Issue Date: August 14, 2006
Biocatalysis Helps Reach A Resolution
Pharmaceutical process chemists use resolutions, or the separation of enantiomers from racemic mixtures, more than 60% of the time to generate chiral compounds. Among resolution methods, stereoselective salt formation followed by crystallization accounts for two-thirds of the cases, chemists from three leading drug companies found in analyzing dozens of syntheses (Org. Biomol. Chem. 2006, 4, 2337). Other common approaches involve chromatography and dynamic kinetic and biocatalytic resolutions. Although resolutions are not new, recent advances in tailoring enzymes for high activity and coupling them with chemocatalytic reactions are expanding the role of biocatalysis in producing enantiopure compounds from racemic mixtures.
Enzymes can be extremely enantioselective, achieving greater than 99% enantiomeric excess (ee) in reactions. They can catalyze the formation of chiral centers from prochiral substrates (see page 15) or selectively discriminate between enantiomers in a racemic mixture. This latter property makes enzymes effective resolving agents. A drawback is that the maximum theoretical yield of the desired enantiomer in resolutions is 50%, and half may be waste material.
A solution offering a potential 100% yield is dynamic kinetic resolution (DKR). It combines enantioselective resolution and in situ racemization to recycle the unwanted enantiomer. A similar alternative is a deracemization process that, instead of separating enantiomers, turns one into the other. Although biocatalysts needn't be involved, their enantioselectivity and proclivity for mild operating conditions may be advantageous over chemical resolutions or synthesis.
"Resolutions were state-of-the-art 20 years ago because at that time nobody cared if you threw away 50% of the material," says Kurt Faber, professor of chemistry at the University of Graz and member of the Research Center for Applied Biocatalysis, in Austria. "Nowadays, the competition is so tough that you try to save every single cent in the production."
A number of multiton industrial processes still use enzymatic resolution, often with lipases that tolerate different substrates. BASF, for example, makes a range of chiral amines by acylating racemic amines with proprietary esters. Only one enantiomer is acylated to an amide, which can be readily separated from the unreacted amine. Although not a DKR process, the unreacted amine can be racemized off-line and fed back into the process to increase the final yield.
Many fine chemicals producers also employ acylases and amidases to resolve chiral amino acids on a large scale. l-Acylases, for example, can resolve acyl d,l-amino acids by producing the l-amino acids and leaving the N-acyl-l-amino acid untouched; after separation, the latter can be racemized and returned to the reaction. Using a d-acylase forms the alternate product. Likewise, DSM and others have an amidase process that works on the same principle: d,l-Amino acid amides are selectively hydrolyzed, and the remaining d-amino acid amide can be either racemized or chemically hydrolyzed.
Dowpharma is experienced with hydrolases and prepares enantiomerically pure α- and β-amino acids from racemic mixtures of the related esters via selective hydrolysis with acylases and lipases, as well as by asymmetric chemoenzymatic routes. The company makes several pharmaceutical intermediates this way at large scale, says Karen Holt-Tiffin, Dowpharma's head of biocatalysis. For example, using a lipase to hydrolyze the corresponding ethyl ester, Dowpharma prepares 99% enantiopure β-phenylalanines from racemic mixtures on a scale of hundreds of kilograms.
The company also has a technology platform of (S)- or (R)-nitrilases to make chiral carboxylic acids via either conventional or dynamic resolutions. A dynamic resolution is possible if one starts with cyanohydrins because these racemize rapidly enough in situ. Enantioselective hydrolysis by the nitrilase produces one α-hydroxycarboxylic acid enantiomer, the unreacted cyanohydrin racemizes, and the desired acid accumulates in high yield and high enantiomeric excess. BASF uses this to make (R)-mandelic acid and derivatives on a multiton scale (Angew. Chem. Int. Ed. 2004, 43, 788).
Dowpharma has developed lactamases as well for bioresolutions. Using an optimized γ-lactamase, the company produces (-)-2-azabicyclo[2.2.1]hept-5-en-3-one, an intermediate for carbocyclic nucleosides, on the multiton scale. Company researchers also have developed an immobilized microbial esterase, replacing one of animal origin, to prepare enantiopure (-)-2′,3′-dideoxy-5-fluoro-3′-thiacytidine, the active ingredient in the anti-HIV drug Emtriva, by selective hydrolysis of the butyrate ester (Org. Process Res. Dev. 2006, 10, 670).
Other established processes, such as the hydantoinase technology practiced by Kaneka, DSM, and Degussa, take advantage of spontaneous racemization. Starting with readily available racemic hydantoins, d- or l-hydantoinase enzymes yield only the d- or the l-hydantoin by forming the corresponding N-carbamoylamino acid; the remaining hydantoin racemizes at dilute alkaline pH. The free amino acid is generated by treatment with a d- or l-carbamoylase. Kaneka uses immobilized recombinant hydantoinase and carbamoylase enzymes to make several thousand metric tons per year of d-amino acids.
Instead of using base to racemize the remaining hydantoin, a hydantoin racemase will work. Degussa has succeeded in combining this racemase and either the l- or d-hydantoinase/carbamoylase pairing in a whole-cell system to make nonnatural d- or l-α-amino acids at industrial scale. "We could do it as well with three isolated enzymes, but it would be much more labor and cost intensive to produce and isolate all of them separately," says Wolfgang Wienand, head of Degussa's Service Center Biocatalysis. "Having them engineered into one cell, we only have to ferment once and end up with an economically attractive biocatalyst that accomplishes all three transformations in one working step."
Meanwhile, Kaneka has developed a one-pot deracemization process for producing nonnatural amino acids in greater than 90% yield and 99% ee. The approach integrates four enzymes. To make the l-amino acid, a d-amino acid oxidase converts the d-isomer to a ketoacid and leaves the l-enantiomer untouched. Then an l-amino acid dehydrogenase (AADH) converts the ketoacid to the l-amino acid as well. Two enzymes, a highly durable formate dehydrogenase and a catalase, are present for recycling cofactors. Because the AADH is substrate specific, the company has developed a library of enzymes to handle different aromatic or aliphatic amino acids.
Similarly, Nicholas J. Turner, chemistry professor at the University of Manchester, in England, and research program director at the new Centre of Excellence for Biocatalysis, Biotransformations & Biocatalytic Manufacturing there, has developed a practical deracemization to separate chiral amino acids or amines.
Although lipase-catalyzed resolution works well for primary amines, it can be applied to only some secondary amines and doesn't work for tertiary ones. And DKR methods often fail since conditions needed to racemize amines are inhospitable to enzymes. For these reasons, Turner's two-step, one-pot process uses an enantioselective amine oxidase and a nonselective chemical reducing agent, such as ammonia-borane. The enzyme oxidizes only the S enantiomer to the corresponding imine, which is then reduced back to the racemic amine. Repeated cycles lead to the R enantiomer accumulating in high yield and enantiomeric excess. Turner's collaborators have included GlaxoSmithKline for modifying enzymes for broader substrate specificity and enantioselectivity and Great Lakes Fine Chemicals, now part of Excelsyn, for finding practical reducing agents.
"We talked to a number of pharmaceutical companies and found that they probably had more problems making chiral amines than pretty much any other type of chiral intermediate," Turner says. "There were far fewer technologies available for chiral amines compared with chiral carboxylic acids or alcohols, so we focused there." Turner and coworkers have recently adapted the deracemization process to work with tertiary amines (J. Am. Chem. Soc. 2006, 128, 2224).
This deracemization approach, which can be applied to more than 100 different natural and nonnatural amino acids and amines, is one technology being implemented by Ingenza, an Edinburgh, Scotland-based bioprocess company. Ingenza has developed the technology from an academic bench-scale reaction to a more robust, economical, and broadly applicable process, currently undergoing scale-up, Ingenza President Ian Fotheringham says.
In contrast to inefficient reductants such as sodium borohydride, employed in 1992 by Kenji Soda at Kyoto University, in Japan, in initial work deracemizing proline and later modified by Turner and collaborators, Ingenza's current process uses even more cost-effective supported-metal catalysts, Fotheringham explains. "It's one of the key changes that has permitted our approach to be scaled up." The company also has applied directed-evolution methods to develop amine oxidases with R enantioselectivity, thus enabling the production of enantiopure (S)-amines.
"We have adapted enzymes toward a number of targets of industrial interest and are currently implementing our process at multiton scale for the production of one such unnatural amino acid," he says. Using this basic technology platform and statistical experimental design methods, Ingenza evolves and optimizes the enzyme and process conditions for particular customer substrates. Chicago-based Richmond Chemical is the exclusive commercialization partner for Ingenza's proprietary technology.
Although the technology has potential applicability to other classes of compounds, Fotheringham believes amines and amino acids offer large market opportunities due in part to the technical challenges in current chemical syntheses. "Chiral amines and unnatural amino acids are two of the largest and most significant classes of pharmaceutical intermediates, so there are more than enough targets to address," he says. "There are several proven biocatalytic and chemical routes to prepare unnatural amino acids with high optical purity, but no single method has ever dominated commercially; each has only really been applied to just one or two compounds."
Chiral alcohols are another important class of intermediates, and although it's popular to use ketoreductases to reduce carbonyls directly to make them, they can be easily prepared by lipase-catalyzed acylation. A practical DKR method combining an immobilized lipase and in situ ruthenium catalyst to racemize enantiomers was developed by Jan-Erling B??ckvall and coworkers, then at Sweden's Uppsala University, in 1997. DSM has since scaled up the process, explains Marcel Wubbolts, DSM's program director for white (industrial) biotechnology, and by "tweaking" it has lowered the enzyme and catalyst loadings to produce chiral alcohols on the ton scale in 77% yield and 99% ee.
Bäckvall's group and others-including Mahn-Joo Kim, Jaiwook Park, and coworkers at South Korea's Pohang University of Science & Technology-have applied the DKR method to a variety of functionalized alcohols. Changes in ruthenium catalyst ligands and reaction conditions have decreased DKR times by orders of magnitude to just a few hours. Also key to the progress is a catalyst that racemizes alcohols in less than 10 minutes, discovered by Bäckvall and coworkers, now in the chemistry department at Stockholm University (Angew. Chem. Int. Ed. 2004, 43, 6535).
Naturally occurring lipases are only R-selective for alcohols. To make secondary (S)-alcohols, Bäckvall's group uses its fast racemization catalyst and a commercially available protease activated and stabilized by surfactants in organic solvents. This process produces the desired alcohols in up to 97% yield and 99% ee at rates four to five times faster than that reported by Kim and Park with a different catalyst (Chem. Eur. J. 2006, 12, 225).
Bäckvall has also been expanding his DKR process to other substrate classes. His group has reported success with primary amines (J. Am. Chem. Soc. 2005, 127, 17620) and has been working more recently with secondary amines. But combining a metal-catalyzed racemization and enzymatic resolution isn't always straightforward, he admits, and processes that separately are quite rapid can even fail completely when coupled.
"For example, because the amines racemize slower, we have to work at 90 oC," Bäckvall explains. Higher temperatures mean, however, that the enzyme must be thermostable. Thermophilic enzymes, and others adapted to unusual conditions, can be found from environmental sources. Or they can be engineered, as his group has been doing in recent collaborations with other research groups. "We also are working on allenes and have a very selective mutant enzyme," he says, "and now are trying to combine it with a racemization catalyst."
An alternative is to make the DKR process entirely biocatalytic. The catch is that racemase enzymes are actually relatively rare. "Nature is extremely specific when it comes to making chiral materials, so there isn't much need for racemases," Faber explains. Racemization also has generally been viewed as an unwanted, rather than synthetically useful, reaction and thus hadn't been explored much.
Faber and coworkers have employed a mandelate racemase whose broad substrate tolerance for b,γ-unsaturated α-hydroxycarboxylic acids makes it a good candidate (Adv. Synth. Catal. 2005, 347, 951). His group and collaborators at BASF have also found a lactate racemase that works with pharmaceutically important saturated aliphatic, arylaliphatic, and aromatic α-hydroxycarboxylic acids (J. Org. Chem. 2005, 70, 4028). Using a whole-cell system under mild conditions, they have achieved what Faber calls "clean" racemization without unwanted side reactions. The same system also works with α-hydroxyketones (Adv. Syn. Catal. 2006, 348, 873).
Faber's group has also been investigating enantioconvergent transformations-that is, the conversion of both enantiomers into the same stereoisomeric product via independent routes. For this to work, the stereochemistry of one enantiomer is retained while the configuration of the other undergoes inversion. Just within the past few years, they have unearthed sulfatases that catalyze the hydrolysis of sulfate esters; whether one of these enzymes inverts or retains the configuration depends on the microbial source.
Such enzymes may be even rarer than the racemases, Faber suggests. Along with alkyl sulfatases, epoxide hydrolases and haloalkane dehalogenases can affect stereochemistry in similar controlled ways (Biochem. Soc. Trans. 2006, 34, 296). "Hopefully we will see more industrial applications within the next five years," he remarks, calling DKR, deracemization, and enantioconvergence methods the front line of biocatalysis.
- Chemical & Engineering News
- ISSN 0009-2347
- Copyright © American Chemical Society