A golden era for catalysts

Published: 20-Jul-2010

Nature’s keys – catalysts – can unlock doors to better product yields and faster reactions. Dr Sarah Houlton reviews the latest research using carbon nanofibres, enzymes and even gold for producing new catalysts

Nature’s keys – catalysts – can unlock doors to better product yields and faster reactions. Dr Sarah Houlton reviews the latest research using carbon nanofibres, enzymes and even gold for producing new catalysts

Catalysts are essential tools in the arsenal of the process chemist, as they can allow reactions to be carried out more quickly and in milder conditions, and often give better yields, atom efficiencies and selectivity than non-catalytic reactions. Modern catalysts enable many molecules to be made that would otherwise be difficult to obtain, if not impossible.

The syntheses of pharmaceutical intermediates and APIs often rely on catalytic reactions, and catalysis continues to provide a productive field of research for chemists. Better, cheaper, more effective catalysts – whether chemical or biological – will always be needed.

Some metals are particularly widely used in synthetic catalysts, and it is common to see metals such as palladium (Pd), rhodium (Rh) and iridium (Ir) featuring in the manufacturing routes for chiral pharma intermediates and ingredients.

A good recent example is a process published by scientists at Merck and Solvias for the synthesis of DPP-IV antidiabetes drug sitagliptin (Januvia).1 This molecule has one stereocentre, which the chemists introduced right at the end of the synthesis using an asymmetric hydrogenation with a chiral Rh catalyst.

The new route, which includes an unusual chiral hydrogenation of a free enamine, provided a significant improvement over the initial process, for which the drawbacks included a Mitsonobu coupling, which is extremely atom-inefficient, and the chirality had to be introduced at a much earlier step via the asymmetric hydrogenation of a β-keto ester to give a β-amino acid moiety (Scheme 1).

Novel ways of using these catalysts are also being developed. It is a common tactic to use precious metal catalysts in a supported form, from the traditional Pd/C hydrogenation catalyst that has been in use for decades, to modern methods such as Ley’s encapsulation techniques.

A different tactic is being employed by Yukihiro Motoyama at Kyushu University in Japan – Rh particles are supported on carbon nanofibres.2 Like Pd, Rh can be supported on activated carbon, and it is more efficient than Pd as a heterogeneous catalyst in the hydrogenation of arenes. However, the catalyst is frequently subject to rapid deactivation as it is very easy for the metal particles to sinter, leading to their leaching from the support. This makes it difficult to recycle the catalyst, adding to the expense. So they looked for an alternative support, and carbon nanofibres did the trick.

There are three types of carbon nanofibres, where the graphite layers are arranged perpendicular to each other giving a platelet form, in parallel to give a tubular structure, or stacked obliquely to give herringbone fibres. The group has worked extensively on these fibres and developed methods to ‘immobilise’ transition metals on them through the decomposition of organometallic precursors, giving Pd, ruthenium and platinum assemblies that can be used as catalysts. Importantly, no leaching or sintering is observed, allowing the catalysts to be reused without loss of activity.

The researchers have now discovered that Rh particles can be immobilised on all three forms of nanofibre, and this overcomes the sintering issues that normally dog Rh/C catalysis. The best results in a model system where phenol was being hydrogenated were obtained using the tubular form of the nanofibres, which was easy to prepare from Rh4(CO)12 and the nanofibres.

The synthetic potential of this form of Rh catalyst was illustrated by the hydrogenation of a range of aromatic compounds bearing a glycidyl moiety. These substrates frequently undergo an epoxide ring opening side-reaction when the aromatic group is hydrogenated, but using this catalyst prevents this unwanted reaction from taking place. In addition, the catalyst could be reused several times without loss of its activity.

precious metals

More unfamiliar precious metals are also being investigated for their catalytic potential. Recently, gold has been become more popular as a catalyst in academic circles, and it is now beginning to feature in industrial syntheses, too. For example, a group from GlaxoSmithKline (GSK) in Stevenage, UK, used a gold catalyst in the final manufacturing route for the synthesis of a 5-HT4 receptor agonist (Scheme 2).3

Scheme 2: Synthesis of a key benzopyran intermediate

Scheme 2: Synthesis of a key benzopyran intermediate

The reaction – a cycloisomerisation of an aryl–propargyl ether to give a key benzopyran intermediate – was being used instead of a traditional Claisen rearrangement that involves high temperatures. The team screened several chlorinated and non-chlorinated precursors against a variety of gold(I) complexes, as well as more traditional platinum(II) and platinum(IV) complexes. The chlorinated starting materials performed pretty poorly, which the group attributed to solubility issues, and possibly some additional stereoelectronic factors coming into play.

They identified triphenylphosphine gold(I) as the best catalyst, with the reaction being carried out in non-polar solvents such as 1,4-dioxane and toluene, at a reaction temperature of 80°C– 100°C. An additional advantage of the gold catalyst over the platinum alternatives was that it suppressed the depropargylation side-reaction that degrades the propargyl ether substrate to the parent phenol. They also tried the reaction using silver(I) catalysts, and found that the product profile resembled that of the high temperature Claisen reaction, and thus this was less effective than the gold mediated process.

Another key trend in metal catalysis is a move towards more environmentally benign – and cheaper – metals such as iron and copper, and work using many of these catalysts has been published in the past few years. For example, Robert Morris at the University of Toronto, Canada, has developed an iron-based catalyst for the asymmetric transfer hydrogenation of ketones – a reaction that typically would involve a platinum group metal catalyst.4

The University reports that it is almost as active at room temperature for acetophenone transfer hydrogenation as the most active ruthenium-based catalysts, with turnover frequencies of the order of 900 an hour. It even gives good conversion with bulky substrates, and while enantio-selectivity for aliphatic ketones is moderate, its chemo-selectivity was good.

Some simple-looking intermediates can actually be quite challenging to make, particularly if they are chiral. Enzymes can prove the saving grace when trying to synthesise them. An excellent example comes from a group of chemists and biochemical engineers in the BiCE (Bioconversion – Chemistry – Engineering interface) programme at University College London (UCL), that has developed an enzymatic route for the synthesis of chiral amino alcohols.5 These intermediates can be used to introduce a chiral motif into a variety of different drugs, with the motif being present in various antibiotics, antiviral glycosidase inhibitors and sphingolipids.

They report that chemical routes to these compounds tend to have many steps, and use solvents and catalysts that are less than environmentally friendly. Existing routes range from those that start from chiral pool compounds such as glucose or serine, or use Sharpless chemistry to introduce the chirality followed by azide substitution and several further steps to add all the required functionality.

Enzymes can make for quicker, cleaner synthetic routes, and the UCL group began the example synthesis of (2S,3S)-2-aminopentane-1,3-diol using an engineered variant of Escherichia coli transketolase (D469T) to introduce the chirality into the corresponding ketone. This was made from the achiral starting materials propanal and hydroxypyruvate.

Model of the Transketolase enzyme showing detail of the active site. Picture courtesy of the UCL BiCE programme

Model of the Transketolase enzyme showing detail of the active site. Picture courtesy of the UCL BiCE programme

Transketolase enzymes can catalyse enantioselective carbon–carbon bond formation in an industrial setting, using β-hydroxypyruvate as a ketol donor. However, yields when using non-hydroxylated aliphatic aldehydes, such as propanal, are typically low. They used active site targeted saturation mutagenesis to identify a number of enzyme mutants with better activities, and the mutant D469T gave a near five-fold increase in specific activity with propanal.

A second enzyme was used to introduce the nitrogen functionality and give the desired chiral amino alcohol. Ω-transaminases are able to aminate a variety of different aldehydes and ketones, and using bioinformatics techniques they found the enzyme from Chromobacterium violaceum DSM30191 (CV2025) could be used to produce the desired chiral aliphatic 2-amino-1,2-diol from the adduct formed by the transketolase enzyme using isopropylamine as the amine donor. Solvent incompatibility meant that the intermediate adduct from the first reaction had to be isolated before the second reaction rather than sequentially in the single pot. But this two-step enzymatic process still represents a significant improvement on the longer chemical synthetic routes (Scheme 3).

Scheme 3: UCL synthesis of (2S, 3S)-2-aminopentane-1,3-diol using enzymes

Scheme 3: UCL synthesis of (2S, 3S)-2-aminopentane-1,3-diol using enzymes

A different transaminase has been used by Nick Turner’s group at the University of Manchester to catalyse the synthesis of enantiomerically pure chiral amines at high concentration.6 Typically, biocatalysis has been used to produce amines by a kinetic resolution of racemic amides and amines using lipases, and not de novo synthesis of a chiral nitrogen.

Transaminases would appear to be the obvious choice for such a reaction, but the equilibrium constant often favours the ketone starting material over the desired amine, and the products can inhibit the enzyme’s activity. Overcoming this typically involves running the reaction at a concentration far lower than is required for a successful industrial process.

Turner’s group has come up with two approaches that overcome these issues. In the first, an Amberlite ion exchange resin is used to remove the product in situ, thus enabling the reaction to be carried out at an industrially viable concentration of 50g/litre. They successfully applied this technique to the transamination of acetophenone, giving either enantiomer of methylbenzylamine in a yield in excess of 90%, and ee above 99%. The second technique involves the amine that is first formed being converted by a spontaneous cyclisation to a product that does not inhibit the biocatalytic reaction. An example reaction, the transamination of ethyl 4-acetylbutyrate, gave either enantiomer of 6-methyl-2-piperidone in similar yield and ee.

B-amino acids are increasingly popular motifs in pharma actives, and Kevin Walker’s group at Michigan State University is using enzymes to convert racemic α-arylalanines into single enantiomer β-arylalanines.7 They found that the Taxus phenylalanine aminomutase enzyme can effect the chemical conversion from α- to β-amino acids, but the ratio of R to S enantiomer ranged between 0.4 and 1.8. By adding a second enzyme to the system, an alanine racemase isolated from Pseudomonas putida, the yield of enantiomerically pure β-arylalanines was increased.

Using a combination of enzymes in this way could have a great deal of potential in the synthesis of β-arylalanine building blocks on a large scale, and the scientists speculate that selectively removing the desired R-enantiomer from the reaction mixture could also increase yield and enantioselectivity.

Re-engineering a process to use enzymes can often lead to far more effective and efficient synthetic routes. An example of this from US enzyme development specialist Codexis, in collaboration with India’s Arch PharmLabs, provides an alternative biocatalytic route to a key intermediate used in the synthesis of Merck’s leukotriene receptor antagonist montelukast (Singulair).8

In the original synthesis, the molecule’s single chiral centre was introduced using a chiral boron–camphor reagent such as (-)-DIP-Cl to reduce a bulky and highly functionalised ketone. They surmised from a survey of the patent literature that generics manufacturers were using a similar tactic in their montelukast syntheses. This reagent is both corrosive and moisture sensitive, requires a long work-up and gives very poor atom economy, but because the substrate was so sensitive, none of the other common chemical enantioselective ketone reduction techniques could be used.

This would appear to be an ideal candidate for an enzyme-catalysed process, because of the lack of chemical alternatives to this far from ideal reagent. Merck had already identified a couple of enzymes that had some activity on the substrate, but these had extremely low volumetric productivity, and back then techniques to improve the enzymes were limited. This is not the case now, and modern enzyme evolution techniques were applied to the problem by the Codexis scientists (Scheme 4).

Scheme 4: Synthesis of montelukast

Scheme 4: Synthesis of montelukast

They identified the biggest challenge in enzymatic catalyst development as the highly hydrophobic nature of the substrate – it is almost insoluble in water, which is the solvent typically used in enzyme-catalysed reactions. However, enzymes can be active in hydrophobic organic solvents, but they said it was at that point unclear whether enzymes could be created that would have sufficient activity and stability when used in the high concentrations of water-miscible organic solvents to get sufficient substrate into solution for a successful industrial process.

Several properties would be needed in an enzyme for this process. It would have to be able to use IPA for cofactor regeneration, and accept both the bulky substrate and the much smaller IPA cosubstrate. It would also have to tolerate as much as 70% polar, water miscible organic solvents, ideally at elevated temperature, work efficiently at a low substrate concentration, not be inhibited or deactivated by the accumulation of the acetone coproduct and, importantly, give enantioselectivity in excess of 99% ee.

An initial screen of commercially available enzymes produced none that had any measurable activity on the ketone, but some of the enzymes in its panel collection of nicotinamide adenine dinucleotide phosphate (NADP) dependent ketoreductases had some activity and, importantly, gave high ees of the correct enantiomer. The activity was far too low for a manufacturing process, and all these enzymes were unstable in organic solvents and inhibited by the acetone, however, and so an enzyme evolution programme was undertaken to find an enzyme that was at least 1000 times more active.

They did not expect enantioselectivity to be a problem, as they presumed the substrate would only enter the catalytic site in one conformation because of its bulk; this supposition turned out to be correct.

After three rounds of evolution, catalytic activity was increased to a product:catalyst mass ratio of 8:1, compared with the ratio of 1:50 for the initial enzyme ‘hit’. This was attributed not only to the enzyme itself being more active, but also an improvement in its thermal stability and the fact that it was not inhibited by acetone.

Swapping the cosolvent from 10% tetra-hydrofuran to 10% toluene also, surprisingly, increased the activity by a further factor of 2.5 without any loss in stability. The final process was successively scaled up from a 20ml vial to a 2-litre flask to a 20-litre kiloreactor, and the technology transferred to Arch’s facility in India. After an initial scale-up to 125kg batches, it is now running at a 230kg scale.

references

1. K.B. Hansen et al. J. Am. Chem. Soc. 2009, 131, 8798

2. Y. Motoyama et al. Org. Lett. 2009, 11, 5042

3. G. Rassias et al. Org. Proc. Res. Dev. 2010, 14, 92

4. A. Mikhailine et al. J. Am. Chem. Soc. 2009, 131, 1394

5. M.E.B. Smith et al. Org. Proc. Res. Dev. 2010, 14, 99

6. M.D. Truppo et al. Org. Proc. Res. Dev. 2010, 14, 234

7. B.M. Cox et al. J. Org. Chem. 2009, 74, 6953

8. J. Liang et al. Org. Proc. Res. Dev. 2010, 14, 193

You may also like