Pure perfection

The need for cost-effective, high yielding selective reactions to create chirally pure nces continues to increase steadily. Dr Sarah Houlton looks at some recent developments.

The vast majority of new chemical entities that reach the market these days are chirally pure. Gone are the days of racemates being routinely approved, followed by a chiral switch as the single isomer form is launched a few years later. Unless there is a good reason why a single isomer is impractical, such as the spontaneous interconversion of a stereocentre either on storage or in vivo, then the regulators are unlikely to allow them onto the market.As a result, the need for effective, selective reactions to create chiral building blocks to make single isomer drugs is only going to increase. Add to this the need for these reactions to be cost effective, high yielding, and avoid the use of dangerous or difficult reagents, and it is hardly surprising that a huge amount of synthetic effort is put in by chemical companies, pharma companies and chemists in universities to create new, and better, processes.

highly efficient

The most cost effective way of creating chirally pure pharmaceutical intermediates is often to begin with something from the chiral pool - a naturally occurring chemical that is chiral, as nature is very efficient at making single isomer molecules. Starch is cheap and comes from renewable vegetable sources, and Korean company Samsung Fine Chemicals has patented a process for making chirally pure S-3-hydroxy-γ-butyrolactone, a hugely useful pharmaceutical intermediate.¹

The starch is first broken down by enzymes to an α-1,4-linked oligosaccharide, which is oxidised by hydrogen peroxide in the presence of aqueous sodium hydroxide. The resulting chirally pure 3,4-dihydroxybutanoic acid (formed as its sodium salt) is then lactonised with sulfuric acid to give the butyrolactone. The process has been run on a kilogram scale, giving yields from starch of up to 57%.

Diastereomeric salt resolution has been used to make chirally pure S-2-amino-5-methoxytetralin in a process developed by Nagase & Co in Japan for the synthesis of an intermediate for the potent dopamine D2 antagonist N-0923, which is being investigated in clinical trials as a potential treatment for Parkinson's disease. The racemic mixture is made in three steps from 5-methoxy-2-tetralone.²

Five different commercially available chiral acids were tried to carry out the resolution by making diastereomeric salts, and S-mandelic acid gave the best results. Experimental parameters that affect crystallisation and recrystallisation (such as stoichiometry), solvent and temperature, were investigated to optimise the efficiency of the resolution. The two diastereomers were separated by crystallisation, and treatment with sodium hydroxide liberated the desired S isomer in 99.7% ee. The unwanted R isomer could be racemised to create more of the S isomer by an oxidation and reduction process.

possible alternatives

Chiral catalytic reactions, however, form the basis of a huge amount of the synthetic effort towards making single enantiomer intermediates. For example, sertraline (Zoloft) is an antidepressant from Pfizer whose patents are nearing the end of their lives.

Ciba Specialties has patented a process for the synthesis of its chiral tetrahydronaphthalene core, starting from a racemic imine.³ Instead of the previously used Pd/C and Raney nickel catalysts, which give a mixture of the four possible diastereomers, Ciba used a barium promoted copper chromite catalyst. In ethanol this gave over 95% of the cis isomers, removing the need for a purification step to separate the trans isomers out.

general procedure

Chiral alcohols are important intermediates in the synthesis of pharmaceutical actives, and Dowpharma has developed a catalytic route to chiral 1-aryl-2-imidazol-1-yl ethanols using asymmetric transfer hydrogenation.⁴

Racemic alcohols with an imidazole group in the alpha position relative to the hydroxyl group are an important structural feature in topical fungicides such as miconazole and econazole. It is possible to separate some imidazoyl alcohols by the resolution of diastereo-meric salts, but asymmetric synthesis would offer a more general procedure to single isomer versions of such drugs.

The Dowpharma team began with imidazoyl acetophenone substrates. Standard bisphosphino ruthenium diamine complexes were found to be ineffective. Instead, when the substrates were treated with [(R,R)-TsDPEN]Ru(Cymene)Cl as the catalyst precursor, and formic acid as the hydrogen transfer agent, the desired chiral alcohols were obtained in ees up to 99%.

Takeda Chemical Industries has been investigating a chiral naphthoquinone derivative as a possible neurodegenerative disease agent. The key intermediate in its synthesis is an optically active acid, initially made by the enantio-selective hydrolysis of an ester of the racemic form using lipase, a reaction that worked on a large scale. Since then, an asymmetric synthesis has been invented by Takeda chemists.⁵

The team found that it could be prepared by the asymmetric reduction of hindered 1-naphthyl-1-aryl ethene derivatives, and that this could be achieved using Josiphos ligands in conjunction with rhodium catalysts in the presence of base, giving the chiral acid up to 93% ee. Numerous ligands were investigated, with Josiphos giving the best results (Scheme 1).

β-amino acids are common structural features in drug molecules, both in small molecules and peptides. As a result they are frequently used as chiral building blocks, and a team at Merck has developed a new, efficient synthesis of β-amino acid derivatives by the asymmetric hydrogenation of enamines.⁶

They claim that this route has not been successful on a large scale in the past because of the requirement for an acyl protecting group on the nitrogen to chelate with the metal in the catalyst; however, direct acylation of enamines is not trivial.

The Merck team has succeeded in developing a route to the asymmetric hydrogenation of unprotected enamines in high yield. They found that β-enamino esters and amides made from commercially available β-keto esters and amides can be hydrogenated with Josiphos type ligands and a rhodium catalyst, and that a variety of unprotected enamine esters and amides could be transformed into the corresponding β-amino acid derivatives in high yields, with good enantio-selectivity (Scheme 2).

halide surprise

A third new application of Josiphos ligands has come out of the labs of Erick Carreira at ETH in Zurich.⁷ His group has developed an enantioselective conjugate reduction of nitroalkenes which, rather than the more common (and expensive) metals such as rhodium, ruthenium or palladium, uses commercially available copper(II) fluoride. The group had previously developed a catalyst system prepared from copper t-butoxide with Josiphos, but the copper t-butoxide is sensitive to both oxygen and water, making a cheap, readily available, more stable metal source preferable.

They were surprised to find that the fluoride salt worked, because it had previously been found that halides inhibit the enantioselective reduction of nitroalkenes. They also found that by using using the additive nitromethane as an activator, more electron rich nitroalkene substrates could be reduced enantioselectively. CuF₂ is easy to handle, and chiral nitroalkanes - very useful building blocks in chiral synthesis - were obtained in yields up to 88%, and with selectivity of up to 98:2.

Salens are commonly used as chiral ligands. A patent from Korean company RS Tech uses cobalt salen catalysts to make chiral epoxides from chlorohydrins, and chiral 1,2-diols by the hydrolysis of epoxides.⁸

The t-butyl salen is used with PF₆^- or BF₄^- as the counterion to ensure the catalyst retains its activity. The company managed to make S-epichlorohydrin from the racemic form by treating it with the cobalt salen catalyst at room temperature in water, followed by distillation. The product was obtained in over 99% ee. The recovered catalyst can be reused, and enantiomerically pure 1,2-epoxyhexane and styrene oxide can be made in a similar way. The catalysts can also be used to effect the stereospecific hydrolysis of racemic epoxides to the corresponding single isomer diols.

novel process

Avecia has recently patented a novel process for the synthesis of primary and secondary amines from imines and iminium salts.⁹ In the process, a ketone is first converted into an oxime and, while both E and Z isomers are formed, the E form can easily be separated by crystallisation. The oxime is then treated with ammonia in a mixture of diethyl ether and dichloromethane, and the resulting solution treated with ClPPh₂, giving a phosphoryl imine. This is then reduced using [Rh(Me₅Cp)Cl₂]₂ as a catalyst, with a chiral monotosyldiamine to induce stereochemistry and a formic acid/ triethylamine mixture as the hydride transfer reagent, giving a clean single enantiomer product.

Numerous other catalyst systems are also described in the patent, including rhodium, ruthenium and iridium complexes, as well as several different chiral monotosyldiamines, and the possibility of using an alcohol for the hydride transfer.

Pfizer has published the chiral synthesis of a range of γ-hydroxy-δ-amino lactones, which are key precursors to hydroxyethylene dipeptide isosteres, a class that has potential of inhibitors of both renin and HIV.¹⁰ While it is possible to make these from commercially available chiral amino acids, the chemistry invariably involves amino aldehyde intermediates that are notably prone to racemisation.

The Pfizer chemists started with a thioester prepared from commercially available Boc-L-3-(fluorophenyl)alanine. However, when this was alkylated with an organozinc homoenolate in the presence of PdCl₂(PPh₃)₂, the resulting aminoketone was completely racemised. They thought the loss of chirality was a result of quantities of ethoxide being produced in the reaction. When they used phthalic anhydride as a thiolate scavenger, the team found that the chiral integrity of the product was preserved.

Another advance from Pfizer is a practical synthesis of a 3-amino-3,4-dihydro-1H-quinolin-2-one building block,¹¹ a common structural feature in a number of drugs, such as antibacterials, ACE inhibitors, tachykinin antagonists and D2 receptor agonists. However, in most cases it is in racemic form, and when the chiral intermediate has been used it has usually been created by resolution, or a multistep synthesis starting with phenylalanine.

The Pfizer scientists developed a two step process to make them, involving an asymmetric alkylation followed by a one pot reduction and cyclisation. They started with the t-butyl ester of N-(diphenylmethylene)glycine, and carried out an asymmetric phase transfer alkylation with 2-nitrobenzyl bromide, using a chiral tetra-alkyl ammonium salt. This selectively shields one face of the enolate, with the alkylation then taking place from the more accessible side. Either enantiomer of the ammonium salt can be used, giving access to both enantiomers of the product.

Catalytic hydrogenation then reduced the nitro group, deprotected the amine and carboxyl groups, and effected the cyclisation, all in one pot. The S-isomer final product was obtained in 65% yield and 89% ee, and the R isomer in 88.5% ee (Scheme 3).

The Julia Colonna epoxidation, catalysed by poly-α-amino acids, is an efficient reaction for functionalising α,β-unsaturated ketones in an enantioselective way, and a broad range of substituted enones can be epoxidised with high yields and high enantio-selectivities. Work at Bayer has developed a new set of conditions for the reaction. The triphasic conditions and the company's own polyamino acid catalyst were used to give the complete conversion of a chalcone into an epoxyketone. They added small quantities of tetrabutylammonium bromide as a phase transfer catalyst, which overcame the previously observed low conversion rates.

The reaction has been scaled up to 100g and uses 10% w/w of the poly-L-leucine catalyst, giving the intermediate the team wanted in 75% yield, with an ee of 95.5% (Scheme 4).

Discodermolide is a potent anti-tumour agent that, like Taxol, is a microtubule stabilising agent. It is currently the most potent of these agents known; hence it is undergoing clinical investigations as a potential cancer treatment.

There is, however, a big problem: the polyketide natural product is isolated in tiny amounts from extracts of the marine sponge Discodermia dissoluta, and has to be harvested using manned submersibles. So, currently, total synthesis is the only practical route to obtaining supplies of discodermolide for the trials work, and this is likely to remain the case until some biochemical route involving getting bacteria to carry out the chemistry is established.

improved synthesis

Chemists at Novartis have been working on an improved synthesis using the best features of the syntheses of Ian Paterson's group at the University of Cambridge, and Amos B. Smith's group at the University of Pennsylvania.¹² The Novartis scientists looked at improving the aldol coupling that forms the C6-C7 bond, and at the same time creates the stereocentre at C7.

Paterson's chiral boron route has several problems, including decomposition, solvent changes and a large excess of enolate. Improvements were found to give the product in a higher yield, raising it to greater than 90%.

The modified process is a great improvement, and should be useful in the creation of supplies of discodermolide until such time as a fermentation route is available.