq.) CH2 Cl2 ,0-5 °C 2) extraction Clearly shikimic acid ( 15)is the starting 3) extraction 3) evaporation material of choice giving access to the eth- O COOE t ylmesylshikimate intermediate 6 in about 80% overall yield compared to about 40% O overall yield starting from quinic acid 3 . OMs 6 The original reductive opening[3] of the ketal 6 with BH3 -SMe2 /TMSOTf proceed- Scheme 4. Synthesis of ketal 6 from shikimic acid ( 15) ed with only moderate regioselectivity and the reagents were considered unsuitable for large-scale production. A much higher se- H O CO2 Et NaHCO CO Et CO Et 3 lectivity was observed using Et SiH/TiCl . O 2 H O 2 H O /EtOH CO2 Et 2 3 4 O O CO2 Et O At –34 °C a 32:1 mixture of the desired O + + OMs H O hydroxyether 7 and its regioisomer 19 was OMs H O OMs OMs O obtained along with 2–4% diol 20. Epoxide The crude hydroxyether 7 was con- 617 9820 verted to the epoxide 8 with NaHCO in 3 Gilead [3] EtOH/H O at 60 °C. Epoxide 8 crystal- 1) BH -Me S 2 3 2 lized from hexane as a white, crystalline 2) TMSOTf CH2 Cl2 , -20°C solid in 80% yield and typically 99% assay Regioselectivity 10 :1 Yield: 63% (Scheme 5). Roche [5]

1) Et3 SiH

2) TiCl4 CH Cl ,-35 °C 4. The Azide Based Transformation 2 2 of the Epoxide 8 to Oseltamivir Regioselectivity 32 :1 Yield: 80% Quality >98% Phosphate (1)

Based on the synthesis of Rohloff et Scheme 5. Regioselective reduction of the ketal 6 al.[3] development work at Roche was performed in close collaboration with several contract manufacturers experienced in azide chemistry. In-depth investigations of CO Et NaN ,NH Cl CO Et CO Et CO Et process safety were required for the whole O 2 3 4 O 2 O 2 O 2 + + sequence. EtOH/H2 O H O N 3 O 60-65 °C For the isolated azidoalcohols 9 and 21 N OH safety tests recommended 40 °C as the high- 3 est permissible safe process temperature 829 122 whereas in solution a safety limit of 70 °C was recommended. The opening of the ep- Scheme 6. Optimized conditions for the epoxide ring opening oxide 8 proceeded highly stereospecifically with 9:1 regioselectivity. Both regioisomers 9 and 21 formed the desired aziridine 10 in available in large quantities, exceptionally The direct conversion of the intermedi- the subsequent step (Scheme 8). An excess malodorous, pyrophoric and highly toxic. ates to the aziridine 10 requires high reac- of sodium azide was required in order to Mechanistically the first step of the trans- tion temperatures and long reaction times. push the reaction to completion and to keep formation is interpreted as a Staudinger type The progress of the reaction was followed the formation of the undesired by-product reaction which forms iminophosphorane by 31P-NMR spectroscopy: the signals for 22 below 3%. and oxazaphospholidine intermediates at the iminophosphorane and oxazaphospho- For the formation of the aziridine 10 room temperature. These intermediates are lidine showed up at 10.99 ppm (mostly trimethylphosphine was replaced by tri- sensitive to traces of water. In order to avoid broad) and –55.28 ppm, respectively, and phenylphosphine, which is cheaper and formation of undesired aminoalcohol 23, disappeared in the course of the reaction. much easier to handle on industrial scale solvents and reagents have to be carefully The triphenylphosphine oxide signal oc- production. Trimethylphosphine is hardly checked for absence of water before use. curred at 28.49 ppm and grew in intensity HOT TOPICS: SANDMEIER PRIZE 2006 96 CHIMIA 2007, 61, No. 3 and the signal for excess triphenylphosphine at –4.95 ppm remained constant. The O COOEt O COOEt + PR H O COOEt mechanism of the aziridine formation is 3 HO HO outlined in Scheme 7. -N2 HO N + N 3 Imino- NH Since the aziridine 10 was found to be R 3 P phosphorane R 3 P thermally unstable acceleration by the ad- 9 NPR dition of catalytic amounts of HOAc as 3 H O COOEt reported[7] was considered. Rohloff et al.[3] R=MeorPh COOEt R' =3-pentyl R'O NPR demonstrated that 5–10 mol% NEt3 -HCl R'O 3 was suitable but the disadvantage of this ap- H O proach was the opening of the aziridine ring reactive conformation by the chloride ion as the nucleophile gen- erating the by-product 24 (Fig. 2). Depend- O COOEt O COOEt O COOEt ing on the reaction conditions, the amount -H+ O O + of the dimer by-product 25 increased. A N + NH H -R3 PO R P NH R P 2 breakthrough was achieved by using trieth- 3 2 3 10 Oxazaphospholidine ylamine and methanesulfonic acid. Acid- NH2 catalysis is required for the conversion of + O COOEt R 3 P azidoalcohol 9 and 21 to aziridine 10 in COOEt R'O NH R'O 2 order to shorten the reaction time and to O R P + keep the reaction temperature below 60 °C. 3 Further reaction to the dimer by-product 25 reactive conformation is minimized when using a slight excess of Scheme 7. Mechanistic considerations of the formation of aziridine 10 from azidoalcohol 9 triethylamine. Originally the aziridine opening to aminoazide 11 was carried out using sodium azide and ammonium chloride in dimeth- ylformamide at 80–85 °C. This procedure turned out to be not technically feasible. Yields of ca. 50% were obtained on lab O COOEt O COOEt scale. Upon scale-up to 0.5–5 kg yields H 2 N AcHN dropped to 25–30%. Calorimetric measure- N N O CO Et CO Et ments showed that aminoazide 11 is unsta- 2 O 2 O O ble at temperatures above 50 °C. With the H O H 2 N finding that aziridine 10 opened smoothly NH2 Cl EtOOC EtOOC with sodium azide in dimethylsulfoxide in the presence of phosphoric acid a much 23 24 25 26 safer process for aziridine opening was available since deposition of ammonium Fig. 2. By-products of the aziridine formation azide sublimating into the cooling system of the equipment was avoided (Scheme 8). In order to prevent unacceptably high levels of HN3 in the headspace, the reaction was carried out below 37 °C. Ph P The next step was the development of a 3 Et3 N-MsOH CO Et CO Et one-pot reaction by carrying out both aziri- O 2 O 2 DMSO, 50 °C, 1h O CO2 Et dine formation and opening to the amino- + N azide 11 in dimethylsulfoxide as solvent. H O 3 H N For the one-pot procedure the reaction N 3 OH mixture obtained after aziridine formation 9 21 10 was treated with sodium azide and heated to 33–37 °C directly followed by aziridine ring opening performed by feeding a solu- 1) NaN3 ,H2 SO4 DMSO, 35 °C, 4h tion of sulfuric acid in dimethylsulfoxide at a rate keeping the temperature below 37 2) neutralization (NH or NaOH °C. An excess of sodium azide is required 3 to pH =8), 0-5 °C to suppress the dimerization but is limited 3) prec. Ph PO due to the solubility of sodium azide in di- 3 methylsulfoxide. In contrast to the original procedure, Ac O O CO2 Et solvent exchange between aziridine forma- O CO2 Et 2 tion and ring opening was avoided. With the H N replacement of ammonium chloride by sul- AcHN 2 N 3 furic acid the temperature of the ring open- N 3 ing reaction was lowered and the time of 12 11 conversion and cycle times were consider- ably reduced as well as the number of unit Scheme 8. Optimized synthesis of acetamidoazide 12 from the azidoalcohols 9 and 21 HOT TOPICS: SANDMEIER PRIZE 2006 97 CHIMIA 2007, 61, No. 3 operations. Besides the impact on process removal of water by azeotropic distilla- 5.2. Azide-Free Transformations safety, higher yields of aminoazide 11 were tion the reaction mixture was used without Although the safe use of azide chemistry reached due to less thermal burden on aziri- further processing in the phosphate salt on an industrial scale is well established, the dine 10. formation. A solution of ortho-phosphoric potential hazards related to its application Reaction calorimetric measurements acid in ethanol was added at 50 °C followed as well as the dependence on third parties for the ring opening showed that the heat re- by seed crystals to induce crystallization of prompted us to search for an efficient azide- leased for the addition of sodium azide can oseltamivir phosphate 1 . After cooling to free transformation of the epoxide 8 to 1 . [7] be neglected. Aziridine ring opening and room temperature 1 was isolated, washed To this end, a suitable non-azide nitrogen formation of aminoazide 11 proceeded with and dried. By this process oseltamivir phos- nucleophile and appropriate conditions had the addition of sulfuric acid solution and phate ( 1) was obtained in a yield of 88–92% to be found compatible with the functional was controlled by the feed rate. Formation related to acetamidoazide 12 and in a very groups present and the pronounced tenden- of acetamidoazide 12 was then performed high quality ( ≥ 99% assay). The crystalliza- cy of such highly functionalized cyclohex- by dosing acetic anhydride to the solution tion of 1 from ethanol proved to be very ene derivatives towards aromatization. of aminoazide 11 in dibutylether between efficient in removing both the N-acetylated These efforts finally led to the selection 0–25 °C. According to reaction calorimetry aziridine dimer 26, the main impurity re- of allylamine as the reagent of choice not heat evolution was fully controlled by dos- maining of the azide steps and tri-n-bu- only for the MgBr2 •OEt2 -catalyzed epoxide age. The only impurity detected in the iso- tylphosphine oxide, the by-product of the ring opening reaction of 8 to 29 and the sub- lated material 12 at a level above 1% was Staudinger reaction. sequent Pd/C-catalyzed deallylation to 30, the acetylated dimer 26. Lab experiments but also for the short, selective and effec- revealed that up to 3% of 26 in the acetami- tive introduction of the second amino func- doazide 12 were acceptable for the manu- 5. The Search for Alternative tion present in 31 followed by the selective facture of oseltamivir phosphate ( 1). Approaches N-acetylation under acidic conditions fol- Thefinalreductionofacetamidoazide12 lowed by deallylation of 32 and phosphate to give oseltamivir ( 1) can be accomplished 5.1. Questions for Synthesis & Pro- salt formation (Scheme 10). This new and by using various methods and reagents. cess Research efficient reaction sequence finally led to the During lab development the originally pre- As described above, the investigations drug substance 1 in good overall yield. ferred heterogeneous catalytic hydrogena- performed finally led to the currently used The reaction cascade developed for the tion proved to be problematic with respect technical process for the production of osel- direct conversion of 30 to 31 with concomi- to the functionalities in the molecule. Lind- tamivir phosphate ( 1) starting from shikimic tant introduction of the second amino func- lar catalyst, Raney Nickel, Pd/C etc. led to acid ( 15). However, for quite some time dur- tion shown in Scheme 11 represents an effi- an unfavorable impurity profile, e.g . due to ing the early development stage two most cient sequence allowing for the combination double bond migration to the ‘1,6’-position important questions remained unanswered: of six chemical steps without the isolation ( 27) or reduction of the double bond ( 28) a) would it be possible to finally acquire of intermediates. This so-called ‘domino (Scheme 9). Once formed both impurities tonnes of shikimic acid to start large-scale sequence’ exemplifies a potentially very 27 and 28 are difficult to remove by the fi- production of theAPI and b) would it be pos- economical way of working and constitutes nal crystallization of the phosphate salt 1 . sible to find strong partners to perform the a new, straightforward and practical conver- To avoid all these problems a Stauding- azide steps on large scale in a safe and ef- sion of an aminoalcohol into a vicinal di- er type phosphine reduction was developed ficient way. Therefore, Synthesis & Process amine employing readily available and safe using commercially available tri-n-bu- Research embarked early on the investiga- reagents. As an interesting variation of the tylphosphine.[8] By adding a slight excess tion on shikimic acid independent approach- ‘allylamine’ route of the epoxide 8 to 1 a of tri-n-butylphosphine to an aqueous etha- es as well as on the exploration of azide-free closely related process has been established nolic solution of 12 at 5 °C a very selective transformations of the key epoxide 8 . and developed.[9] and smooth conversion to the free base of Since the results of these investigations 1 was obtained. Subsequently the reaction are well documented in an earlier CHIMIA 5.3. Shikimic Acid Independent mixture was warmed to room temperature publication[4] we summarize here the final Approaches in order to complete the reduction. After results for the sake of completeness. 5.3.1. The Need for Shikimic Acid Independent Approaches Due to the pace of development of oseltamivir phosphate ( 1 ) and especially due

Bu3 P, cat. AcOH to the initial uncertainty regarding the com- H PO ,EtOH EtOH/H2 O, 3 4 O mercial tonne-scale availability of the start- O CO Et 5-20°C CO2 Et 50-20 °C O CO Et 2 2 ing material shikimic acid ( 15), the inten- AcHN AcHN AcHN sive search for shikimic acid independent N NH . 3 2 NH2 H 3 PO4 routes to the API was mandatory and was Yield: 88-92%, 12 1 free base Quality > 99% 1 therefore pushed forward in parallel to Acetamidoazide Oseltamivir the technical development of the routes Phosphate described above. The main synthetic challenges regarding the target molecule 1 are O O CO2 Et CO2 Et a) the efficient induction of three stereo- genic centers, AcHN AcHN b) the regioselective introduction of the NH NH 2 2 1,2-double bond, 27 28 c) the introduction of the two amino func- tions and Scheme 9. Selective azide reduction of the acetamidoazide 12 to oseltamivir ( 1 ) (free base) and d) the formation of the 3-pentyl ether side phosphate salt formation chain. HOT TOPICS: SANDMEIER PRIZE 2006 98 CHIMIA 2007, 61, No. 3

From the various proposals and concepts collected from Roche chemists via a 1) AllylNH 2 , 2eq. MgBr . OEt ,0.2 eq. 1) 10% Pd/C,EtOH global request, not surprisingly Diels-Alder 2 2 H NCH CH OH TBME /MeCN 9:1 CO Et 2 2 2 O 2 O CO2 Et approaches were prevalent followed by a O CO2 Et 55 °C, 16 h rflx, 3h large variation of further proposals. 2) NH Cl / H O 4 2 H O 2) H 2 SO4 ,1.1 eq. H O O H N 97% 77% NH2 5.3.2. Diels-Alder Approaches 8 29 30 From a large number of Diels-Alder Epoxide 1) PhCHO ,TBME, -H O concepts tested[4] the furan-ethyl acrylate 2 "Domino" 2) MsCl,NEt3 ,filtr. Sequence concept shown in Scheme 12 finally led to 3) AllylNH ,3eq., 111 °C, 16 h 2 80% + the first shikimic acid independent access 4) H /H2 O to oseltamivir phosphate ( 1 ). The synthe- 1) 10% Pd/C, EtOH, Ac2 O ,1eq. sis started with the Zn-chloride promoted H 2 NCH2 CH2 OH AcOH,10eq. rflx.,3h MeSO H ,1eq. Diels-Alder reaction of cheap furane and 3 O CO Et 2) H 3 PO4 , EtOH O CO2 Et TBME, 15 h, r.t. O CO2 Et ethyl acrylate leading preferentially to the 2 AcHN 70% 83% desired exo-isomer 33, the product of ther- AcHN H 2 N NH •HPO H N H N modynamic control. Enzymatic resolution 2 3 4 then allowed access to the R -isomer (–)-33 1 32 31 in about 20% yield (not optimized). Oseltamivir overall ~35-38% phosphate from 8 Addition of diphenylphosphoryl azide in a [3+2] fashion to (–)-33 followed by thermal Scheme 10. Azide-free transformation of the epoxide 8 to oseltamivir phosphate ( 1 ) nitrogen extrusion and trans-esterification at phosphorus delivered – unexpectedly[4] – the endo-isomer of aziridine 34 which smoothly underwent eliminative ring opening to form 35 followed by direct O-mesylation and re- O CO Et O CO2 Et MsCl O CO2 Et gio- and stereoselective introduction of the 2 -NEt . HCl 3-pentyl-ether side chain under Lewis-acid H O -H2 O H O 3 MsO N N NH catalysis to provide 36. Acid-catalyzed N–P- NH2 2 bond cleavage then led to 37 isolated as the 30 hydrochloride. The specific configuration - N of this intermediate finally allowed for an 1) PhCHO, TBME, -H2 O O CO2 Et azide-free transformation to the drug sub- 2) MsCl, NEt3 ,filtr. 80% Büchi 3) AllylNH ,3eq., 111 °C, 16 h stance 1 in analogy to the processes shown 2 Autoclave MsO 4) H + /HO in the Schemes 10 and 11. 2 111 °C / 16 h NH TBME 2 5.3.3. Aromatic Ring Transformations: -MsOH O CO2 Et O CO Et H + ,HO H N The Desymmetrization Concept 2 2 2 O CO2 Et Taking advantage of a desymmetriza- N H N H N tion protocol over racemate cleavage re- 2 H N H N garding effectiveness, the meso-approach shown in Scheme 13 is based on the enzy- 31 matic hydrolysis of the all-cis meso-diester Scheme 11. Reaction cascade combining six chemical steps but only one final work-up and isolation [4] 41 to the optically active mono-acid 42. of 31 Starting from cheap dimethoxy phenol 38 etherification with 3-pentyl mesylate, di- 1) enzymaticresolution R/S bromination and ethoxycarbonylation gave CO2 Et Chirazyme L-2 R 0.7 eq. 2) (distillation) the symmetrically substituted isophthalic CO2 Et O CO2 Et diester 38 which was hydrogenated over ZnCl2 1.0 eq. ~20% O Ru-Al O providing exclusively the all-cis neat, 50 °C, 72 h O 2 3 77% meso-diester 40. 33 (-)-33 After selective O-demethylation the re- exo/endo ~9:1 1) DPPA, toluene, 70 °C, 18 h 53% 2) NaOEt, EtOH, 1h, r.t. (chromat.) sulting meso-diester 41 was enzymatically OEt EtO 1) MsCl, NEt hydrolyzed using pig liver esterase (PLE) 3 P NaHMDS, THF N H CO Et CH2 Cl2 ,r.t. OEt CO Et O affording the optically active mono-acid EtO O 2 2 -60°C, 15 h EtO EtO P N P CO2 Et 42 in almost quantitative yield and with N 2) 3-pentanol O 94% (crude) O . H BF3 OEt2 ,CH 2 Cl2 high optical purity ( ee: 96–98%). The con- OMs OH O 62% (cryst.) version of the β -hydroxy-acid 42 into the 36 35 34 drug substance 1 was straightforward and 1) 20% H SO in EtOH 68% 2 4 started with Yamada-Curtius degradation (cryst.) 70 °C, 22 h 2) HCl /EtOH cryst. of 42 allowing the introduction of the 5- Allylamine Route amino-functionality by formation of the O CO Et O CO Et 2 2 Oseltamivir oxazolidinone 43. In an efficient ‘one-pot’ phosphate AcHN procedure the N-Boc-protected oxazolidi- H 2 N NH •HPO none was treated with catalytic amounts of •HCl OMs 2 3 4 sodium hydride in refluxing toluene trig- 37 1 gering the effective and selective formation Scheme 12. Furan-ethyl acrylate Diels-Alder approach to oseltamivir phosphate ( 1 ) HOT TOPICS: SANDMEIER PRIZE 2006 99 CHIMIA 2007, 61, No. 3

Int. Appl. WO 9116320, priority date: OMs all-cis hydrogenation 24.04.1990.

O 1) KOtBu, DMSO O O [3] J. C. Rohloff, K. M. Kent, M. J. Postich, H (100 bar) CO Et 2 H O O 2 O CO2 Et 2) NBS, DMF 5% Ru-Al2 O 3 M. W. Becker, H. H. Chapman, D. E. Kel-

O 3) CO, EtOH, KOAc O EtOAc, 60 °C O ly, W. Lew, M. S. Louie, L. R. McGee, E. J. 82% Pd(OAc)2 /dppp cat. CO Et 2 CO2 Et Prisbe, L. M. Schultze, R. H.Yu, L. Zhang, 85% (3 steps) 38 39 40 J. Org. Chem. 1998, 63, 4545. TMSCl, NaI 97% [4] S. Abrecht, P. Harrington, H. Iding, M. MeCN, 40 °C Karpf, R. Trussardi, B. Wirz, U. Zutter, desymmetrization by OH enzymatic hydrolysis Chimia 2004, 58, 621. O CO Et OH OH 2 [5] M. Federspiel, R. Fischer, M. Hennig, H. CO Et O CO Et DPPA, NEt3 O 2 PLE, H 2 O 2 O J. Mair, T. Oberhauser, G. Rimmler, T. Al- pH 8.0, 96% NH CH2 Cl2 ,40°C H O H O 81% ee: 96-98% biez, J. Bruhin, H. Estermann, C. Gandert, O CO H CO Et 2 2 V. Göckel, S. Götzö, U. Hoffmann, G. 43 42 41

1) (Boc)2 O, DMAP cat. Huber, G. Janatsch, S. Lauper, O. Röckel- 83% 2) NaH cat., tol. rflx. Stäbler, R. Trussardi, A. G. Zwahlen, Org. 3) Tf2 O, py, DCM Proc. Res. and Dev. 1999, 3 , 266. [6] S. C. Chandran, J.Yi, K. M. Draths, R. von 1) H 2 , Ra-Co O O CO Et 2) Ac2 O/Et3 N O CO Et CO2 Et 2 2 Daeniken, W. Weber, J. W. Frost, Biotech- NaN3 3) HBr-AcOH nol. Prog. 2003, 19, 808; D. R. Knop, K. TfO acetone/H O, r.t. N AcHN 2 3 4) H 3 PO4 ,EtOH NH •HPO HNBoc 78% HNBoc 2 3 4 M. Draths, S. C. Chandran, J. L. Barker, 83% 44 45 1 R. von Daeniken, W. Weber, J. W. Frost, J. Oseltamivir phosphate Am. Chem. Soc. 2001, 123, 10173. [7] M. Karpf, R. Trussardi, J. Org. Chem. Scheme 13. Synthesis of oseltamivir phosphate ( 1 ) via desymmetrization of meso-diester 41 2001, 66, 2044. [8] H. J. Mair, F. Hoffmann-La Roche AG, EP 1112999-A2, priority date: 03.12.1999. of the 1,2-double bond and – at the same the experienced organic chemists with [9] P. J. Harrington, J. D. Brown, T. Fodera- time – the cleavage of the oxazolidinone manufacturing experience to evaluate the ro, R. C. Hughes, Org. Proc. Res. & Dev. system, followed by the introduction of technical feasibility of these syntheses in 2004, 8 , 86. a good leaving group to provide the tri- comparison to the currently used commer- [10] V. Farina, J. D. Brown, Angew. Chem. Int. flate 44 in an overall yield of 83%. The cial route using readily available shikimic Ed. 2006, 45, 7330. 4-amino functionality was finally intro- acid ( 15) and also to the alternative routes [11] Y.-Y. Yeung, S. Hong, E. J. Corey, J. Am. Chem. Soc. 2006, 128, 6310. duced via S N 2-substitution of the triflate described above. 44 using sodium azide with concomitant [12] Y. Fukuta, T. Mita, N. Fukuda, M. Kanai, inversion of configuration under mild, ba- Acknowledgements M. Shibasaki, J. Am. Chem. Soc. 2006, sic conditions. Azide reduction, N-acety- The authors thank all their coworkers for 128, 6312; T. Mita, N. Fukuda, X. Roca, lation, Boc-deprotection and phosphate their full commitment and their hard work in all M. Kanai, M. Shibasaki, Org. Lett. 2007, salt formation gave the final product osel- our attempts to finally succeed in this difficult 9 , 259; K. Yamatsugu, S. Kamijo, Y. Soto, tamivir phosphate 1 in an overall yield of task. The careful reading of the manuscript by M. Kanai, M. Shibasaki, Tetrahedron Let- Ulrich Widmer and Roland Thieme is gratefully ters 2007, 9 , 259. 30%, starting from 1,6-dimethoxy phenol acknowledged. 40 comparing favorably with the shikimic acid based route. Received: January 23, 2007 Recently several academic groups embarked on shikimic acid independent syn- [1] N. W. Bischofberger, C. U. Kim, W. Lew, theses of oseltamivir phosphate ( 1). [10] An H. Liu, M. A. Williams, Gilead Sciences elegant enantioselective Diels-Alder based Inc., USA, PCT Int. Appl. WO 9626933, route by Corey et al. [11] as well as synthe- priority date: 27.02.1995. ses applying desymmetrization and Diels- [2] L. M. von Itzstein, W. Y. Wu, P. T. Van, Alder concepts by Shibasaki et al.[12] have B. Danylec, B. Jin, Biota Scientific been published. We have to leave it up to Management Pty. Ltd., Australia, PCT