Asymmetric Catalysis on Industrial Scale: Challenges, Approaches and Solutions

This second edition of the pioneering work on this hot topic captures the major trends and latest achievements in the art of asymmetric catalysis on an industrial scale. A number of completely new real-life case studies written by the world leaders in their respective areas provide a compact and qualified insight into this developing field. The resulting ready reference and handbook collates first-hand and valuable information within a context where it can be easily found.
The high-quality contributions illustrate the relevant environments and situations, such as time pressure, how the catalytic step fits into the overall synthesis, or competition with other synthetic approaches, as well as the typical problems encountered in the various phases, including finding/developing the catalyst and optimization of the process or choice of equipment. Both successful and unsuccessful approaches to solve these problems are described.

• Language: English
• Publisher: Wiley
• Release date: Aug 4, 2011
• ISBN: 9783527642168

Book preview

Part I

New processes for Existing Active Compounds (APIs)

1

Some Recent Examples in Developing Biocatalytic Pharmaceutical Processes

Junhua Tao, J. Liu, and Z. Chen

1.1 Introduction

A confluence of factors is driving biocatalysis into a premier platform for the production of pharmaceuticals. First, the technology itself is more practical than ever for commercialization as a result of easy access of biocatalyst tool boxes from the GenBank, efficient expression systems for their production, and robust protein engineering techniques to improve their specificity, selectivity, and stability. Second, to improve the therapeutic index and absorption, desorption, metabolism, excretion, and toxicity (ADMET) profile, new chemical entities (NCEs) as pharmaceutical ingredients are structurally increasingly more complex, which conversely demand more selective transformations for bond connection and disconnection, manipulation of functional groups, and stereoselectivity. Third, catalytic process technology is posed to be the most crucial component in commercializing drug substances and even drug products as drug innovators or branded pharmaceutical companies are entering the generic business by launching generic versions of branded drugs. The premium paid to ‘the first mover’ by a generic company will be significantly decreased. Not only is biocatalysis intrinsically process efficient under the principles of green chemistry, it also provides a stronghold to generate novel routes with freedom to operate (FTO) and/or proprietary intellectual property (IP). This chapter focuses on the development of three chemoenzymatic routes to illustrate the dynamics of the field and the importance of strategic integration of chemical and biological transformation to shorten synthetic sequences, reduce energy input, and enhance process safety.

1.2 Levetiracetam (Keppra®)

Levetiracetam, (S)-α-ethyl-2-oxo-1-pyrrolidinacetamide, is the active pharmaceutical ingredient (API) of Keppra for the treatment of epilepsy, with sales of over US$1.5 billion in 2008. Most existing processes require either chromatographic separation or chemical resolution using stoichiometric amounts of chiral acids or bases, and therefore excess amounts of chemical or solvents, resulting in high process mass intensity (PMI), whereas others use asymmetric hydrogenation or start from chiral pools (Scheme 1.1) [1, 2]. In addition, most reported syntheses employ a hazardous alkyl halide to install the pyrrolidinone ring in the final step. There is a need for efficient, cost-effective, and safer routes for the large-scale production of levetiracetam.

Scheme 1.1 Synthesis of levetiracetam by chemical resolution or asymmetric hydrogenation.

images/c01_image001.jpg

Recently, a biocatalytic process was disclosed in which the strategic step involves the kinetic resolution of a racemic 2-pyrrolidinonylnitrile catalyzed by nitrile hydratases (Scheme 1.2), which was prepared from 2-pyrrolidinone by N-alkylation with racemic 2-chloro-n-butanonitrile [3]. It is difficult to execute such an approach using chemical transformations alone.

Scheme 1.2 Chemoenzymatic synthesis of levetiracetam by nitrile hydratases.

images/c01_image002.jpg

Nitrile hydratases are metalloenzymes that contain either iron or cobalt in their catalytic centers, and consist of α and β subunits with the active site being located at the interface of the two subunits. There is no homology between the two subunits. However, each subunit is highly homologous in amino acid sequences [4]. In this work, nearly 30 nitrile hydratases were discovered from the GenBank. Initial screening showed that all of them have poor selectivity though many with modest to good reactivity. The best nitrile hydratase, NH33 from Bradyrhizobium japonicum, has a poor E-value of less than 5.0 or approximately 60% enantiomeric excess (ee) at a conversion of 20%.

Subsequently, NH33 was engineered to improve its enantioselectivity through a combination of structure-based rational design and saturation mutagenesis [5, 6]. The three-dimensional structures of nitrile hydratases revealed almost superimposable metal coordination sites. Of the several residuals participating in recognition of substrates, several conserved aromatic residuals in the β-subunit form a hydrophobic pocket, which is thought to accommodate the side-chain of a nitrile substrate [4]. The three-dimensional structure of NH33 was constructed by homology modeling from the known crystal structure of a nitrile hydratase from Bacillus smithii, SC-J05-1 [7]. Computer modeling of the interaction between NH33 and the racemic 2-(2-oxopyrrolidin-1-yl)butanonitrile was performed to identify amino acid residues making up the substrate binding pocket and/or that reside within approximately 10-15 Å from the substrate binding pocket and catalytic center. These residues were then replaced by 19 other amino acids using saturation mutagenesis to screen up mutants with enhanced enantioselectivity. The results show that all positive mutation occurred in the β-unit and most mutation in the α-unit resulted in either deactivation or complete loss of enzymatic activity. The key mutation is βArg38Cys and additional scanning of double and triple mutation based on βArg38Cys led to nearly a ninefold increase in enantioselectivity with the E-value being improved to over 45 from less than 5 for the wild type (Table 1.1). It should be noted that although additional mutations around Ala42, Ala43, Leu76, and Val113 led to synergistic effects in improving enantioselectivity from the βArg38Cys template (Seq. ID 3-9, Table 1.1), further combinations of these upper mutation actually led to diminished E-values (Seq. ID 10-14, Table 1.1).

Table 1.1 Exemplary results from active site scanning of α-subunit and site saturation mutagenesis.

Medium engineering was then carried out for additional process optimization. For example, the optimal pH was found to be 6.5, and the enantioselectivity is higher in the Tris buffer than either acetate or phosphate buffers. Addition of CoCl2 increased the reaction rate without adversely affecting the enantioselectivity, and its optimum concentration was set at 0.2 mM. At lower temperatures, higher enantioselectivity was obtained, as expected. To balance reactivity and enantioselectivity, the final process was conducted at 4°C. Under these conditions, the enzymatic resolution of the racemic substrate proceeds with a high substrate loading of 100g l−1·per day, and good stereoselectivity of 94% ee for the amide product at a resolution yield of 43%, which was further enriched to over >99% ee upon recrystallization. Since the undesired R-enantiomer could be recycled by base-mediated racemization, the biocatalytic process is more atom efficient than reported chemical routes by resolution or chromatographic separation. Furthermore, the hazardous alkylation chemistry was circumvented since this chemoenzymatic process starts with pyrrolidinone (Scheme 1.2).

1.3 Atorvastatin (Lipitor®)

Atorvastatin is the API of Lipitor, a cholesterol-lowering drug with sales exceeding US$13 billion in 2008. A number of chemical processes have been reported and nearly all of them proceed through (R)-6-cyano-5-hydroxy-3-ketohexanoate (Scheme 1.3), which was obtained from either (R)-4-cyano-3-hydroxybutyrate by Claisen condensation or its ketone precursor under asymmetric hydrogenation conditions. This intermediate was then converted to the statin side-chain upon borane reduction at −70°C, followed by protection of the two hydroxyl groups to give tert-butyl (3R, 5S)-3,5-O-iso-propylidene-3,5-dihydroxyhexanoate-6-nitrile (TBIN) and reduction of the nitrile group to finalize the statin side-chain [8-10] (Scheme 1.3). These approaches suffer from the lack of synthetic convergence and high energy input as a result of relying on cryogenic reactions.

Scheme 1.3 Chemical approaches for the syntheses of the atorvastatin side-chain.

images/c01_image003.jpg

The use of 2-deoxy-D-ribose-5-phosphate aldolase (DERA) was also reported for the synthesis of statin side-chain (Scheme 1.4). The chemoenzymatic synthesis is significantly shorter, more cost effective, and safer by avoiding cryogenic reactions and hazardous reduction. In Nature, DERAs catalyze reversible aldol reaction between a d-glyceraldehyde-3-phosphate (acceptor) and acetaldehyde (donor) to form 2-deoxy-d-ribose-5-phosphate. The discovery that the acceptors could also be non-phosphate substrates led to significantly expansion of DERA’s synthetic applications [11]. The DERA-catalyzed synthesis of the statin side-chain starts with 1 equiv. of 2-chloroacetaldehyde and 2 equiv. of acetaldehyde to form a chiral lactol, which was subsequently oxidized to a lactone. The amino moiety in the statin side-chain was introduced through displacement of the chloro atom in the lactone with sodium cyanide followed by transesterification and Ni-catalyzed hydrogenation (Scheme 1.4). Almost all initial work focused on a DERA from Escherichia coli and limited success was achieved due to its strong substrate inhibition and relatively poor stability. Subsequently, the E. coli DERA was engineered by directed evolution which led to an almost 100-fold increase in volumetric activity [12]. Separately using a metagenomics approach by screening environmental DNAs, a DERA with high activity was discovered from environmental DNAs, where the substrate inhibition was partially overcome by a semi-continuous process [13]. However, the process is still linear where the cyano group has to be introduced using NaCN and then hydrogenated to the desired amino group of the statin side-chain.

Scheme 1.4 First-generation chemoenzymatic synthesis of the statin side-chain.

images/c01_image004.jpg

More recently, a more convergent DERA-catalyzed synthesis of the statin side-chain was reported using aminoaldehydes as the acceptors [14]. In this work, a representative library of DERAs were cloned from GenBank to study their activity in catalyzing sequential aldol condensation between 1 equiv. of an aminoaldehyde and 2 equiv. of acetaldehyde to form an amino lactol. It is striking that some DERAs are fairly promiscuous and active towards even bulky and lipophilic aminoaldehydes considering that the native substrate is a glyceraldehyde-3-phosphate, which is hydrophilic and anionic at pH 7 (Figure 1.1).

Figure 1.1 Some representative aminoaldehydes accepted by DERAs.

images/c01_image005.jpg

The O-benzyloxycarbonyl (Cbz)-3-aminopropionaldehyde was selected for further process development due to commercial availability. The biotransformation proceeds with high throughput (200 g l−1·per day), high yields (90-95%), excellent stereocontrol (98% ee), and 97% diastereomeric excess (de), which could subsequently be converted to the statin side-chain upon oxidation, protection, and esterification (Scheme 1.5). This new DERA process is two steps shorter than the first-generation DERA route and the hazardous cyanation chemistry was circumvented.

Scheme 1.5 Second-generation chemoenzymatic synthesis of the statin side-chain by DERAs.

images/c01_image006.jpg

1.4 Pregabalin (Lyrica®)

Pregabalin is the API of Lyrica for the treatment of neuropathic pain and epilepsy with sales of over US$2.5 billion in 2008. A number of syntheses have been reported involving diastereomeric resolution, asymmetric hydrogenation, or chiral desymmetrization followed Hoffman degradation (Scheme 1.6) [15-17]. The main issue with late-stage resolution of the racemic amino acid is that the undesired enantiomer could not be recycled efficiently and as a result the route suffers from poor atom economy.

Scheme 1.6 Chemical syntheses of pregabalin.

images/c01_image007.jpg

To address the green chemistry and cost issues, an enzymatic resolution route was recently developed with excellent process efficiency (Scheme 1.7) [18]. The key step in this process is stereoselective hydrolysis of a cyano diester (CNDE), which was prepared from isovaleraldehyde by Knoevenagel condensation and cyanation.

Scheme 1.7 Biocatalytic process for the production of pregabalin.

images/c01_image008.jpg

One of the most difficult issues in this route development is to identify reliable analytical methods. Here both direct and indirect chiral gas chromatographic (GC) methods were developed using reference standards (Scheme 1.8). Under GC conditions at 130°C, the resulting carboxylic acids from enzymatic hydrolysis are decarboxylated to give cyano esters, allowing the measurement of ees. Alternatively, all four thermostable diastereomers could be prepared by esterification (Scheme 1.8).

Scheme 1.8 Synthesis of reference standards for chiral GC analysis.

images/c01_image009.jpg

Initial screening of commercially available hydrolases showed that seven enzymes showed reasonable to good enantioselectivity with E-values >35, including lipases from Thermomyces lanuginosus, Rhizopus delemar, R. niveus, Pseudomonas sp., Mucor miehei, R. rryzae, and an esterase from M. miehei. Both lipases from T. lanuginosus and R. delemar have E-values >200 (Table 1.2). For the final optimization, the lipase from T. lanuginosus, known commercially as Lipolase, was selected due to high enantioselectivity and superior activity, and its commercial availability at large scale. Although there is no substrate inhibition, significant product inhibition was observed at a high substrate (CNDE) loading of 1-3 M. Extensive optimization showed that calcium acetate was effective in overcoming product inhibition. At a substrate loading of 3 M, only 170 mM of calcium was needed. High temperatures tend to deactivate the enzyme, and the best pH appears to be 8.0.

Table 1.2 Active enzymes screened from the hydrolase library.

The optimized process parameters are 1.5% (w/w) enzyme loading, pH 8.0, 25°C, and 170 mM Ca(OAc)2. Under these conditions, the enzymatic step has an excellent volumetric activity with a substrate loading of 765 g l−1. The resulting (S)-monoacid was obtained in high enantioselectivity (>98% ee) at a conversion of 40-45%, which could be readily converted to the final API upon thermal decarboxylation followed by saponification and hydrogenation. Since the undesired R-enantiomer could be readily racemized to CNDE by NaOEt in toluene, the overall yield was improved to over 40-45% after one recycling from <30% by the chemical resolution. Moreover, the final three steps were conducted in water and intermediates were telescoped, resulting in significant reductions in wastes and process costs. This process has currently been adopted for the manufacture of pregabalin [18].

1.5 Conclusion

Nature makes all sorts of small molecules and polymeric materials and their conjugates by enzymes, which catalyze a wide range of transformations that are challenging by chemical methods. Recent advances in large-scale DNA sequencing and enzyme-directed evolution rendered biocatalysis a more practical technology than ever for industrial applications [19]. The demand for sustainable products and efficient process technologies provides a great opportunity to advance enzyme technologies further for the chemical synthesis of fuels, chemicals, and materials. As illustrated in the example of levetiracetam, atorvastatin, and pregabalin, biotransformations provide a new dimension in route design to satisfy both the process and green chemistry metrics [20, 21].

Acknowledgments

The authors would like to thank Kim Albizati, John Tucker, Lan Xu, Jean Xie, Ningqing Ran, Shanghui Hu, Robert Scott, Weihong Yu, Stephen Bowlin, Lishan Zhao, and Carlos Martinez, whose contributions made this chapter possible.

References

1. Dolityzky, B.Z. (2004) Process for producing levetiracetam. Patent WO 2004069796.

2. Gade, S.R., Mallepalli, S.R., Muvva, V., Amirisetty, R.T., Harikeerthi, N.M., Ramasamy, V.A., Bandichhor, R., Mylavarapu, R.K., Kopparapu, R.J., Manudhane, K.S. et al. (2008) Processes for the preparation of levetiracetam. Patent WO 2008077035.

3. Tucker, J.L., Xu, L., Yu, W., Scott, R., Zhao, L., and Ran, N. (2009) Modified nitrile hydratases and chemoenzymatic processes for preparation of levetiracetam. Patent WO 2009009117.

4. Miyanaga, A., Fushinobu, S., Ito, K., Shoun, H., and Wakagi, T. (2004) Mutational and structural analysis of cobalt-containing nitrile hydratase on substrate and metal binding. Eur. J. Biochem., 271, 429–438.

5. Chica, R.A., Doucet, N., and Pelletier, J.N. (2005) Semi-rational approaches to engineering enzyme activity: combining the benefits of directed evolution and rational design. Curr. Opin. Biotechnol., 16, 378–384.

6. Morley, K.L. and Kazlauskas, R.J. (2005) Improving enzyme properties: when are closer mutations better? Trends Biotechnol., 23, 231–237.

7. Hourai, S., Ishii, T., Miki, M., Takashima, Y., Mitsuda, S., and Yanagi, K. (2005) Cloning, purification, crystallization and preliminary X-ray diffraction analysis of nitrile hydratase from the themophilic Bacillus smithii SC-J05-1. Acta Crystallogr., 61 (Part 11), 974–977.

8. Ohrlein, R. and Baisch, G. (2003) Chemo-enzymatic approach to statin side-chain building blocks. Adv. Synth. Catal., 345, 713–715.

9. Müller, M. (2004) Chemoenzymatic synthesis of building blocks for statin side-chains. Angew. Chem. Int. Ed., 44, 362–365.

10. Brower, P.L., Butler, D.E., Deering, C.F., Le, T.V., Millar, A., Nanninga, T.N., Palmer, C.W., and Roth, B.D. (1992) The convergent synthesis of (4R-cis)-1,1-dimethylethyl 6-cyanomethyl-2,2-dimethyl-1,3-dioxane-4 -acetate, a key intermediate for the preparation of CI-981, a highly potent, tissue selective inhibitor of HMG-CoA reductase. Tetrahedron Lett., 33, 2279–2282.

11. Machajewski, T.D., Wong, C.-H., and Lerner, R.A. (2000) The catalytic asymmetric aldol reaction. Angew. Chem. Int. Ed., 39, 1352–1374.

12. Jennewein, S., Schuermann, M., Wolberg, M., Hilker, I., Luiten, R., Wubbolts, M., and Mink, D. (2006) Directed evolution of an industrial biocatalyst: 2-deoxy- D-ribose 5-phosphate aldolase. Biotechnol. J., 1, 537–548.

13. Greenberg, W.A., Varvak, A., Hanson, S.R., Wong, K., Huang, H., Chen, P., and Burk, M.J. (2004) Development of an efficient, scalable, adolase-catalyzed process for enantioselective synthesis of statin intermediates. Proc. Natl. Acad. Sci. USA, 101, 5788–5793.

14. Hu, S., Tao, J., and Xie, J. (2006) Process for producing atorvastatin, pharmaceutically acceptable salts thereof and intermediates thereof. Patent WO 2006134482.

15. Hamersak, Z., Stipetic, I., and Avdagic, A. (2007) An efficient synthesis of (S)-3-aminomethyl-5-methylhexanoic acid (pregabalin) via quinine-mediated desymmetrization of cyclic anhydride. Tetrahedron: Asymmetry, 18, 1481–1485.

16. Hoekstra, M.S., Sobieray, D.M., Schwindt, M.A., Mulhern, T.A., Grote, T.M., and Huckabee, B.K. (1997) Chemical development of CI-1008, an enantiomerically pure anticonvulsant. Org. Process Res. Dev., 1, 26–38.

17. Burk, M.J., DeKoning, P.D., Grote, T.M., Hoekstra, M.S., Hoge, G., Jennings, R.A., Kissel, W.S., Le, T.V., Lennon, I.C., Mulhern, T.A. et al. (2003) An enantioselective synthesis of (S)-(+)-3-aminomethyl-5-methylhexanoic acid via asymmetric hydrogenation. J. Org. Chem., 68, 5731–5734.

18. Martinez, C.A., Hu, S., Dumond, Y., Tao, J., Kelleher, P., and Tully, L. (2008) Development of a chemoenzymatic manufacturing process for pregabalin. Org. Process Res. Dev., 12, 392–398.

19. Tao, J., Lin, G., and Liese, A. (2009) Biocatalysis for the Pharmaceutical Industry – Discovery, Development, and Manufacturing, Wiley-VCH Verlag GmbH, Weinheim.

20. Ran, N., Zhao, T., Chen, Z., and Tao, J. (2008) Recent applications of biocatalysis in developing green chemistry for chemical synthesis at industrial scale. Green Chem., 10, 361–372.

21. Tao, J. and Xu, J. (2009) Biocatalysis: greening the pharmaceutical industry. Curr. Opin. Chem. Biol., 13, 43–50.

2

Enantioselective Hydrogenation: Applications in Process R&D of Pharmaceuticals

Kurt Püntener and Michelangelo Scalone

2.1 Introduction

In the last 10-15 years, various new selective and efficient catalysts have increased the potential of asymmetric catalysis to deliver short entries into complex chiral non-racemic molecules [1]. This higher versatility is of particular importance to the pharmaceutical industry because the new chiral drugs usually have complex structures and are marketed almost exclusively as single enantiomers [2]. The activities at Roche’s Catalysis Group have traditionally been focused on asymmetric reactions requiring special equipment (e.g., autoclaves, continuous reactors) and gases under pressure (e.g., hydrogen, carbon monoxide, carbon dioxide, ammonia). Asymmetric hydrogenation has been and still is the preferred reaction to generate chiral centers with the desired configurations although, more recently, asymmetric transfer hydrogenation (ATH) has also been applied with success. As a consequence, we have been able to hydrogenate enantioselectively an increasing number of substrates with functional groups other than C=C double bonds. Indeed, out of the five examples selected from our recent Synthesis and Process Research work discussed in this chapter, four involve the reduction of a carbonyl group (three by hydrogenation, one by transfer hydrogenation) and one deals with the reduction of an imine salt (by hydrogenation).

2.2 Carbonyl Hydrogenations

2.2.1 Asymmetric Hydrogenation with Dynamic Kinetic Resolution of Racemic 1,4-Dibenzylpiperidin-3-one

Ro 67-8867 [(S,S)-4] is a high-affinity, selective, and activity-dependent antagonist of the N-methyl-d-aspartate (NMDA) receptor (Scheme 2.1). Among the various drawbacks with regard to scale-up which rendered the original Discovery Chemistry synthesis of (S,S)-4 unsuitable for technical development, two are particularly worth mentioning: (i) the piperidinone rac-1 was isolated as the free base but turned out to be unstable even at room temperature; therefore, it could not be purified and had to be brought rapidly into the next step; (ii) the key chiral intermediate (S,S)-2 was prepared by resolution of its racemate by crystallization of the diastereomeric salts with O,O′-dibenzoyl-d-tartaric acid with only 10-12% yield.¹) (Jaeschke, G. and Waldmeier, P., F. Hoffmann-La Roche Ltd., unpublished results.).

Scheme 2.1 The resolution-based and the enantioselective synthesis of Ro 67-8867.

images/c02_image001.jpg

After a short series of exploratory experiments, Noyori’s catalysts of type [RuCl2(chiral diphosphine)(chiral diamine)] [3] proved to be the most suitable for the desired hydrogenation of rac-1 to (S,S)-2 under conditions of dynamic kinetic resolution. The investigations in our laboratories then included (i) the selection of a suitable salt of rac-1, (ii) the fine tuning of the structure of both chiral ligands in the ruthenium catalyst, and (iii) the optimization of various reaction parameters such as the solvent, the hydrogen pressure and the base used.

The most suitable salt of rac-1 in terms of simplicity of preparation, stability, and purity was the hydrochloride rac-1.HCl. The latter was obtained in two steps in 78% yield as a white solid which proved to be stable at room temperature for weeks [4]. All diphosphines tested belong to the MeOBIPHEP family (Figure 2.1 and Table 2.1) [5]²). They are easily accessible in analogy with the parent compound MeOBIPHEP (Figure 2.1) [5, 6]. Moreover, various chiral diamines were tested; the results obtained with 1,2-diphenylethylenediamine (DPEN) were clearly better than those with 1,2-di-tert-butylethylenediamine (DTBEN), 1,2-diaminocyclohexane (DACH), and 2,2′-diaminobinaphthyl (DABN). The S-R,R combination of the configuration of the diphosphine and the diamine afforded the higher enantioand diastereoselectivity than the R-R,R combination (entries 1 and 2). The best catalysts contained (R,R)-DPEN and members of the MeOBIPHEP family which had substituents in the 3- and 5-positions of the aryl moieties at phosphorus. The highest ee values (96-97%) were obtained when two isopropyl groups were present (entry 4). However, if an additional substituent was present in the para position or if the meta substituents were bulkier, the ee and in part also the cis/trans ratio decreased considerably (entries 5 and 6). Among the bases tested (Li-, Na-, or KOtBu, KOSiMe3, KOMe, Cs2CO3, K2CO3, Rb2CO3, KOH), KOtBu and its sodium analog generated the catalysts with the highest enantioselectivity. The amount of base employed strongly influenced the activity of the catalyst, the highest activity being obtained with a substrate/base ratio of 5-10. Under the optimal conditions (20°C and 40 bar of hydrogen pressure), the asymmetric hydrogenation of rac-1.HCl was complete within a few hours in the presence of a tiny amount of catalyst. Specifically, the highest molar substrate to catalyst ratios (S/Cs) were achieved with 3,5-Xyl-MeOBIPHEP (800 000) and 3,5-iPr-MeOBIPHEP (200 000). The (S,S)-2 obtained was of sufficient chemical and enantiomeric purity to be used directly as a crude material in the next step. Finally, (S,S)-4 was obtained in 53% overall yield compared with 3.5% of the original resolution-based synthesis [7]. The new synthesis was established rapidly on a technical scale such that 50 kg of (S,S)-4 could be produced within 6 months from the first order.³)

Figure 2.1 Ligand structures and abbreviations.

images/c02_image002.jpg

Table 2.1 Enantio- and diastereoselective hydrogenation of piperidinone rac-1. HCla.

images/c02_image003.jpg

a[RuCl2 (diphosphine)[(R,R)-DPEN]], S/C 1000-2000, 1.4 molar equiv. KOtBu, iPrOH, 20°C, 40 bar H2. Since the substrate rac-1 was employed as its hydrochloride, 1 molar equiv. of base simply liberated the amine.

2.2.2 Asymmetric Hydrogenation of Methyl 3-Oxotetradecanoate

Orlistat (12), the active pharmaceutical ingredient of Xenical™, is an effective anti-obesity agent that selectively inhibits gastrointestinal lipases. The first-generation synthesis [rac−6 → rac−7 → (S, S, R)−8 → 12] was based on the resolution of the racemic intermediate 8 (Scheme 2.2). The overall yield of 18% is a remarkable achievement considering that the undesired diastereomer (R,R,S)-8 could not be recycled [9]. In order to develop a more efficient process with an increased throughput, a second-generation synthesis has been developed. The strategy has been on the one hand to retain the final reaction sequence from (S,S,R)-8-12 in order to affect the impurity profile of 12 as little as possible. On the other hand, however, the late-stage resolution of rac-8 has been avoided by the use of enantiomerically pure intermediates (S)-6 and (S,S,R)-7 rather than their racemates. In this synthetic concept, (R)-hydroxy ester 10 became the first chiral non-racemic intermediate and the hydrogenation of methyl 3-oxotetradecanoate (9) the enantioselective step in the synthesis.

Scheme 2.2 First-(resolution based) and secondgeneration (enantioselective) syntheses of orlistat.

images/c02_image004.jpg

Based on the first encouraging results obtained with a Ru-(R)-BIPHEMP catalyst [10], a new process based on Ru-(R)-MeOBIPHEP was elaborated. The catalytically active species [RuCln (diphosphine)] was generated by addition of hydrogen chloride to the corresponding [Ru(OAc)2 (diphosphine)] complex [11]. Although only two molar equivalents of hydrogen chloride are formally required to form the proposed dichloro catalyst by replacement of the two coordinated acetates [12], the best results in terms of enantioselectivity and reaction rate were obtained at S/C 50000 with a about a 20-fold excess of hydrogen chloride. A short screening of diphosphines with various backbone types confirmed that ee values of 99% were reached only when atropisomeric ligands such as (R)-BIPHEMP, (R)-MeOBIPHEP, (R)-3,5-tBu-MeOBIPHEP, and (R)-Fur-MeOBIPHEP or (R)-TMBTP [13] were used. Finally, Roche proprietary (R)-MeOBIPHEP was preferred owing to its easier accessibility on a large scale.

The complex [Ru(OAc)2 [(R)-MeOBIPHEP]] proved to be air stable for at least 6 months. In contrast, the active species formed in methanol solution by addition of aqueous HCl was highly air sensitive. Therefore, careful exclusion of air during both the catalyst formation step and the hydrogenation were crucial for high catalyst performance. In the course of the optimization of the process, it was found that technical-grade solvent (methanol) could be used and that the hydrogenation proceeded satisfactorily even at a concentration of 96 wt.% or at an S/C ratio of 250000. Finally, the hydrogenation was run at 40 wt.% concentration at S/C 50 000 under 40 bar of hydrogen at 80°C to achieve full conversion within 4 h. The scale-up from a 2l to a 2 m³ autoclave proceeded uneventfully such that a total of 2.2 tons of (R)-10 with 99% ee and 99% yield was produced in up to 240kg batches.

2.2.3 Asymmetric Hydrogenation of 1,1,1-Trifluoroacetone

(S)-1,1,1-Trifluoro-2-propanol [(S)-14] is the chiral building block in the synthesis of a variety of new Type 1 Glycine Transporter (GlyT-1) inhibitors with general structure (S)-15 under development for the treatment of schizophrenia and psychotic disorders (Scheme 2.3) [14].

Scheme 2.3 Product and by-products of the asymmetric hydrogenation of 1,1,1-trifluoroacetone.

images/c02_image005.jpg

A technically feasible process for (S)-14 had to meet the following stringent requirements: (i) the ee value for (S)-14 of 98.9% present in the specifications had to be achieved already by the hydrogenation since the ee upgrade of a more advanced synthetic intermediate towards (S)-15 had proved very laborious and low yielding; (ii) the low boiling points of 13 (22°C/1 atm) [15] and (S)-14 (79°C/1 atm) [16] called for an operationally simple process where the separation from the process solvent would be an important issue; and (iii) trifluoroacetone has a high propensity to self-condense in the presence of base [17]. Exploratory experiments with a set of standard Ru or Rh catalysts containing chiral diphosphine ligands evidenced in general poor catalyst activity and/or enantioselectivity. Interesting results were achieved with [Rh(CF3 CO2)[(S)-Cy-oxo-ProNOP]]2, the same catalyst which had been successfully used with analogous substrates [18]. At S/C 200 in toluene, complete conversion was achieved and (S)-14 was isolated with up to 86% ee. At lower catalyst loadings, however, only incomplete conversion was observed.

In the presence of Noyori’s ruthenium catalysts [RuCl2(diphosphine)(diamine)], KOtBu as base and isopropanol as solvent (the use of a strong base and isopropanol as solvent is reported to be crucial for high catalytic activity) [19], (S)-14 was obtained with 90% ee. However, the selectivity was only 50% due to the formation of various unsaturated and hydrogenated aldol by-products (Scheme 2.3). Additionally, (S)-14 could not be separated from isopropanol by distillation on a laboratory scale due to the very small difference between their boiling points (79 and 82°C, respectively) [20].

In contrast, the base-free catalysts with general formula [RuH(BH4)(diphosphine) (diamine)] [21] afforded (S)-14 in up to 97.8% ee and >95% purity. The best system contained (S)-3,5-iPr-MeOBIPHEP as diphosphine and (R,R)-DPEN as diamine. However, the highest S/C ratio achieved was only 2000 at 40 bar and 40°C. Moreover, these hydridotetrahydroborato catalysts showed partial decomposition within weeks even when stored in a refrigerator under argon or in a glove-box (<1 ppm O2). Therefore, the work on the robust and stable ruthenium dichloro catalysts was resumed and particular attention was devoted to the type of base used. A wide screening led to the observation that weak bases (e.g., acetate, formate, hydrogencarbonate) were able to activate efficiently the ruthenium dichloro catalysts and promoted the formation of only traces, if any, of self-condensation products of 13 [22]. Under optimized conditions, (S)-14 was obtained in 99.2% ee and 99.8% purity using [RuCl2[(S)-3,5-tBu-MeOBIPHEP][(R,R)-DPEN]] at an S/C ratio of up to 30000 in the presence of 0.005 molar equivalent of sodium formate as base and 3 wt. % of water (Table 2.2, entry 4). The hydrogenation was run neat; (S)-14 was easily separated from the catalyst and from traces of self-condensation by-products by distillation.

Table 2.2 Enantioselective hydrogenation of 1,1,1-trifluoroacetonea.

images/c02_image006.jpg

a[RuCl2 [(S)-diphosphine][(R,R)-DPEN]] S/C 20 000, 40 bar H2, 3 wt.% H2 O, no solvent, 0.005 molar equiv. NaOCHO, 40-60°C.

bAfter distillation.

In the course of the scale-up work, we were confronted with the fact that the advantages of running a reaction without solvent with regard to high throughput and simplicity of work-up also posed major safety issues. The heat developed by the hydrogenation was estimated⁴) to be ΔrH = −66 kJ mol−1, corresponding to a calculated total adiabatic temperature increase Δadia Tmax of about 359°C. An excessive temperature increase would favor self-condensation reactions (estimated Δadia Tmax ≈ 140°C) and thermal decomposition of 13 [T(onset) 144°C in differential scanning calorimetry]. Clearly, the heat generated had to be removed efficiently and reliably. Therefore, factors such as the operating fill level, the geometry of the stirrer, the mode of hydrogen feed, and the type of heating/cooling system were optimized depending on the autoclave employed in addition to the usual parameters (such as temperature, pressure, quality of substrate, gases, and catalyst). Finally, the reactions in autoclaves with a volume larger than 1 l were run at a fill level of at least 80%. Under these conditions, even in the case of a simultaneous failure of both the standard cooling and the independent emergency cooling system, the hydrogen valve would be immediately closed such that the consumption of the hydrogen present in the autoclave would bring about a maximum temperature increase of only about 12°C. Accordingly, the temperature of the reaction mixture would remain well under the 144°C decomposition threshold of 13. Finally, this procedure was successfully scaled up: five pilot runs were carried out on a 50 kg scale in a 50 l autoclave and delivered (S)-14 in 96% yield, 99.8% purity, and 99.2% ee after distillation.

2.2.4 Asymmetric Transfer Hydrogenation of Levodione

(3R,3′R)-Zeaxanthin (21) is undergoing intensive investigation for the treatment of age-related macula degeneration (AMD), a disorder associated with the macula, the area where visual acuity is best [23]. In Nature, the highest overall density of 21 can be found in egg yolk and in maize. The first chiral intermediate in the synthesis of 21 [24], levodione [(R)-18], is readily accessible by fermentation (Scheme 2.4) [25]. In contrast, the selective conversion of the latter to trans-actinol [(R,R)-19] had still remained elusive.

Scheme 2.4 Synthesis of (3R,3′ R)-zeaxanthin.

images/c02_image007.jpg

The Raney nickel-catalyzed hydrogenation of (R)-18 had been reported to afford an 80: 20 mixture of (R,R)- and (S,R)-19 [25], whereas the Ru-BINAP-catalyzed hydrogenation proceeded with modest chemoselectivity affording (R,R)-19 in only 25% isolated yield after intensive purification [26]. Work in our laboratories with ruthenium- and rhodium-based catalysts only confirmed the poor chemo- and enantioselectivity of the asymmetric hydrogenation of (R)-18. The ATH of rac-18 with concomitant dynamic kinetic resolution afforded only very complex mixtures in the presence of base-activated ruthenium catalysts such as [RuCl[(S, S)−Ts−DPEN{−H}](p-cymene)] [27]. Finally, the ATH of (R)-18 was achieved with high selectivity by using catalysts of general formula [Ru(Ts-diamine{−2H})(η⁶-arene)] [28] which were prepared by treatment of the corresponding monochloro complexes [RuCl(Ts-diamine{−H})(η⁶-arene)] with base in a preceding step and offered accordingly a neutral reaction medium (Table 2.3). The presence of the stereocenter in the 6-position of (R)-18 was sufficient to control the diastereoselective formation of the center in the 4-position [29]. Accordingly, the results obtained with the dianion of N-Ts-ethylenediamine matched fully those obtained with the more expensive (S,S)-Ts-DPEN (entries 1 and 5). The process has been carried out in our laboratories on a few hundred grams scale, affording (R,R)-19 in 74% yield and 99.9% ee (after crystallization); it has been transferred to the Technical Process Development Department for further scale-up and technical implementation.⁵)

Table 2.3 Asymmetric transfer hydrogenation of (R)-18 with base-free catalystsa.

images/c02_image008.jpg

a[Ru(Ts-diamine{-2H})(η⁶ -arene)], S/C 20-100, room temperature, 3-24 h.

bGas chromatography (GC) area%.

2.3 Imine Hydrogenation

2.3.1 Asymmetric Hydrogenation of Hexabase Hydrogensulfate

Dextromethorphan hydrobromide (26.HBr), an important ingredient in various cough-relieving medications, is manufactured by a resolution-recycling route (bottom sequence in Scheme 2.5). The first generation of enantioselective routes entailed the Ru-diphosphine-catalyzed asymmetric hydrogenation of the (Z)-N-acylenamides 24a or 24b (diphosphine: members of the BINAP [30] or BIPHEMP/MeOBIPHEP [31] families). Particularly with 24b as substrate, the outcome of the hydrogenation step was excellent with (S)-MeOBIPHEP [S/C up to 20 000 at 60 wt. % concentration, 99% ee, and 97% isolated yield of (S)-25b].

Scheme 2.5 Resolution and enantioselective hydrogenation routes to dextromethorphan.

images/c02_image009.jpg

A few years later, work at Lonza and in our laboratories demonstrated the viability of the asymmetric hydrogenation of hexabase salts 22.HX (HX = H3PO4[32] or H2SO4[33]). On the one hand, this approach represented the smallest possible deviation from the resolution route. On the other hand, both iridium-based systems achieved only relatively low S/C ratios (1000-1500) and moderate ee values (84-89%). Moreover, the chemoselectivity (in particular overhydrogenation) was an issue. To overcome this problem, a very broad screening of chiral diphosphines with diverse backbones was carried out.⁶) As a result, we found that the asymmetric hydrogenation of 22.H2SO4 could be run chemoselectively and with low catalyst loading in the presence of iridium or rhodium catalysts containing members of the MeOBIPHEP diphosphine family (Table 2.4). The best results in terms of both chemo- and enantioselectivity were achieved with bulky substituents (e.g., tert-butyl) on the aryl groups bound to phosphorus [34].

Table 2.4 Enantioselective hydrogenation of 22. H2SO4a.

images/c02_image010.jpg

aCat*: [M[(S)-diphosphine](COD)]Cl- 4 Bu4 NI, S/C 1000, 40 bar H2, 0.1 iPr2 NEt, 80°C, 16 h, 5-10wt.%intoluene-methanol(1:1).

bSelectivity = area% of 23/(100 - area% of 22) (by GC).

cOf crude (S)-octabase.

Under optimized conditions, both catalytic systems gave rise to high S/C ratios and virtually perfect chemoselectivity (>99%) (Table 2.5). With the rhodium-based system, the ee was high enough to eliminate an upgrade step, whereas with the iridium-based system the ee upgrade step would still be required. The asymmetric hydrogenation of 22.H2SO4 and the subsequent isolation were scaled up in our laboratories to a scale of several hundred grams and transferred to the Technical Process Development Department.

Table 2.5 Current status of the enantioselective hydrogenation of 22. H2 SO4a.

aCat*: [M[(S)-3,5-tBu-MeOBIPHEP](COD)]Cl.

bSelectivity = area% of 23/(100 - area% of 22) (by GC).

cOf crude (S)-octabase.Notes

2.4 Conclusion

The enantioselective processes described confirm the high potential and the usefulness of catalytic methods and specifically of the transition metal-catalyzed asymmetric hydrogenation. With these processes, intermediates for both active pharmaceutical ingredients in clinical development and for marketed substances have been produced in kilogram to ton amounts. High S/C ratios were achieved by using hydrogenation substrates of high quality and catalysts containing diphosphines with the appropriate structure. Consequently, the contribution of the catalyst cost to the total economy of the synthesis was very low.

An important additional aspect was common to all projects and is worth mentioning, that is, the necessity to make the new synthesis of a complex target available within a short time frame. For this purpose, it is essential for chemists in the Chemical Process Research Department of a pharmaceutical company that a number of requirements are met. These are the availability of a well-stocked library of chiral ligands and of metal complexes, the availability of modern equipment to run reactions under pressure on small to large scale, and the availability of a staff with wide expertise also in other efficient catalytic (e.g., the metal-catalyzed formation of C−C or C−N bonds, bio- or organocatalytic reactions) and non-catalytic methodologies. The result will be the sound integration of the most advantageous methodologies in the synthetic scheme and finally the creation of well scalable processes.

References

1. (a) Mikami, K. and Lautens M. (eds.) (2007) New Frontiers in Asymmetric Catalysis, John Wiley & Sons, Inc., Hoboken, NJ; (b) Blaser, H.U. and Schmidt, E. (eds.) (2004) Asymmetric Catalysis on Industrial Scale, Wiley-VCH Verlag GmbH, Weinheim; (c) Ojima, I. (ed.) (2000) Catalytic Asymmetric Synthesis, 2nd edn, Wiley-VCH, Inc., New York; (d) Jacobsen, E.N., Pfaltz, A., and Yamamoto, H. (eds.) (1999) Comprehensive Asymmetric Catalysis I–III, Springer, Berlin.

2. Thayer, A.M. (2007) Chem. Eng. News, 85, 11–19.

3. (a) Kitamura, M. and Noyori, R. (1993) J. Am. Chem. Soc., 115, 144–152; (b) Noyori, R., Sandoval, C.A., Muñiz, K., and Ohkuma, T. (2005) Philos. Trans. R. Soc. London A, 363, 901–912, and references therein.

4. Scalone, M. and Waldmeier, P. (2003) Org. Process Res. Dev., 7, 418–425.

5. (a) Schmid, R., Foricher, J., Cereghetti, M., and Schönholzer, P. (1991) Helv. Chim. Acta, 74, 370–389; (b) Schmid, R., Broger, E.A., Cereghetti, M., Crameri, Y., Foricher, J., Lalonde, M., Müller, R.K., Scalone, M., Schoettel, G., and Zutter, U. (1996) Pure Appl. Chem., 68, 131–138.

6. (a) Schmid, R., Cereghetti, M., Heiser, B., Schönholzer, P., and Hansen, H.-J. (1988) Helv. Chim. Acta, 71, 897–929; (b) Schmid, R. and Scalone, M. (2008) (R)- and (S)-2,2′-bis(diphenylphosphino)-6,6′- dimethoxy-1,1′-biphenyl, in eEROS Encyclopedia of Reagents for Organic Synthesis, (ed. L.A. Paquette), John Wiley & Sons, Ltd., Chichester.

7. Crameri, Y., Scalone, M., Waldmeier, P., and Widmer, U. (2001) European Patent Application EP 1136475, to F. Hoffmann-La Roche AG.

8. D`echamps, I., Gomez Pardo, D., Karoyan, P., and Cossy, J. (2005) Synlett, 1170–1172.

9. (a) Karpf, M. and Zutter, U. (1991) European Patent Application EP 443449, to F. Hoffmann-La Roche AG. (b) Birk, R., Karpf, M., Püntener, K., Scalone, M., Schwindt, M.A., and Zutter, U. (2006) Chimia, 60, 561–565.

10. (a) Heiser, B., Broger, E., and Crameri, Y. (1991) Tetrahedron: Asymmetry, 2, 51–62; (b) Broger, E., Karpf, M., and Zutter, U. (1995) European Patent Application EP 0643052, to F. Hoffmann-La Roche AG.

11. (a) Schwindt, M.A., Fleming, M.P., Han, Y.K., Hodges, L.M., Johnston, D.A., Micheli, R.P., Roberts, C.R., Snyder, R., Topping, R.J., Püntener, K., and Scalone, M. (2007) Org. Process Res. Dev., 11, 524–433; (b) Harrington, P., Hodges, M., Püntener, K., and Scalone, M. (2001) European Patent Application EP 1127886, to F. Hoffmann-La Roche AG.

12. (a) Noyori, R., Ohkuma, T., Kitamura, M., Takaya, H., Sayo, N., Kumobayashi, H., and Akutagawa, S. (1987) J. Am. Chem. Soc., 109, 5856; (b) Noyori, R. (1996) Acta Chem. Scand., 50, 380–390.

13. Benincori, T., Cesarotti, E., Piccolo, O., and Sannicolo, F. (2000) J. Org. Chem., 65, 2043–2047.

14. Jolidon, S., Narquizian, R., Norcross, R., and Pinard, E. (2007) International Patent Application WO 2007/147770, to F. Hoffmann-La Roche AG.

15. (a) Sykes, A., Tatlow, J.C., and Thomas, C.R. (1956) J. Chem. Soc., 835–839; (b) Haszeldine, R.N. and Leedham, K. (1952) J. Chem. Soc., 3483–3490.

16. (a) Bucciarelli, M., Forni, A., Moretti, I., and Torre, G. (1983) Synthesis, 11, 897–899; (b) Horvath, A.L. (2001) Chemosphere, 44, 897–905.

17. (a) Mead, D., Loh, R., Asato, A.E., and Liu, R.S. (1985) Tetrahedron Lett., 26, 2873–2876; (b) Sheppard, W.A. and Sharts, C.M. (1969) Organic Fluorine Chemistry, W.A. Benjamin, New York, pp. 437–438.

18. Kuroki, Y., Sakamaki, Y., and Iseki, K. (2001) Org. Lett., 3, 457–459.

19. Noyori, R. and Ohkuma, T. (2001) Angew. Chem. Int. Ed., 40, 40–73.

20. Lammiman, S.A. and Satchell, R.S. (1972) J. Chem. Soc., Perkin Trans. 2, 2300–2304.

21. Okhuma, T., Koizumi, M., Muñiz, K., Hilt, G., Kabuto, C., and Noyori, R. (2002) J. Am. Chem. Soc., 124, 6508–6509.

22. Püntener, K. and Waldmeier, P. (2008) International Patent Application WO 2008/012240, to F. Hoffmann-La Roche AG.

23. (a) Zhao, L. and Sweet, B.V. (2008) Am. J. Health-Syst. Pharm., 65, 1232–1238; (b) Cho, E., Hankinson, S.E., Rosner, B., Willet, W.C., and Colditz, G.A. (2008) Am. J. Clin. Nutr., 87, 1837–1843; (c) Samimi, D.B. (2005) Nutr. Bytes, 10, Article 7.

24. (a) Widmer, E., Soukup, M., Zell, R., Broger, E.A., Wagner, H.P., and Imfeld, M. (1990) Helv. Chim. Acta, 73, 861–867; (b) Englert, G., Noack, K., Broger, E.A., Glinz, E., Vecchi, M., and Zell, R. (1991) Helv. Chim. Acta, 74, 969–982.

25. (a) Leuenberger, H.G.W., Boguth, W., Widmer, E., and Zell, R. (1976) Helv. Chim. Acta, 59, 1832–1849;0 (b) Sode, K., Kajiwara, T., Tamiya, E., Karube, I., Hori, N., and Yanagimoto, T. (1987) Biocatalysis, 1, 77–89.

26. Brunner, H. and Fisch, K. (1993) J. Organomet. Chem., 456, 71–75.

27. (a) Hashiguchi, S., Fujii, A., Takehara, J., Ikarija, T., and Noyori, R. (1995) J. Am. Chem. Soc., 117, 7562–7563; (b) Noyori, R. and Hashiguchi, S. (1997) Acc. Chem. Res., 30, 97–102; (c) Noyori, R., Yamakawa, M., and Hashiguchi, S. (2001) J. Org. Chem., 66, 7931–7944.

28. Haack, K.J., Hashiguchi, S., Fujii, A., Ikarija, T., and Noyori, R. (1997) Angew. Chem. Int. Ed. Engl., 36, 285–288.

29. Crameri, Y., Püntener, K., and Scalone, M. (1999) European Patent Application EP 915076, to F. Hoffmann-La Roche AG.

30. Kitamura, M., Hsiao, Y., Noyori, R., and Takaya, H. (1987) Tetrahedron Lett., 28, 4829–4832.

31. (a) Broger, E.A., Crameri, Y., and Heiser, B. (1989) European Patent Application EP 315886, to F. Hoffmann- La Roche AG. (b) Heiser, B., Broger, E.A., and Crameri, Y. (1991) Tetrahedron: Asymmetry, 2, 51–62.

32. (a) Imwinkelried, R. (1997) Chimia, 51, 300–302; (b) Werbitzky, O. (1998) Chim. Oggi, 16, 86–89.

33. (a) Schmid, R. and Scalone, M. (1999) in Comprehensive Asymmetric Catalysis I–III (eds. E.N.Jacobsen, A. Pfaltz, and H. Yamamoto), Springer, Berlin, pp. 1439–1449; (b) Broger, E.A., Scalone, M., and Wehrli, C. (1998) European Patent Application EP 850931, to F. Hoffmann-La Roche AG.

34. Püntener, K., Scalone, M., and Wang, S. (2003) International Patent Application WO 2003/078399, to Roche Vitamins AG.

---

1) Jaeschke, G., and Waldmeier, P., F. Hoffmann-La Roche Ltd., unpublished results.

2) The chiral diphosphines of the MeO-BIPHEP family are now commercially available from Solvias AG, Basel.

3) Recently, a new formal synthesis of (S,S)-3 based on an enantioselective ring enlargement has been published [8].

4) Based on the heat of formation for the liquid starting material 13, the product 14, and aldol by-product 16, all three calculated with a group contribution method at 25°C/1 bar.

5) In October 2003, Roche’s Vitamins and Fine Chemicals Division was sold to DSM. Zeaxanthin is now marketed as OPTISHARP™ by DSM Nutritional Products for use in human nutrition and health products.

6) Part of this work was contracted out to Solvias AG.

3

Chiral Lactones by Asymmetric Hydrogenation - a Step Forward in (+)-Biotin Production

Werner Bonrath, Reinhard Karge, Thomas Netscher, Felix Roessler, and Felix Spindler

3.1 Introduction: (+)-Biotin as an Example for the Industrial Production of Vitamins

The business of vitamins and fine chemicals has become a field of strong competition during recent decades [1]. The success of manufacturing within this industry segment is mainly determined by economic and ecological conditions. The efficiency of chemical transformations is a prerequisite to meet the requirements for commercialization. For the synthesis of chiral products or intermediates, asymmetric homogeneous catalysis with metal complexes, and in particular asymmetric hydrogenation, serves as a potent toolbox for developing highly selective processes [2-4]. Such achievements on moving towards the ideal scenario of green chemistry [5-7] not only allow us to establish environmentally benign procedures, but also provide economic benefits.

(+)-Biotin (vitamin H, 1) is a member of the water-soluble B-vitamins and is of importance for human and animal health. (+)-Biotin acts as a cofactor of enzymes catalyzing carboxylation (carboxyl group transfer) reactions, and is an essential growth factor in all living cells. An adult human needs about 0.03-0.1 mg of (+)-biotin per day. Biotin possesses three stereogenic centers. The only isomer exhibiting full biological activity is the one with the configuration (3aS,4S,6aR), d-(+)-biotin (1, Figure 3.1) [8]. In addition to its function as a nutrient, 1 has found applications in various technologies, for example, in the avidin-streptavidin binding assay based on the exceptionally high ligand-protein affinity, and photoaffinity labeling [9].

Figure 3.1 (+)-Biotin (1).

images/c03_image001.jpg

The chemical history of biotin starts with the publications of the first total synthesis of racemic biotin and subsequent optical resolution by Harris et al. at Merck in 1943 [10]. Goldberg and Sternbach of F. Hoffmann-La Roche applied for patents on the first commercially applicable biotin synthesis in 1946 (publication in 1949) [11-13]. Since then, many industrial and academic research laboratories have directed their work towards an optimum synthetic approach or, alternatively, a biotechnological method. The general production method still applied today is multi-step chemical synthesis. The world market for 1 is about 100 t per year. Current manufacturers of (+)-biotin (1) are DSM and several Chinese producers.

From a chemical point of view, the following general problems accompanied with efficient routes to 1 have to be solved in an economically and ecologically satisfactory manner: introduction of nitrogen and sulfur functionalities to form the highly functionalized bi-heterocycle, introduction of the C5 side-chain, and generation of the three stereogenic centers of the all-cis-thiophane ring. An excellent review of approaches to biotin, with discussion of synthetic strategies, was published by De Clercq [14]. An update of newer literature is given in the compilation of Seki [9].

3.2 Commercial Syntheses and Other Routes to (+)-Biotin by Total Synthesis

Although the Goldberg-Sternbach concept described in their patents dates back to 1946 [11-13], this lactone-thiolactone approach is still valuable today. The cyclic anhydride 6 was obtained by starting from readily available fumaric acid (2) via the meso-compounds 3-5 (Scheme 3.1). After several functional group transformations with racemic thiolactone rac-7 as an intermediate, rac-8 was transformed to the racemic sulfonium salt rac-9. (+)-Biotin (1) was produced by a C2 elongation (→ 11) decarboxylation sequence. Optical resolution by use of d-camphorsulfonic acid delivered the chiral salt 10.

Scheme 3.1 The Goldberg-Sternbach concept.

images/c03_image002.jpg

The original Goldberg-Sternbach concept was improved significantly by Gerecke et al. [15]. They found that (chiral) lactone 12 can be directly converted with potassium thioacetate to (chiral) thiolactone 7 (Scheme 3.2). The optical resolution step, taking place advantageously at a relatively late stage, delivered d-lactone 12 via the crystalline ephedrine salt of the diastereomeric half-ester, reduction, and cyclization. The undesired stereoisomer was recycled by acidic back-hydrolysis to diacid 5. This procedure was operated on a commercial scale until the 1990s.

Scheme 3.2 The improved Goldberg-Sternbach concept: direct conversion of lactone to thiolactone and late optical resolution.

images/c03_image003.jpg

A further improvement was the approach of diastereoselective ring opening of anhydride 6 with a chiral alcohol, devised by Pauling and Wehrli [16] (Scheme 3.3), replacing the optical resolution step in production. d-Lactone was thus formed by reduction of the selectively formed diastereoisomeric half-ester by treatment with a complex hydride and ring closure.

Scheme 3.3 The Pauling-Wehrli concept of diastereoselective ring opening.

images/c03_image004.jpg

An interesting and very short route involving an Rh(I)-catalyzed asymmetric hydrogenation step as a key transformation has been applied by Lonza (Scheme 3.4) [17-20]. Tetronic acid (13) prepared from diketene was used as a cheap starting material. In cooperation with the catalysis group of the former Ciba-Geigy, the selectivity of the heterogeneous diastereoselective hydrogenation of intermediate 14 could be improved from originally 70:30 of diastereomers 16a:16b to >99:1 with the diphosphane josiphos2 (15) as a ligand. The production via lactone 17 was performed on multi-ton scale, but had to be terminated due to a severe drawback of this approach: in the final deprotection step by hydrogenation of alkene 18 to (+)-biotin (1), the chirality of the (expensive) auxiliary is destroyed, which has a dramatic impact on the overall production cost.

Scheme 3.4 The Lonza concept: (+)-biotin process using asymmetric hydrogenation.

images/c03_image005.jpg