Chirality remains a core feature of many nces. Dr Sarah Houlton explores some of the new routes being devised by development chemists to ensure the most efficient synthesis is achieved
The number of chirally pure drug molecules has increased dramatically in recent times. While about half of the new drugs that reached the market in the past dozen years have no chiral centres, most of the rest are single enantiomers, in line with the regulatory authorities" reluctance to approve racemates unless there is a very good reason for it not to be chirally pure.
This means that a huge amount of effort has been put into developing a whole range of chiral syntheses that work on a production scale, not just in the lab. Techniques include the separation of racemates by resolution, chiral catalysts and auxiliaries, and biosynthetic processes, as well as starting with a molecule from the chiral pool that is readily available in large quantities, such as amino acids and sugars. This introduces chirality into the molecule right at the start using a chiral building block.
Even when no chiral pool molecule is available, it is often most efficient to introduce one or more chiral centres right at the beginning of the synthesis, because the starting materials are less complex and thus less likely to interfere, and the centres can then be used to induce further chirality in subsequent synthetic steps. As a result, the syntheses of many chiral drugs start with smaller chiral building blocks.
HIV treatment
In the past few years, HIV infection has largely been transformed from a death sentence to a manageable condition, thanks to the introduction of cocktails of antiviral drugs that keep the virus at bay. However, as the virus mutates to evade the drugs" action, new agents are required if the cocktails are to remain active in the long term. A new protease inhibitor, A-792611, being developed by Abbott, also has the advantage of having better bioavailability than many of its predecessors, and something which has been a significant issue in the past.
The drug has a chiral amino alcohol core, and the Abbott team felt that an ideal intermediate would be a chiral epoxide derived from phenylalanine because it already has the correct 1,2-anti stereochemistry.1 However, there was a problem - existing routes gave a 1.6:1 mixture of diastereomers when it was elaborated to the amino alcohol. They felt that it ought to be possible to exploit the existing chirality of the epoxide to direct the chirality of the third stereocentre.
By treating the epoxide with diethylmalonate, a 1:1 mixture of epimers of the intermediate lactone was obtained. This was then reacted with 2-(a)-bromotolyl)pyridine to give a quaternary intermediate ester, which was then decarboxylated to get the desired lactone. They found that this product had been formed in a 5:1 ratio of diastereomers, which could be improved to 97:3 by recrystallisation. The vital diastereoselective step in this process is the protonation of the enol that results from the decarboxylation reaction, and molecular modelling indicated that the side of attack that results in the desired product is by far the easiest. This substrate directed kinetic enol protonation proved to be an effective, selective route into the chiral core of A-792611.
Chiral ß-amino acids are common features in drug molecules, and there are a number of different ways in which they can be synthesised selectively. A group at Kyoto University in Japan has developed a new synthesis for ß-amino acids, including the chiral cores of two antibiotic molecules, otamixaban and premafloxacin.2 The chirality is controlled by the ligands included in the reaction.
They started with either a cyclic or an acyclic enoate, and carried out a conjugate addition of lithium arylmethyl(trimethylsilyl)amides or allyl(trialkylsilyl) amides. This gives the corresponding ß-alkylaminoalkanoates after the protonation of the intermediate enolate, and allows the construction of two chiral centres in one pot, one after the other. The chirality is controlled by a chiral ligand, and the absolute chirality of the product is determined by the enantiomer of the ligand that is included in the reaction. The major products of these reactions resulted from anti-alkylation, and were isolated in ees of up to 98%.
The standard protocol for removing the benzyl group did not work in this case because other functionality within the molecule interfered with the reaction. Instead, the team developed an alternative route, involving N-chlorination of the benzylamines and DBU-mediated elimination, giving benzylimines which could then be transoxaminated with hydroxylamine to give the desired free ß-amino acid.
Roche's oseltamivir (Tamiflu) has been much in the news in recent months because of its potential to treat avian influenza, and the World Health Organization and many governments have been stocking up on the drug in case a pandemic does occur. The current commercial production route uses the natural product (-)-shikimic acid as starting material, which has three chiral centres. However, this is expensive and it is not easily available in large quantities. As a result, alternative syntheses that use cheaper ways of introducing the chirality would be extremely useful in ramping up production to meet demand. A group at the University of Tokyo has developed a route that employs the desymmetrisation of meso-aziridines by nitrogen nucleophiles to create a chiral 1,2-diamine with the correct stereochemistry for oseltamivir.3
Unusually, the group used a rare earth catalyst. After screening several metal cores, including gadolinium, dysprosium, erbium and scandium, they found that 10mol% of a yttrium complex with isopropyloxy ligands gave the best results, in combination with a fluorinated chiral ligand to induce asymmetry in the product, and the nucleophile trimethylsilyl azide.
Using this procedure, they achieved yields of chiral C-2 symmetric 1,2-diamines in excess of 90%, with ees also greater than 90%. Good results were obtained from a variety of aziridines, both cyclic and acyclic. To prove the route's applicability to oseltamivir, they managed to elaborate the aziridine derived from cyclohexane to the drug product, using the stereoselective aziridine opening reaction to introduce the chirality.
An alternative, simpler synthesis of oseltamivir has been developed by EJ Corey's group at Harvard.4 The first chiral centre is introduced by a Diels-Alder reaction of trifluoromethyl acrylate and butadiene in the presence of the chiral boron catalyst derived from (S)-proline, originally invented in the Corey group. The reaction is simple to run - it gives good yields and enantioselectivities in multigram quantities, with the procedure being carried out at room temperature. In this case, the group achieved a yield of 97% and an ee in excess of 97%. As a bonus, it is a straightforward procedure to recover and recycle the chiral ligand, which dramatically improves the economics of the process.
Camptothecin is an alkaloid isolated in the 1950s from the Chinese Camptotheca accuminata tree. It has found use as an anticancer agent, but more importantly provided the starting point for other anticancer drugs, notably topotecan and irinotecan. The lactone ring of camptothecin itself cleaves very readily in vivo, and alternatives where this ring is less fragile could have great potential.
One such derivative is diflomotecan, being developed by Roche, in which the modifications include the six-membered ring lactone is replaced by a seven-membered ring lactone with the addition of an extra methylene unit. This lactone contains a chiral centre, and the group has developed two different routes, one involving an acetal and the second an amide.5
The acetal route is derived from the original racemic route based on the Reformatsky reaction, which has been improved and made chirally selective. Instead, they based it on a stoichiometric asymmetric aldol addition of an acetate to a ketone substrate which gave the desired chiral quaternary centre. The best results were achieved at a temperature of -95°C, which may be an issue ultimately in scale-up. The bicyclic target was obtained in nine steps and an overall yield of 8.9%, which while it seems low is better than the original route and has the added advantage of needing no chromatographic separations.
The second, amide, route uses 2-chloroisonicotinic acid as the starting material, and they used a secondary amide directing group to ensure the lithiation of the pyridine 3-position proceeded selectively. This is then used as the starting point for an asymmetric acetate aldol addition, and this time the overall yield for the nine steps was 11.1%, although this time a chromatographic separation was required. Again, the asymmetric aldol reaction needed to be carried out at very low temperature.
Boehringer Ingelheim has been developing the drug BILN 2061 as a potential treatment for hepatitis C infection. It is an NS3 protease inhibitor that has good oral bioavailability and antiviral effect, and it contains a 15-membered tripeptide macrocyclic ring. However, making this macrocycle requires three unnatural chiral amino acid fragments to be synthesised first.
One of these contains a cyclopentene sidechain, used to introduce a carbon chain into the ring, and they identified a route that included an enzyme catalysed kinetic resolution step.6 The group felt that the inexpensive enzyme acylase I would be good for this, as it has already been applied to the L-selective cleavage of N-acetyl amino acids. These enzymes are commercially available, derived from both porcine kidney and Aspergillus fungi, and have the advantages of being stable in aqueous solution and having high specific activity.
The racemic compound identified for separation was a 2-acetamido carboxylic acid bearing the desired , which was prepared in three steps from 1-bromo-4-chlorobutane, with each step giving yields in excess of 80%. This was then subjected to kinetic resolution, using 0.5% w/w of the acylase enzyme in water at pH between 7.5 and 8.0, in the presence of a small amount of a Co(II) enzyme cofactor. After seven hours, the solid product was collected by filtration, giving the desired amino acid fragment in yields of at least 45% and an ee in excess of 99%. Unusually, the pH remained stable throughout the process, and no sodium hydroxide needs to be added during the process to adjust the pH, as is common in this kind of enzymatic resolution. The unwanted enantiomer could be racemised by heating in water containing acetic anhydride, and after two cycles of racemisation an overall yield of 71% of the correct isomer was obtained. The company has also reported the synthesis of the full molecule, with the macrocyclic unit constructed from the chiral fragments.7
Resolution was also the key to the synthesis of the chiral core of the piperidine-based cocaine analogue CTDP 31,446, which is a dopamine and norepinephrine reuptake inhibitor. A team at Albany Molecular used a literature8 Grignard addition of 4-chlorophenylmagnesium bromide to arecholine, which gave a mixture of cis- and trans-isomers, both racemic.9 The desired cis-isomer was crystallised in 22% yield, and a further 34% was isolated from the mother liquor, along with 18% of the "wrong" trans isomer. However, this original reaction had significant problems for scale-up: it needed enormous quantities of diethyl ether as solvent, and the reaction mixture tends to form a gel as the mixture is made less dilute. Chromatography was also required to extract the product from the mother liquor.
The reaction was optimised by using methyl-t-butyl ether as a cosolvent, allowing the volume of diethyl ether required to be reduced by more than half. They also managed to maximise the quantity of the (±)-cis isomer formed by quenching the reaction rapidly, and ultimately a combined mass recovery of the desired isomer of up to 80% was achieved. This was then separated by resolution via (-)-dibenzoyl-L-tartaric acid, and the (-)-cis form was obtained in 98% ee and 36% mass recovery.
Racemisation
Lilly's antidepressant duloxetine is an inhibitor of both serotonin and noradrenaline reuptake, and as a result is also useful for treating stress incontinence as well as depression itself. Its predecessor, fluoxetine - better known as Prozac - was licensed as a racemate, but duloxetine is sold as a single enantiomer. The most industrially viable manufacturing published process for the chiral core involves a resolution, racemisation and recycling procedure, and a group at Nagase in Japan has revisited the synthesis and made improvements.10
The Lilly route involved the Mannich reaction of diethylamine and 2-acetylthiophene, which was then treated with base to give a ß-aminoketone and reduced using sodium borohydride. On work-up, the racemic building block (±)-(N,N)-dimethyl)aminoalcohol was formed. This was then separated by resolution using (S)-mandelic acid in t-butylmethoxide solution. The (S,S)-salt precipitated out, and the other diastereomer remained in solution and treated with HCl to racemise it. This could then be resolved, too, to increase the yield.
The Nagase team made several process optimisations to the synthesis of the racemic compound. The initial reaction mixture tended to "bump" as the product crystallised out abruptly, so they added the thiophene in two batches, with the reaction mixture being seeded to induce partial crystallisation before the second batch was added. Ultimately, they obtained the racemate in 90% yield. The resolution step was also improved, with toluene and methanol being substituted for t-butylmethoxide as solvent, as the original solvent poses environmental concerns.