Chirals continue to spiral

The dominance of chirals in pharma pipelines is still growing, as are the variety of technologies for making chirally pure chemicals. Dr Sarah Houlton reports on latest developments

The number of single enantiomer molecules in pharma companies' pipelines has risen dramatically in the past few years, to the extent that these days around 80% of all developmental drugs are now chirally pure.It is estimated that 15% of all advanced intermediates, building blocks and auxiliaries are chiral, too. Optically active molecules have become an extremely important part of the market for speciality chemicals manufacturers, and as a result a variety of new methods of making chirally pure chemicals on an industrial scale have been developed in recent years.

efficient catalysts

There are several strategies that can be applied to chiral synthesis, but some are much more industrially useful than others. The cheapest way can often be to take advantage of the pool of common chiral molecules as a starting point, such as sugars or amino acids.

Catalytic procedures - whether using a chemical catalyst or an enzyme in a biocatalysis route - are frequently the most efficient, if a good, selective catalyst is available. Auxiliaries can be used to direct the chirality of the reaction, although this can involve unacceptable loss of mass. And sometimes the most efficient route remains making a racemic mixture, and finding a way of separating the enantiomers.

If a suitable commercially available chiral building block can be pinpointed, this is often the fastest and most cost-effective strategy for introducing chirality into an intermediate or active. As a result, many suppliers offer molecules off the shelf, in the hope that they will be picked up as a potential building block by a pharma company.

building blocks

For example, recent additions to the product line of US company BioCatalytics, of Pasadena, CA, include the non-natural amino acids L-cyclopentylglycine and D-allo-leucine, both of which are made using the company's enzyme-based technologies. Both are available as free amino acids and a range of N-protected derivatives. The company has also recently launched a group of 3-hydroxy-4-aminocarboxylic acids, called statines, which are useful building blocks for several drug candidates in fields such as Alzheimer's disease, cancer and anti-infectives.

A huge amount of synthetic effort has been put into the development of metal catalysts that will direct the stereochemistry in chemical reactions, particularly asymmetric hydrogenations. Dowpharma, for example, has access to more than 250 catalysts based on rhodium, ruthenium and iridium, as Ian Lennon explained at the Modern Synthetic Methods conference held earlier this year by Scientific Update alongside the Biofine exhibition in Berlin. 'We have many catalysts, whether developed in house or licensed in, because no one catalyst can do every reaction,' he said. 'You need a broad range to handle every real world substrate that comes along.'

scalable process

A particularly useful catalyst is [(R,R)-Me-DuPhos Rh (COD)]BF₄, which has been applied to the synthesis of three separate actives: candoxatril from Pfizer, tipranavir from Pharmacia & Upjohn, and pregabalin from Warner Lambert.¹ 'These used to be separate companies, but of course now they're all Pfizer!' Lennon said. The catalyst has been made in multiple kilogramme quantities, and the process is scalable. 'It's a red compound, with uniform crystal size, and is reasonably air stable. We've made it on a 2kg scale, and believe we could make 10kg batches, if we wanted to risk that much expensive rhodium and ligand in one reactor.'

Pinning down the best combination of metal and ligand for a particular transformation gets ever more complicated as the number of potential ligands spirals upwards. The majority of these ligands are bidentate, as mono-dentate ones were initially thought less likely to give high ees because of their inability to form chelating complexes with the metal. However, more recently it has been shown that they can indeed be effective ligands in rhodium catalysed asymmetric hydrogenations.

monodentate ligands

The first successful monodentate ligands, such as MonoPhos, now commercially available from Strem Chemicals, were based on a chiral diol backbone, with the stereochemistry of this directing the chirality of the product.

A completely different class of monodentate ligands for rhodium catalysed asymmetric hydrogenations has now been developed by Ben Feringa and his team at the University of Groningen in the Netherlands, as he described at Johnson Matthey's Forum on Asymmetric Synthesis and Technologies (FAST) conference earlier this year. The cyclic phosphoramidite ligands are based on an achiral catechol backbone, and the chirality is directed solely by the amine moiety. The ligand shown proved comparable to MonoPhos for the hydrogenation of amino acid precursors, and better for enamides, where ees of up to 99% were observed.²

While, historically, chemists have tended to gravitate towards using metals as catalysts, nature uses enzymes to give incredibly selective and effective chemical transformations. Recent years have seen chemists becoming increasingly attracted towards enzyme catalysts, and have developed a variety of strategies that make them easier to handle as reagents, such as immobilisation and cross linking. Nature developed enzymes to be enantioselective and regioselective, not to mention chemo-selective, as they frequently carry out transformations at a single centre in a complex molecule, without affecting the rest of the molecule, which one might normally expect to be sensitive to a traditional chemical reagent. Reaction conditions are usually mild, and one enzyme can often replace several chemical steps.

Enzymes can also be used as catalysts for resolutions, as a recent example from Organon shows (figure 1).³ The team needed access to large quantities of R-seudenol for a project and, while the racemic form was easily acquired, producing the single enantiomer was much less simple, especially as the team wanted to avoid the use of chromatography. They looked at enzymatic hydrolysis, enzymatic transesterification and enzymatic acylation, followed by hydrolysis of the resulting ester. The first of these was too slow to be practical, and the second needed too high a catalyst loading. The third strategy proved much more successful.

optimum solution

After trying several different enzymes and esters, they established that the best solution was immobilised Candida antarctica lipase, which is commercially available under the name Novozym 435 from Novozymes. This was used in conjunction with vinyl butyrate or vinyl laurate as the acyl donor. The reaction has been scaled up to 10kg of the laurate, giving around 2.5kg of the desired enantiomer of seudenol, with 96% ee.

Dowpharma has also been investigating enzymes as catalysts for dynamic kinetic resolutions. A good example is the synthesis of travoprost (Travatan), Alcon Laboratories' treatment for glaucoma (Figure 4). As Dowpharma's Ian Lennon explained to the Berlin conference, a bioresolution was an essential step in the synthesis of the active, which the company developed for Alcon as part of a five-year collaboration.⁴

A particularly important biocatalysis route was developed by Dowpharma to manufacture (-)-Lactam. When the project was started, no enzymes were available to carry out the resolution. The product was originally made by resolution using Pseudomonas putida which, while it was developed into a workable process to satisfy the tonne scale demand, needed a high biocatalyst loading, and the lactamase enzyme itself was too unstable to isolate. It was also a messy process, with a complex isolation procedure required to separate the product. 'The enzyme has now been cloned, and is a pink liquid that is simply added to the reactor,' Lennon explained. It also has an extremely good catalyst loading: 500g/l in water.⁵

Dowpharma has also been collaborating with Diversa on a range of nitrilase enzymes. These can be used for both dynamic kinetic resolutions and desymmetrisation reactions.⁶

novo synthesis

Although many amino acids are readily available, others, particularly unnatural ones, are not. Some are amenable to synthesis from simpler, available amino acids, but in many cases de novo synthesis is required, which involves introducing chirality into the molecule.

Degussa has been working in this area, as the company's Kai Rossen explained at the FAST conference. 'While methionine is made on a huge scale for chicken feed, and lysine for pig feed, there are much more complex targets than these,' he said. 'And a good approach to synthesis could actually be the old approach! Ritter, Strecker and Claisen reactions are all very old, but still have huge potential.' They would be particularly powerful if the reactions could be combined with enzymes to create the chirality.

This method has been applied to the synthesis of 6,6-dimethyllysine (figure 2). 'We came up with a short, high yielding and efficient synthesis using readily available bulk chemicals as reagents,' he said. 'HCN is cheaper than the more traditional MeCN, and also results in a formyl group that is easier to take off. The only real problem was with the kinetic resolution step, but there is an enzyme that can be used to racemise it.'

biocatalysis strategy

A biocatalysis strategy has been used by a team of chemists at Nagase & Co in Japan to make a chiral intermediate.⁷

Initially developed for use in liquid crystals, the molecules have potential as pharma intermediates. 4-Propylcyclohexanone was treated with Galactomyces geotrichum JCM 6359, which was identified through a screen of 288 bacteria, 384 yeasts and 84 fungi. The hydride transfer reaction led to 99% of the product formed being the cis isomer. Although the reaction did not go to completion, the unreacted starting material was simple to remove, and the cis hydroxide was formed preferentially. The alcohol could be further elaborated to make more complex intermediates.

This route used a screening process to pinpoint a suitable biocatalyst. However, it is possible to design enzymes to perform specific reactions, as Manfred Reetz of the Max Planck Institute fuer Kohlenforschung in Muelheim, Germany, explained in Berlin. Reetz is working on the directed evolution of enantioselective enzymes. 'What nature does is amazing. Can we replicate it in the lab?' he asks.

A natural gene makes a wild type enzyme, but to alter this enzyme, the gene has to be mutated. Reetz and his group did this by using the error prone polymerase chain reaction (figure 3).

'Site specific mutagenesis is not trivial,' Reetz adds. 'Which amino acid around the active site of the enzyme needs to be changed, and what to? So we create a library of mutated mutant genes, and express them to give thousands of altered enzymes. Now the major problem is that you have maybe 2000 catalysts, and these need evaluating.'

Using classical GC or HPLC, this is slow, so Reetz and his team have worked on high throughput assays to pinpoint the enzymes whose activity is better than wild type. The cycle is then repeated on the relevant mutated genes, and the resulting enzymes re-screened.

'This simulates evolutionary pressure,' he explains. 'Without knowing anything about structure, we can optimise the enzymes. But it is rational, because it relies on the evolutionary pressure.' The group is particularly interested in transformations that are difficult - or even impossible - to carry out by chemical catalysis.

These include the hydroxylation of alkanes, the epoxidation of difficult olefins, and Baeyer-Villiger reactions. 'Partial oxidation is a real challenge,' he said. His group has been looking at cyclohexanone mono-oxygenase as a stereoselective enzyme for Baeyer-Villiger reactions.

The wild type enzyme gave a ratio of 55:45 R:S. After one round of error prone PCR, mutants with ratios of 77:23 and 11:89 were identified. The former has been subjected to a second round of error prone polymerase chain reactions, and this increased the selectivity to 95:5. The team had yet to carry out a second round on the latter.

alternative substrates

They have also tried their enzymes on alternative substrates, such as p-methoxycyclohexanone, and the existing mutants are already very good for this. Other substrates have also been successful, including methyl, ethyl, chloro, bromo and iodo, giving ees of 95-99%. 'There is no synthetic counterpart to this reaction yet,' Reetz explained. 'It works well in thioether oxidation, too, despite the absence of structural homology.'

As the importance of enantiomerically pure molecules rises, so it is inevitable that the development of strategies and syntheses that can be applied on an industrial scale will accelerate too. Processes will become faster, more efficient and cleaner, with higher yields and greater ees possible in the products.

Enzymes are now considered routine synthetic reagents, whereas even a decade ago this was not the case. Further developments are inevitable, and the use of ever more complex chemical reactions will become industrially viable chiral processes.