key: cord-0997630-9scinswr
authors: Sagandira, Cloudius R.; Mathe, Francis M.; Guyo, Upenyu; Watts, Paul
title: The evolution of Tamiflu synthesis, 20 years on: Advent of enabling technologies the last piece of the puzzle?
date: 2020-07-26
journal: Tetrahedron
DOI: 10.1016/j.tet.2020.131440
sha: 6a8b77eb1042a8aa5b3da7e7f18e69e1311b4f5f
doc_id: 997630
cord_uid: 9scinswr

Influenza is a serious respiratory disease responsible for significant morbidity and mortality due to both annual epidemics and pandemics; its treatment involves the use of neuraminidase inhibitors. (−)-Oseltamivir phosphate (Tamiflu) approved in 1999, is one of the most potent oral anti-influenza neuraminidase inhibitors. Consequently, more than 70 Tamiflu synthetic procedures have been developed to date. Herein, we highlight the evolution of Tamiflu synthesis since its discovery over 20 years ago in the quest for a truly efficient, safe, cost-effective and environmentally benign synthetic procedure. We have selected a few representative routes to give a clear account of the past, present and the future with the advent of enabling technologies.

Tamiflu, also known as oseltamivir phosphate 1 is a potent chemotherapeutic agent for influenza treatment. [1] [2] [3] Influenza is a severe viral infection of the respiratory system regarded as the most serious respiratory disease, which is responsible for significant morbidity and mortality due to both annual epidemics and predictable pandemics. [1] [2] [3] More specifically, the avian influenza H5N1 has a mortality rate of about 60 %. 4 Unfortunately, little has been done to change the influenza infection patterns in past decades despite influenza being the most studied viral infection before the arrival of the human immunodeficiency virus (HIV). 5, 6 Now in this global society, a highly aggressive strains of the influenza virus such as the H5N1 can mutate to become easily transmitted from human to human and spark another deadly pandemic just as with current Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Infection (COVID19) pandemic, 7, 8 which has brought the world to its knees. The surge in drug-resistant influenza strains resulting from naturally occurring mutations reminds us of the need for continued research to discover more potent neuraminidase inhibitors. [9] [10] [11] [12] [13] The current anti-influenza drugs are useful templates in the development of new neuraminidase inhibitors through structure-activity relationship studies. As the research towards new and better treatment remains a top priority, it is equally important to improve the availability of the current anti-influenza drugs by developing better synthetic procedures to guard the world against influenza. Drugs such as oseltamivir phosphate (Tamiflu) 1a, amantadine HCl 1b, rimantadine HCl 1c, zanamivir 1d and peramivir 1e have been developed for the treatment of influenza over the years 1-3 and more recently, baloxavir marboxil 1f was developed ( Figure 1 ). 14 The last four are FDA approved and are currently the recommended influenza drugs by Centres for Disease Control and Prevention (CDC) . 15 Of all the drugs, Tamiflu is the most commonly used since its FDA approval in 1999 making its synthesis an important research area. Hoffmann-La Roche Ltd and marketed by F. Hoffmann-La Roche and commercially launched in November 1999. [16] [17] [18] [19] [20] [21] In the early years of its discovery, (-)-shikimic acid was used as starting material for the synthesis of Tamiflu and furthermore, the current and only industrial synthetic route still uses (-)-shikimic acid. In response to an increasing threat of an influenza pandemic, diverse synthetic approaches have been developed and very insightful reviews of their relative merits have been published. [1] [2] [3] [22] [23] [24] [25] [26] [27] However, there were legitimate (-)-shikimic acid availability concerns in the early years of the development of this drug.

Shikimic acid, which is a natural product isolated from a plant of Chinese star anise was unavailable in consistent purity and enough quantity, which prompted extensive studies into (-)-shikimic acid free synthetic routes in both industry and academia. Fortunately, this has been addressed by the development of more efficient extraction and purification processes or alternatively by fermentation using a genetically engineered E.coli bacteria. 25, [28] [29] [30] Furthermore, Yoshida and Ogasawara demonstrated an enantioconvergent synthesis of shikimic acid via a palladium mediated elimination reaction. 31 Although more studies utilising alternative starting materials are still ongoing, they have not been as efficient as the current shikimic acid based production route. The use of the potentially hazardous azide chemistry for the introduction of amino and acetomido groups to the ring was, and is still, a major concern. 1,2,32 Azide chemistry poses many safety concerns because of its hazardous and highly exothermic nature, which become more pronounced at a large scale. 2,33-37 As a result, numerous studies towards azide-free synthetic routes were done, which unfortunately have not been as good as the current production route. Herein, this review highlights the evolution towards efficient and safe synthetic routes of Tamiflu since its first approval 20 years ago.

Since there has been over 70 published synthetic routes and some review articles, 1-3,22-26 a few selected representative routes will be used to give a clear account of the past, present and the future with the arrival of enabling technologies 38 such as flow chemistry.

Oseltamivir carboxylate 2 was the first molecule identified by Gilead scientists for development, but the ethyl ester prodrug oseltamivir phosphate (Tamiflu) 1a was ultimately chosen as the clinical candidate based on its potent in vitro and in vivo activities and its good oral bioavailability after extensive diversity-oriented discovery chemistry studies by Kim et al. [16] [17] [18] [19] [20] [21] Gilead Sciences researchers first synthesised the oseltamivir carboxylate 2 from a natural product, (-)-shikimic acid 29, as the starting material (Scheme 1). 20 (-)-Shikimic acid derivative 3 was treated under Mitsunobu conditions resulting in selective activation of the least selectively hindered OH-group at C-5 whilst the C-3 OH is MOM protected, affording epoxide 4. 20 Epoxide 4 was subsequently opened regio-and stereospecifically using azide chemistry, selective azidating the C-5 to afford azido alcohol 5. Mesylation of 5, followed by azide reduction afforded aziridine 6. Once more, azide chemistry was utilised in regioselective aziridine-opening at C-5 followed by MOM group cleavage affording amino alcohol 7. 20 Aziridine 8 was synthesised from 7 by a two-step, one-pot process: (1) protection of the amino functionality with a trityl group, and (2) mesylation of the hydroxyl group.

Regio-selective ring-opening of aziridine 8 with 3-pentanol in the presence of Lewis acid catalyst BF 3 ·OEt 2 subsequently followed by acetylation of the resulting amine afforded the corresponding amido ether. The azide group on the resulting amido ether was reduced, followed by hydrolysis of the methyl ester under basic conditions affording oseltamivir carboxylate 2 in 15 % overall yield over the 14 steps despite using protecting group chemistry. 20 The choice of their starting material (-)-shikimic acid was justified; it has the carbocyclic system with chirality which is also present in the target compound 2 or which can be used to handle the introduction of the desired stereochemistry. However, at that time, (-)shikimic acid availability was one of the major drawbacks since effective extraction and purification methods had not been developed. The use of potentially explosive azidecontaining intermediates is another drawback associated with this synthetic route, which restricted the synthesis to milligram scale.

Scheme 1: Gilead Sciences' synthetic route of the first candidate 2 for development 20 Due to scarcity of (-)-shikimic acid in large quantities at the time, 17, 39 Gilead scientists went on to prepare Tamiflu 1a at multi-gram scale from more available (-)-quinic acid 9 (Scheme 2). 39 The first large scale route by Gilead sciences from (-)-quinic acid consisted of 12 steps and afforded an overall yield of 4.4 %. 39 Despite the relative low yield, it was successfully implemented in a standard pilot plant producing kilogram quantities of Tamiflu 1a and the potentially hazardous azide chemistry was safely handled. Furthermore, minimal protecting group manipulations were employed and no chromatography was required for isolation.

First large scale synthesis of Tamiflu 1a by Gilead Sciences 39

There has been extensive research by Gilead Sciences, [16] [17] [18] [19] [20] 39 Roche 30, [40] [41] [42] [43] [44] [45] [46] and other numerous scientific laboratories to develop an efficient, safe large-scale route towards Tamiflu since 1995. [1] [2] [3] [22] [23] [24] [25] [26] The efforts led to the 12-steps Roche industrial route starting from (-)-shikimic acid affording an overall yield of ~35 % (Scheme 3) 46 after (-)-shikimic acid availability had been improved by the development of more efficient extraction and purification processes or alternatively by fermentation using a genetically engineered E.coli bacteria. 25, 28, 29 As with Gilead Sciences approaches (Scheme 1 and 2), 20,39 the Roche industrial route utilised azide chemistry in structuring the 1,2-diamine moiety in 28. 46 Contrary to the Gilead Sciences' routes, where the pentyloxy group is introduced in the latest stage (Scheme 1 and 2), 20, 39 Roche introduced the 3-pentyloxy moiety early at C-3 by regio-selective reduction of acetal 18. 46 However as with the Gilead routes (Scheme 1 and 2), the Roche approach has drawbacks that include the challenge to safely handle the thermally unstable azide reagents and intermediates on large scale. Another drawback was associated with the utilisation of shikimic acid which was scarce at that time. These drawbacks prompted the scientific community to extensively develop numerous alternative routes.

Roche's industrial synthesis 46

Although the Roche industrial route is currently supplying the world with enough tonnes of Tamiflu, the route raised three concerns as aforementioned: a) the use of shikimic acid, which had limited availability in the early days of development b) the use of potentially explosive azide chemistry and c) long synthetic route with low overall yield. Consequently, this prompted the development of numerous alternative synthetic approaches to address the concerns. [1] [2] [3] [22] [23] [24] [25] [26] 47 These approaches can generally be categorised into two main classes:

shikimic acid-dependent and shikimic acid-independent approaches in which both the azidedependent and azide-independent approaches can be subclasses. Herein, a few selected and representative alternative synthetic strategies that can potentially be scaled-up after minimal modifications are highlighted in each class.

After (-)-shikimic acid availability improvement with time, researchers embarked on the development of shorter and higher yielding synthetic routes. These routes can be classified into two groups: namely azide chemistry dependent and azide-free routes.

Azide-dependent routes utilize potentially hazardous azide chemistry to introduce two amino groups on the cyclohexene ring system. The authors started with (-)-shikimic acid 29 esterification to afford ethyl shikimate 35

according to a reported procedure. 46 The synthesis proceeded via the O-trimesylate 36

followed by regio-and stereoselective nucleophilic substitution of the allylic O-mesylate at C-3 by an azide group resulting in azide 37. In the presence of triethyl phosphite (P(EtO) 3 ) and under reflux, aziridine 38 was formed from the azide 37. This was followed by regio-and stereoselective aziridine ring opening at the allylic position with 3-pentanol in the presence of a Lewis acid catalyst providing 39. N-P bond cleavage in 39 followed by N-acetylation afforded mesylate 40, which went on to react with sodium azide to afford 27. The azide group was reduced to the amine group followed by the addition of phosphoric acid (H 3 PO 4 )

affording the drug Tamiflu 1a. Protecting group manipulations, chromatographic separations and tedious purifications were not required in this route. It proceeded with an unoptimised overall yield of 20 %, utilising cheap and commercially available chemicals. 30 Although the route could neither avoid azide chemistry nor improve the overall yield, it presents a potentially scalable, elegant and shorter synthetic route. In addition to Karpf and Trussardi 30 procedure, Trussardi 44 disclosed a similar process in which compound 39 could alternatively be azidated first before acetylation.

In 2013, Kalashnikov et al. 28 reported almost a similar synthetic route to that of Karpf and Trussardi. 30 The route afforded an optimised overall yield of 27 %. Although this 10-step procedure utilises azide chemistry and is accompanied by lower overall yield, it uses minimum amount of expensive reagents making it attractive.

Utilising the experience accumulated from their first 13 steps approach (44 % overall yield) from (-)-shikimic acid, 48 Shi's group developed an optimised 8 steps route (Scheme 5). 49 The route is almost similar to Karpf's approach, 30 A short and practical approach towards Tamiflu 1a with an impressive 47 % overall yield was developed by Shi et al. 49 The overall yield improved slightly (35 to 47%) and the number of transformations were considerably reduced relative to the Roche industrial approach (12 steps to 8 steps). 46, 49 This approach represents a model of atom economy since no protecting group manipulations were needed. Unfortunately, the researchers resorted to the potentially hazardous azide chemistry on two occasions after other safer nitrogen-containing nucleophiles such as ammonia, benzylamine and allylamine failed to afford the desired products. 49 This drawback is however more than compensated by the elegance, simplicity and efficiency of the approach. This synthetic route is evidently a major player in the goal to develop an efficient scalable process towards oseltamivir phosphate synthesis. However, there is need to find ways of dealing with the potentially hazardous steps involved to guarantee a truly scalable and safe process, which is applicable in industry.

Building on their earlier work in which they synthesised Tamiflu 1a from shikimic acid via cyclic sulfite intermediates, 50 Shi and coworkers reported an improved 11-step route starting from shikimic acid via a 3,4-cyclic sulfite intermediate 43 affording Tamiflu in 55 % overall yield (Scheme 6), 51 which is significantly better than the industrial route (35 % overall yield).

All the transformations were clean with each step affording yields greater than 90 %, meaning that subsequent steps could be performed without purification of crude products.

This approach generally shares some similarities with other shikimic acid dependent routes such as the use of azide chemistry and can be scaled-up easily.

Shi's 11-step Tamiflu 1a synthetic route via a 3,4-cyclic sulfite intermediate 43 51 Generally, the shikimic acid-azide chemistry dependent procedures proceeded in good yields.

They can potentially be performed at large scale if the azide chemistry safety is guaranteed, as with the current industrial route. This can be achieved by performing holistic process calorimetric studies before scale-up, use of highly skilled personals and working under very strict conditions. This is usually not easy to achieve. However, the use of either enabling technologies such as continuous flow technology that are known to enhance process safety or alternative safe chemistry to introduce the two amino groups on the cyclohexene ring system can be considered.

Most recently, Watts group reported a 8-step total flow synthesis of Tamiflu starting from ethyl shikimate 35 derived from shikimic acid (Scheme 7). 52, 53 Taking lessons from the previously reported shikimic acid-based routes, 28, 30, 48, 49, 51, 54 the authors aimed to ensure azide chemistry safety, processing time reduction and process overall yield improvement by taking advantage of continuous flow chemistry technology. Flow chemistry technology is an enabling technology, which has attracted considerable attention in synthetic chemistry and pharmaceutical industry owing its efficiency, easy scale-up, safety and reproducibility;

industry is now using the technology up to 2000 tonnes per annum. [55] [56] [57] [58] [59] This has seen numerous approaches for pharmaceutical drugs being redesigned into continuous flow synthesis. 56, 58, [60] [61] [62] [63] [64] [65] The technology allows for in situ generation and consumption of dangerous intermediates, preventing their accumulation thus enhancing process safety. 55, [66] [67] [68] Additionally, microreactors can handle exotherms extremely well, due to the inherent high surface area to volume ratio and rapid heat dissipation unlike the conventional batch process. 55 The authors demonstrated an efficient synthetic route for Tamiflu 1a with 58 % overall yield and 3.5 minutes total residence time starting from ethyl shikimate 35. This process elegantly handled the hazardous azide chemistry involved in this procedure by taking advantage of flow chemistry technology. The overall yield of the process is literature comparable, however, processing time is significantly shorter than all the reported procedures, which are mostly greater than 30 h. 1,2 Without doubt, this presents a safe, efficient and scalable procedure for the synthesis of the drug.

Continuous flow synthesis towards Tamiflu 1a by Sagandira and Watts

Due to azide safety concerns, azide-free routes were developed to introduce the two amino groups on the cyclohexene ring system. 39, 70, 71 The goal was to identify a non-azide nucleophile that is compatible with the rest of the functional groups on the molecule and with to afford Tamiflu 1a. The overall yield for this azide-free approach was 35-38 %, which is comparable to the industrial route. 45, 46 This approach represented the first example that avoids the utilisation of potentially hazardous azide chemistry, the convectional way of introducing nitrogen functionality on the ring. 

The legitimate (-)-shikimic acid availability concerns in the early years of the development of Tamiflu led to the development of shikimic acid-free routes. Although the shikimic acid availability improved with time, alternative routes are still being explored to date. Unlike the shikimic acid-dependent approaches which take advantage of the already present chiral cyclohexene backbone to introduce the groups at C3, C4, and C5 with the desired stereochemistry, the shikimic acid independent approaches construct the cyclohexene backbone through various strategies such as Diels-Alder reaction, Horner-Wadsworth-Emmons reaction, aldol condensation, Michael addition, sugars, nitroalkenes by Curtius rearrangement. These approaches display ingenuity in the construction of the cyclohexene ring system, the induction of the three stereogenic centres, the introduction of the two amino groups, the introduction of the 3-pentylether side chain and the regioselective introduction of the 1,2-double bound on the cyclohexene ring of the drug starting from readily available and affordable starting materials.

Approaches in this class utilise the Diels-Alder reaction to construct the cyclohexene ring system of the drug based on the investigations by Brion. 72 This procedure was characterised by extensive protection group chemistry, very low yield as well as the use of azide chemistry. 78 In 2007, Shibasaki and coworkers reported a 12-step approach towards Tamiflu 1a via the Diels-Alder reaction and Curtius rearrangement reaction as the key steps. 77 In this approach, chirality is introduced with the help of a chiral ligand, and relies on the preparative chiral HPLC to obtain enantiomerically pure material (Scheme 11). 77 The cyclohexene skeleton was synthesised through a Diels-Alder reaction between diene 70

and dienophile 79 affording 80 on a 58g scale. 47 This approach is unique in its chirality introduction as it is introduced in the first step through asymmetric Diels-Alder reaction. This considerably increased the overall yield (16%) of the approach compared to their previous approach. 47, 77 The synthetic route is also characterised by low catalytic loading which is advantageous. An interesting way of introducing an ester group on the cyclohexene using malononitrile was unearthed. Although the procedure is low yielding compared to the current industrial, it has the potential to become a scalable process if Although the authors demonstrated an improved and effective iron-catalyzed procedure for

highly functionalized yet electronically deactivated substrates that have been otherwise problematic and an oligomeric iron-azide catalyst was uniquely effective for the stereoselective diazidation, the procedure was accompanied by an unimpressively low overall yield (5 %). Although the chemical hazard assessment of olefin diazidation using both differential scanning calorimetry (DSC) and the drop weight test (DWT) demonstrated the feasibility of performing this olefin diazidation reaction for large scale synthesis of Tamiflu 1a, an excess of TMSN 3 (5 equiv.) was necessary for the reactivity and the reaction was only scaled up to 5 g scale. 

Hayashi's group first 'one-pot' sequence towards oseltamivir 28 91 Hayashi's second 'one-pot' sequence towards oseltamivir starts with the treatment of compound 108 (made in the first 'one-pot' sequence) with TFA, thus cleaving the tert-butyl ester, to afford carboxylic acid 109 which was subsequently converted to acid chloride 110 via oxalyl chloride and catalytic DMF treatment. 91 To address the aforementioned safety concerns associated with the use of potentially explosive acyl azide 111, 91 Hayashi and coworkers 32 

Affordable and abundantly available sugars such as D-xylose, 101 D-ribose, 102 D-mannitol, 103 D-glucal, 104 and D-glucose 105 have also been used as starting material for Tamiflu 1a synthesis. 1, 2, 24, 26 However, sugar dependent synthetic procedures are characteristically long Furthermore, the synthesis had a mechanistic step which required reaction at high temperature in a sealed tube in addition to protection group chemistry. These limitations make it difficult for the synthesis process to be scaled up to industrial scale. 

As with the current Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Infection (COVID19) pandemic, a highly aggressive strain of the influenza virus such as the H5N1 can mutate and spark another deadly pandemic. This reminds us of the need for continued research to discover more potent neuraminidase inhibitors. As the research towards new and better treatment remains a top priority, it is equally important to improve the availability of the current anti-influenza drugs by developing better synthetic procedures to guard the world against influenza. Though other influenza antiviral drugs such as peramivir, zanamivir and baloxavir marboxil are on the market, Tamiflu remains the most used antiinfluenza drug 20 years after its approval. Consequently, Tamiflu synthesis remains an important research area. Although the Roche industrial route is currently supplying the world with enough tonnes of Tamiflu but it remains to be seen if will be suffice in the face of a pandemic, the route raised three concerns; a) the use of shikimic acid, which had limited availability in the early days of development b) the use of potentially explosive azide chemistry and c) long synthetic route with low overall yield. To address these concerns, academic and industrial researchers have made tremendous efforts in the quest for a truly efficient, safe, cost-effective and environmentally benign synthetic procedure resulting in more than 70 synthetic procedures to date since its discovery.

As reflected in this review and complimented by the published reviews, [1] [2] [3] [22] [23] [24] [25] [26] [27] 47 both the shikimic acid-free and azide chemistry-free approaches are mostly low yielding compared to their competitors despite their ingenuity in Tamiflu assembling. Having solved the shikimic acid availability concerns over the years by developing more efficient extraction and purification processes or alternatively by fermentation using genetically engineered E.coli bacteria, 25,28-30 the use of hazardous azide chemistry needs more attention. The application of continuous flow technology in Tamiflu synthesis proved to be a potential enabling tool for safe handling of the hazardous azide chemistry as well as improving efficiency. 32, 53, 92 Continuous flow synthesis has attracted considerable attention in synthetic chemistry and pharmaceutical industry in the last decade owing to its well-documented advantages, [55] [56] [57] [58] [59] 65, 67, 79, [96] [97] [98] [99] [100] resulting in numerous pharmaceutical drugs approaches being redesigned into continuous flow synthesis. 56, 58, [60] [61] [62] [63] [64] [65] 106 In this light, we envisage that Tamiflu synthesis can hugely benefit from continuous flow technology application to afford truly efficient synthetic procedures. Furthermore, the promising Tamiflu synthetic approaches which were previously ruled out for large scale synthesis in batch based on either safety concerns or poor efficiency can be reconsidered in flow. We envisage that the incorporation of other enabling technologies such as artificial intelligence, machine learning for "Big data" analysis, can challenge the dogma of the past and come up with a truly efficient, safe, cost-effective and environmentally benign Tamiflu synthetic procedure. With this in mind, we are looking

forward to see what the future of Tamiflu synthesis holds.

Laude). In the same year, he joined Professor Paul Watts' Group where worked on the development of an integrated continuous flow system for the synthesis of biodiesel from waste cooking oil and graduated in 2018 with a Msc Chemistry degree. Currently, he's in the same group working towards attaining his PhD on the flow synthesis of silicon compounds from low-grade silicon. Processing'. He has published over 120 highly cited papers. He strongly believes that scientists should conduct research that impacts society; the biggest project underway involves the local production of key drugs as the morbidity and mortality from major diseases are much more devastating in Africa than in other regions of the world. The vision

Standard guidelines for the clinical management of severe influenza virus infections

Influenza : The Once and Future Pandemic

Semin Respir Crit Care Med

Organic Azides: Syntheses and Applications

Angew Chemie -Int Ed

Angew Chemie -Int Ed

New Technology for Modern Chemistry

Angew Chemie -Int Ed

Synthesis (Stuttg)

Angew Chemie -Int Ed

Angew Chemie -Int Ed

Org Process Res Dev. 2020. is that new technology will be used within South Africa to manufacture generic drugs

We thank the National Research Foundation (NRF SARChI Grant), Council for Scientific and Industrial Research (CSIR-DST Grant) and Nelson Mandela University for financial support.

Cloudius R. Sagandira 

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: