Biocatalytic Asymmetric Synthesis of Chiral Amines from Ketones Applied to Sitagliptin Manufacture Christopher K. Savile et al. Science 329 , 305 (2010); DOI: 10.1126/science.1188934

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others , you can order high-quality copies for your colleagues, clients, or customers by clicking here. Permission to republish or repurpose articles or portions of articles can be obtained by following the guidelines here.

The following resources related to this article are available online at www.sciencemag.org (this information is current as of October 25, 2012 ):

Updated information and services, including high-resolution figures, can be found in the online version of this article at: http://www.sciencemag.org/content/329/5989/305.full.html Supporting Online Material can be found at: http://www.sciencemag.org/content/suppl/2010/06/17/science.1188934.DC1.html A list of selected additional articles on the Science Web sites related to this article can be found at: http://www.sciencemag.org/content/329/5989/305.full.html#related This article cites 24 articles , 2 of which can be accessed free: http://www.sciencemag.org/content/329/5989/305.full.html#ref-list-1 This article has been cited by 6 articles hosted by HighWire Press; see: http://www.sciencemag.org/content/329/5989/305.full.html#related-urls This article appears in the following subject collections: Chemistry http://www.sciencemag.org/cgi/collection/chemistry

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright 2010 by the American Association for the Advancement of Science; all rights reserved. The title Science is a registered trademark of AAAS.

REPORTS

occur, according to the existing data, and the orderdisorder transition can thus indeed be a real melting process or a transition into a glassy state.

counterintuitive. Based on the finding of a much stronger molecule-substrate interaction in the LT phase, we believe that the local atomic order at the interface, although different from molecule to molecule (no long-range order), is considerably higher in the LT (compared to the RT) phase, connected with a shorter distance between substrate and adsorbate atoms. Thus, heating leads to a reduction of the short-range order (DS > 0), which is not completely compensated for by the appearance of long-range order (DS < 0).

This scenario may bear some analogy to the description of inverse melting in metallic alloys (4). There, the phenomenon is possible because the (low-temperature) amorphous phase gains energy against the crystal because of increased chemical short-range order (4). The very pronounced anisotropy of the forces involved in the present strongly anisotropic (quasi-two-dimensional) case, where the ratio of the respective energy scales is extremely crucial, requires a different theoretical description.

The reduction of the bonding strength is most probably accompanied by an increase of the vertical bond lengths, perhaps induced by the excitation of vertical molecular vibrations (20). Consequently, the lateral interaction between the molecules gains more weight and induces the lateral ordering (16).

References and Notes

1. N. Schupper, N. M. Shnerb, Phys. Rev. E Stat. Nonlin. Soft Matter Phys. 72, 046107 (2005).

2. A. L. Greer, Nature 404, 134 (2000).

At low temperature, the structural reorganization may be explained by different, energetically favorable adsorption sites of the NTCDA molecules if they are near the substrate, perhaps involving a distortion of the adsorbed molecule. If the molecules thereby differ slightly, such as in their rotational and/or translational alignment (the local bonding geometry), a statistic population of the respective sites would result in the observed loss of long-range order. The occurrence of both translational and rotational disorder upon cooling may

3. P. M. Tedrow, D. M. Lee, Phys. Rev. 181, 399 (1969).

4. W. Sinkler, C. Michaelsen, R. Bormann, D. Spilsbury, N. Cowlam, Phys. Rev. B 55, 2874 (1997).

5. H. Y. Bai, C. Michaelsen, R. Bormann, Phys. Rev. B 56, R11361 (1997).

6. M. R. Feeney, P. G. Debenedetti, F. H. Stillinger, J. Chem. Phys. 119, 4582 (2003).

7. U. Stahl, D. Gador, A. Soukopp, R. Fink, E. Umbach, Surf. Sci. 414, 423 (1998).

8. L. Kilian, thesis, Universität Würzburg, Würzburg, Germany (2002).

9. D. Gador et al., J. Electron Spectrosc. Relat. Phenom. 101–103, 523 (1999).

Biocatalytic Asymmetric Synthesis of Chiral Amines from Ketones Applied to Sitagliptin Manufacture

Christopher K. Savile,¹ * Jacob M. Janey,² * Emily C. Mundorff,¹ Jeffrey C. Moore,² Sarena Tam,¹ William R. Jarvis,¹ Jeffrey C. Colbeck,¹ Anke Krebber,¹ Fred J. Fleitz,² Jos Brands,² Paul N. Devine,² Gjalt W. Huisman,¹ Gregory J. Hughes²

Pharmaceutical synthesis can benefit greatly from the selectivity gains associated with enzymatic catalysis. Here, we report an efficient biocatalytic process to replace a recently implemented rhodium-catalyzed asymmetric enamine hydrogenation for the large-scale manufacture of the antidiabetic compound sitagliptin. Starting from an enzyme that had the catalytic machinery to perform the desired chemistry but lacked any activity toward the prositagliptin ketone, we applied a substrate walking, modeling, and mutation approach to create a transaminase with marginal activity for the synthesis of the chiral amine; this variant was then further engineered via directed evolution for practical application in a manufacturing setting. The resultant biocatalysts showed broad applicability toward the synthesis of chiral amines that previously were accessible only via resolution. This work underscores the maturation of biocatalysis to enable efficient, economical, and environmentally benign processes for the manufacture of pharmaceuticals.

insufficient stereoselectivity (requiring further upgrading of the product), are still being sought. Even though biocatalysis is rarely constrained by these shortcomings, enzymes may suffer from other limitations, such as low turnover numbers, instability toward the demanding conditions of chemical processes, and postreaction processing issues.

symmetric hydrogenation technologies are being increasingly used for commercial-

A

scale manufacture of fine chemicals and pharmaceuticals (1). Solutions that address remaining shortcomings of such technologies, including the need for high-pressure hydrogen, the use and subsequent removal of precious and toxic transition metals, the oftentimes lengthy process of ligand screening and synthesis, and the generally

The current synthesis of sitagliptin (2, 3) involves asymmetric hydrogenation of an enamine

10. C. Stadler, S. Hansen, A. Schöll, C. Kumpf, E. Umbach, N. J. Phys. 9, 50 (2007).

11. J. Stöhr, NEXAFS Spectroscopy, R. Gomer, Ed. (Spinger Series in Surface Science, Springer, Berlin, 1992), vol. 25.

12. A. Schöll et al., Phys. Rev. Lett. 93, 146406 (2004).

13. Y. Zou et al., Surf. Sci. 600, 1240 (2006).

14. A. Bendounan et al., Surf. Sci. 601, 4013 (2007).

15. A. Schöll, Y. Zou, T. Schmidt, R. Fink, E. Umbach, J. Phys. Chem. B 108, 14741 (2004).

16. L. Kilian et al., Phys. Rev. Lett. 100, 136103 (2008).

17. F. S. Tautz, Prog. Surf. Sci. 82, 479 (2007).

18. J. Ziroff, P. Gold, A. Bendounan, F. Forster, F. Reinert, Surf. Sci. 603, 354 (2009).

19. N. Schupper, N. M. Shnerb, Phys. Rev. Lett. 93, 037202 (2004).

20. O. Skibbe et al., J. Chem. Phys. 131, 024701 (2009).

21. Fruitful discussions with M. Sokolowski and Th. Schmidt are gratefully acknowledged. This work was funded by the German Bundesministerium für Bildung und Forschung (contracts 05 KS1WWA-5 and 05 KS7WE1) and by the Deutsche Forschungsgemeinschaft (contract DFG Re 1469/3-2). Experimental support by the BESSY (Berlin Electron Storage Ring Society for Synchrotron Radiation) staff during beam time is gratefully acknowledged.

Supporting Online Material

www.sciencemag.org/cgi/content/full/329/5989/303/DC1 Methods SOM Text Figs. S1 and S2 References

4 March 2010; accepted 8 June 2010 10.1126/science.1189106

at high pressure [250 pounds per square inch (psi)] using a rhodium-based chiral catalyst (Fig. 1A) (4). Within the manufacturing scheme, the chemistry suffers from inadequate stereoselectivity and a product stream contaminated with rhodium, necessitating additional purification steps at the expense of yield to upgrade both enantiomeric excess (e.e.) and chemical purity. By using a transaminase (5–8) scaffold and various protein engineering technologies, we have developed a catalyst and process that substantially improve the efficiency of sitagliptin manufacturing (Fig. 1B).

Transaminases have a limited substrate range, most accepting only substrates with a substituent no larger than a methyl group at the position adjacent to the ketone (Fig. 2A) (9–14). Not surprisingly, screening a variety of commercially available transaminases provided no enzyme with detectable activity for amination of the prositagliptin ketone (Fig. 2B). We therefore applied a combination of in silico design and directed evolution in an effort to confer such activity.

An (R)-selective transaminase [ATA-117 (15), a homolog of an enzyme from Arthrobacter sp.] (16) was previously used for R-specific transamination of methyl ketones and small cyclic ketones

1Codexis, Incorporated, 200 Penobscot Drive, Redwood City, CA 94063, USA.² Department of Process Research, MerckResearch Laboratories, Merck and Company, Incorporated, Rahway, NJ 07065, USA.

*To whom correspondence should be addressed. E-mail: christopher.savile@codexis.com (C.K.S.); jacob_janey@ merck.com (J.M.J.)

305

www.sciencemag.org SCIENCE VOL 329 16 JULY 2010

REPORTS

(11, 12). To assess the feasibility of developing an enzyme for sitagliptin synthesis, we generated a structural homology model of ATA-117 (17) to develop hypotheses for initial library designs. Docking studies using this model suggested that the enzyme would be unable to bind prositagliptin ketone (Fig. 2B) because of steric interference in the small binding pocket and potentially undesired interactions in the large binding pocket (Fig. 2C). By using a substrate walking approach (18) with a truncated substrate (Fig. 2D), we first engineered the large binding pocket of the enzyme and then evolved that enzyme for activity toward prositagliptin ketone.

Consistent with the model, ATA-117 was poorly active on the truncated methyl ketone analog (Fig. 2D), giving 4% conversion at 2 g/l substrate loading (table S2). Site saturation libraries of residues lining the large pocket of the active site provided new variants with increased activity toward the methyl ketone analog. The best variant contained a S²²³ → P²²³ [S223P (19)] mutation and showed an 11-fold activity improvement (Fig. 3 and table S1). On the basis of this improved variant (ATA-117: S223P), we generated a small library of enzyme variants for potential activity on prositagliptin ketone. Analysis of the enzyme model suggested four residues that could potentially interact with the trifluorophenyl group [V69, F122, T283, and A284 (19)]. Each of these positions was individually subjected to saturation mutagenesis and also included in a combinatorial library that evaluated several residues at each position on the basis of structural considerations [V⁶⁹ → G⁶⁹ andV⁶⁹ → A⁶⁹ (V69GA),F122AVLIG, T283GAS, and A284GF; library size of 216 variants]. A variant containing four mutations, three in the small binding pocket and one in the large pocket, provided the first detectable transaminase activity on prositagliptin ketone (Fig. 3 and table S3). No detectable activity was identified in any of the variants from the single amino acid site saturation libraries. Initial activity was accomplished via an F122I, V, or L mutation in combination with V69G or A284G. Docking studies indicated that these mutations may relieve the steric interference in the small binding pocket (Fig. 2E). Enzyme loading of 10 g/l provided 0.7% conversion of 2 g/l of ketone over 24 hours, corresponding to an estimated turnover of 0.1 per day. Screening the same combinatorial library in the ATA-117 context without the S223P large binding pocket mutation did not provide any variant with detectible activity toward prositagliptin ketone. Having attained activity through computer-aided catalyst design, we started evolving an enzyme variant for a practical, large-scale process.

The variant with the highest activity toward prositagliptin ketone from round 1b was chosen as the parent for the second round of evolution, and all the beneficial mutations from both the small-pocket combinatorial library and the largepocket saturation mutagenesis libraries were combined into a new library. Screening of this library resulted in a variant with 75-fold increased

Fig. 1. (A) The current synthesis of sitagliptin involves enamine formation followed by asymmetric hydrogenation at high pressure (250 psi) using a rhodium-based chiral catalyst, providing sitagliptin in 97% e.e., with trace amounts of rhodium. Recrystallization to upgrade e.e. followed by phosphate salt formation provides sitagliptin phosphate. (B) Our biocatalytic route features direct amination of prositagliptin ketone to provide enantiopure sitagliptin, followed by phosphate salt formation to provide sitagliptin phosphate.

Fig. 2. Previous substrate range studies suggested that the active site of transaminase consists of large (L) and small (S, typically limited to substituents about the size of a methyl group) binding pockets as mapped on the structure of acetophenone (A). Accordingly, the structure of prositagliptin ketone (B) can be mapped on these binding pockets and docked into the active site of the homology model (C). A prositagliptin ketone analog (D) was designed to fit the large pocket for initial optimization of this part of the active site. After initial engineering of the large pocket, an enzyme variant was generated with activity on the desired substrate (E) by excavating the small pocket (gray/blue, transaminase homology model; orange, large binding pocket; turquoise, small binding pocket; green, PLP and catalytic residues).

306

16 JULY 2010 VOL 329 SCIENCE www.sciencemag.org