Development of a pratical synthisis of DPP IV inhibitor LY2497282

Article abstract of DOI:10.1021/op700235c

A new synthetic route to LY2497282 (1), a potent and selective DPP IV inhibitor for the potential treatment of diabetes, suitable for the preparation of multikilogram quantities is described. The key step involved a stereoselective addition of the dianion of nicotinamide 8 to iV-dibenzyl-protected α-amino aldehyde 12, which was derived from iV-acetyl-protected amino ester 14 without epimerization. The desired Felkin-Anh nonchelation controlled anti-amino alcohol 11 was isolated with >99% HPLC area and > 99% ee by crystallization. After removing the dibenzyl protecting group under transfer hydrogenation conditions, LY2497282 (1) was finally obtained in 39% overall yield with a six-step longest linear sequence starting from N-acetyl-protected amino ester 14.

Full text of DOI:10.1021/op700235c

Organic Process Research & Development 2008, 12, 218–225
Development of a Practical Synthesis of DPP IV Inhibitor LY2497282†
Hannah Yu,* Rachel N. Richey, James R. Stout, Mark A. LaPack, Ruilin Gu, Vien V. Khau, Scott A. Frank, Joel P. Ott,
Richard D. Miller, Michael A. Carr, and Tony Y. Zhang
Chemical Product Research and DeVelopment, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana,
46285-4813, U.S.A.
Abstract:
type 2 diabetes.3 We detail here the development of our
synthetic route for the preparation of 1. The key step involved
a stereoselective addition of the dianion of nicotinamide 8 to
N-dibenzyl protected R-amino aldehyde 12, which was derived
from N-acetyl protected amino ester 14 without epimerization.
The desired Felkin-Anh nonchelation controlled anti amino
alcohol 11 was isolated with >99% HPLC area and >99% ee
by crystallization. After removing the dibenzyl protecting
group under a transfer hydrogenation condition, LY2497282
(1) was finally obtained in 39% overall yield with a six-
step longest linear sequence starting from N-acetyl-
protected amino ester 14.
A new synthetic route to LY2497282 (1), a potent and selective
DPP IV inhibitor for the potential treatment of diabetes, suitable
for the preparation of multikilogram quantities is described. The
key step involved a stereoselective addition of the dianion of
nicotinamide 8 to N-dibenzyl-protected r-amino aldehyde 12,
which was derived from N-acetyl-protected amino ester 14 without
epimerization. The desired Felkin-Anh nonchelation controlled
anti-amino alcohol 11 was isolated with >99% HPLC area and
>99% ee by crystallization. After removing the dibenzyl protecting
group under transfer hydrogenation conditions, LY2497282 (1)
was finally obtained in 39% overall yield with a six-step longest
linear sequence starting from N-acetyl-protected amino ester 14.
Introduction
Type 2 diabetes (formerly known as non-insulin-dependent
diabetes mellitus) is a severe and increasingly prevalent disease.
Inhibition of GLP-1 degradation by dipeptidyl peptidase IV
(DPP-IV)1 has emerged as a promising approach for the
treatment of type 2 diabetes and has been aggressively pursued
by numerous pharmaceutical companies.2
Results and Discussion
The original route used presents a very attractive pathway
that is based on a key ketone intermediate 9a (Scheme 1).3 The
synthesis started by coupling commercially available aldehyde
2 and N-Boc phosphonate 3 to give enamine intermediate 4.
Asymmetric hydrogenation of enamine 4 afforded N-Boc
protected amino ester 5 with >99% ee. N-Boc protected amino
ester 5 was then converted to Weinreb amide 7a, which was
coupled with the dianion of 8 to give ketone intermediate 9a.
Reduction of ketone afforded alcohol 10a, which was subse-
quently deprotected to give the amino alcohol LY2497282 (1).
The Scheme 1 synthesis of amino alcohol 1 also represented
a few issues: (1) the expense of preparing Boc-protected amino
ester 5 was high; (2) more than 2 equiv of 8 was needed to
subside the deprotonation of NH in 7a and the extra 8 was
difficult to remove; (3) epimerization occurred in the coupling
reaction between nicotinamide 8 and Weinreb amide 7a, which
resulted in chiral chromatography at 10a stage to improve ee;
(4) low diastereoselectivity (2:1) was obtained in the reduction
of 9a, and the two resulting diastereomers were difficult to
separate.
LY2497282 (1) was identified at Eli Lilly and Company as
a potent and selective DPP IV inhibitor for the treatment of
† We dedicate this paper to the memory of Dr. Chris Schmid, a colleague,
mentor, and friend, who passed away December 26, 2007.
* To
whom
correspondence
should
be
addressed.
E-mail:
yu_hannah@lilly.com.
(1) (a) Weber, A. J. Med. Chem. 2004, 47, 4135. (b) Drucker, D. J.
Exp.Opin. InVest. Drugs 2003, 12, 87. (c) Wiedeman, P. E.; Trevillyan,
J. M. Curr. Opin. InVest. Drugs (Thompson Sci.) 2003, 4, 412.
(2) (a) Sheehan, S. M.; Mest, H.; Watson, B. M.; Klimkowski, V. J.;
Timm, D. E.; Cauvin, A.; Parsons, S. H.; Shi, Q.; Canada, E. J.; Wiley,
M. R.; Ruehter, G.; Evers, B.; Petersen, S.; Blaszczak, L. C.; Pulley,
S. R.; Margolis, B. J.; Wishart, G. N.; Renson, B.; Hankotius, D.;
Mohr, M.; Zechel, J.; Kalbfleisch, J. M.; Dingess-Hammond, E. A.;
Boelke, A.; Weichert, A. G. Bioorg. Med. Chem. Lett. 2007, 17, 1765.
(b) Kowalchick, J. E.; Leiting, B.; Pryor, K. D.; Marsilio, F.; Wu,
J. K.; He, H.; Lyons, K. A.; Eiermann, G. J.; Petrov, A.; Scapin, G.;
Patel, R. A.; Thornberry, N. A.; Weber, A. E.; Kim, D. Bioorg. Med.
Chem. Lett. 2007, 17, 5934. (c) Luebbers, T.; Boehringer, M.; Gobbi,
L.; Hennig, M.; Hunziker, D.; Kuhn, B.; Loeffler, B.; Mattei, P.;
Narquizian, R.; Peters, J.; Ruff, Y.; Wessel, H. P.; Wyss, P. Bioorg.
Med. Chem. Lett. 2007, 17, 2966. (d) Feng, J.; Zhang, Z.; Wallace,
M. B.; Stafford, J. A.; Kaldor, S. W.; Kassel, D. B.; Navre, M.; Shi,
L.; Skene, R. J.; Asakawa, T.; Takeuchi, K.; Xu, R.; Webb, D. R.;
Gwaltney, S. L. J. Med. Chem. 2007, 50, 2297. (e) Yoshida, T.;
Sakashita, H.; Akahoshi, F.; Hayashi, Y. Bioorg. Med. Chem. Lett.
2007, 17, 2618. (f) Pei, Z.; Li, X.; Von, G.; Thomas, W; Longeneck,
K.; Pireh, D.; Stewart, K. D.; Backes, B. J.; Lai, C.; Lubben, T. H.;
Ballaron, S. J.; Beno, D. W. A.; Kempf-Grote, A. J.; Sham, H. L.;
Trevillyan, J. M. J. Med. Chem. 2007, 50, 1983. (g) Tran, T.; Quan,
C.; Edosada, C. Y.; Mayeda, M.; Wiesmann, C.; Sutherlin, D.; Wolf,
B. B. Bioorg. Med. Chem. Lett. 2007, 17, 1438.
Despite the issues, we felt that the ketone approach described
in Scheme 1 still represented a very attractive and potentially
scaleable approach to amino alcohol 1. Our initial attempt was
to improve the route by developing alternative reaction and work
up conditions to make it amenable for scale up.
(3) Blaszczak, L. C.; Mathes, B. M.; Pulley, S. R.; Robertson, M. A.;
Sheehan, S. M.; Shi, Q.; Watson, B. M.; Wiley, M. R. WO 2007/
015767 A1, 2007.
218
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development
10.1021/op700235c CCC: $40.75
2008 American Chemical Society
Published on Web 03/04/2008

Scheme 1. Original synthesis of LY2497282 (1)
Table 1. Stereoselective reduction of ketones 9a and 9b
yield (10+10′)
entry
R
reagent
solvent
T (°C)
anti:syn (10:10′)
(%)
1
2
3
4
5
6
Cbz (9b)
Boc (9a)
Boc (9a)
Boc (9a)
Boc (9a)
Cbz (9b)
Al(OiPr)3
IPA
50
0
no reaction
2.5:1
2.3:1
1:1
0
91
84
81
88
86
NaBH4
EtOH
EtOH
THF
NaB(OCH3)3H
LiAlH(OtBu)3
LiAlH(OtBu)3
LiAlH(OtBu)3
0
-78
-78
-78
EtOH
EtOH
8:1
15:1
The development of synthetic methods for the preparation
of anti-1,2-amino alcohols through the reduction of N-protected
amino ketones has received significant attention in recent years.4
Carbamate groups are ubiquitous protecting groups for amines.
Stereocontrol in the reduction of carbamate-protected amino
ketones would appear to favor the formation of chelation
controlled anti-amino alcohols since both the ketone oxygen
and the carbamate nitrogen are sterically accessible for coor-
dination to a Lewis acid.5 To improve the Scheme 1 route, our
first objective was to address the poor diastereoselectivity issue
observed in the ketone 9a reduction step. We started our
screening studies using racemic ketone starting materials 9a and
9b, with N-Boc and N-Cbz being the protecting groups,
respectively (Table 1). The initial attempt was to reduce 9b
with Al(OiPr)3 in refluxing IPA, the classical Meerwein-
Pondorf-Verley reduction condition which had been reported
to afford excellent diastereoselectivity in reducing N-carbamate
protected R-amino ketones.4c–e Presumably due to the enone
characteristic of ketone starting material 9b,6 no reaction was
observed under these conditions (Table 1, entry 1). As previ-
ously observed, an anti/syn (10:10′) ratio of 2.5:1.0 was
observed when Boc-protected amino ketone 9a was reduced
with NaBH4 in EtOH at 0 °C (entry 2). A similar ratio of 2.3:
1.0 was obtained when NaB(OCH3)3H was used as the reducing
agent (entry 3). LiAlH(OtBu)3 in EtOH at -78 °C5 proved to
afford the best diastereoselectivity (entries 5 and 6). Interest-
ingly, Cbz-protected ketone 9a gave much better selectivity than
Boc-protected ketone 9b (15:1 vs 8:1) when the reduction was
carried out with LiAlH(OtBu)3 in EtOH. Therefore, Cbz-
protected ketone 9b was identified as the best substrate and
LiAlH(OtBu)3 in EtOH at -78 °C was identified as the best
condition for the reduction (entry 6).
(4) (a) Buckley, T. F.; Rapoport, H. J. Am. Chem. Soc. 1981, 103, 6157.
(b) Fujita, M.; Hiyama, T. J. Am. Chem. Soc. 1984, 106, 4629. (c)
Satoh, H.; Turk, G. H. Eur. Pat. Appl. EP934923A1, 1999. (d) Urban,
F. J.; Jasys, V. J. Org. Process Res. DeV. 2004, 8, 169. (e) Yin, J.;
Huffman, M. A.; Conrad, K. M.; Armstrong, J. D. J. Org. Chem. 2006,
71, 840.
With acceptable diastereoselectivity results in hand, we next
turned our attention to improving the coupling reaction between
Weinreb amide 7b and nicotinamide 8. We started our
(5) (a) Hoffman, R. V.; Maslouh, N.; Cervantes-Lee, F. J. Org. Chem.
2002, 67, 1045. (b) Vabeno, J.; Brisander, M.; Lejon, T.; Luthman,
K. J. Org. Chem. 2002, 67, 9186.
(6) Ketones 9a and 9b were observed as a mixture of ketone and enone
forms in NMR.
Vol. 12, No. 2, 2008 / Organic Process Research & Development
•
219

Scheme 2. Coupling reaction between Weinreb amide 7b
nucleophilic addition of dianion-8 to an appropriately protected
R-amino aldehyde.10 Boc, Cbz and dibenzyl protecting groups11,12
had been used as the most common protecting groups for related
R-amino aldehydes. The stereoselectivity of the addition was
strongly dependent on the nature of the substrates and reaction
conditions. Since N-dibenzyl-protected R-amino aldehydes often
led to improved selectivity towards the desired nonchelation-
controlled anti product, we decided to investigate the aldehyde
route using dibenzyl-protected aldehyde 12. The aldehyde 12
could be derived from N-acetyl-protected amino ester 14, an
and nicotinamide 8
investigation with two major development objectives: (1)
improve the reaction conditions to avoid low yield and reduce
the amount of nicotinamide 8 used and (2) suppress the
epimerization occurring during the reaction workup.
(13) (a) To fund the lab development of the subsequent steps, we also
developed the synthesis of amino ester 14.
Because the deprotonation of the exchangeable amino proton
of 7b was faster than the nucleophilic attack of dianion-8 at
the Weinreb amide, more than 2 equiv of dianion-8 was needed.
When only 1.3 equiv of dianion-8 was used, almost no product
9b was observed. To avoid the use of excess of 8, we therefore
adopted a predeprotonation protocol by first deprotonating the
amino group of 7b with 0.95 equiv of i-PrMgCl/THF,7 followed
by addition of 1.25 equiv of dianion-8.8 As a result, ketone 9b
was obtained in ∼70% yield without the use of excess
nicotinamide starting material (Scheme 2).
We next attempted to suppress the epimerization which was
reported to occur during the workup of the reaction. It was
precedent in the literature that N-carbamate protected R-amino
ketones could be obtained by coupling N-protected R-amino
Weinreb amides and metal nucleophiles with retention of
stereochemistry.9 We then examined multiple reaction and
workup conditions starting from enantiomerically pure Weinreb
amide 7b. To our disappointment, epimerization was observed
under a variety of conditions. The ee of isolated ketone 9b
ranged from 0% to 80%, with lower ee observed when the
reaction was performed at larger scale (Scheme 2). We
speculated that hydrogen bonding between the ketone oxygen
and pyridine nitrogen in 9b might make the compound more
sensitive to epimerization.
The synthesis involved a classical Erlenmeyer condensation of 2,4-
difluorobenzyaldehyde with N-acetyl glycine, followed by azlactone
ring-opening to afford the corresponding dehydroamino ester. Amino
ester 14 was further obtained from the dehydramino ester in >99%
ee through enzyme resolution or asymmetric hydrogenation approach.
For references on the synthesis of amino acids and derivatives, see:
Roper, J. M.; Bauer, D. P. Synthesis 1983, 1041. (b) Duthaler, R. O.
Tetrahedron 1994, 50, 1539. (c) Boaz, N. W.; Large, S. E.; Ponasik,
J. A.; Moore, M. K.; Barnette, T.; Nottingham, W. D. Org.Process
Res. & DeV. 2005, 9, 472. (d) Perdih, A.; Sollner Dolenc, M. Curr.Org.
Chem. 2007, 11, 801. (e) Cativiela, C.; Diaz-de-Villegas, M. D.
Tetrahedron: Asymmetry 2007, 18, 569. (f) Humphrey, C. E.; Furegati,
M.; Laumen, K.; Vecchia, L. L.; Leutert, T.; Muller-Hartwieg, J. C. D.;
Vogtle, M. Org.Process Res. DeV. 2007, 11, 1069.
Because of the difficulties we encountered in preserving the
chiral integrity in the ketone formation step, we decided to
investigate alternative approaches for the synthesis of LY2497282
(1). An alternative strategy for the preparation of 1 would be
(14) (a) Lahm, G. P.; Selby, T. P.; Freudenberger, J. H.; Stevenson, T. M.;
Myers, B. J.; Seburyamo, G.; Smith, B. K.; Flexner, L.; Clark, C. E.;
Cordova, D. Bioorg. Med. Chem. Lett. 2005, 15, 4898. (b) Okada, E.;
Kinomura, T; Higashiyama, Y.; Takeuchi, H.; Hojo, M. Heterocycles
1997, 46, 129.
(7) A similar predeprotonation protocol has been reported in the synthesis
of R-(N-Boc-amino)ketones from R-(N-Boc-amino)Weinreb amide: Liu,
J.; Ikemoto, N.; Petrillo, D.; Armstrong, J. D Tetrahedron Lett. 2002,
43, 8223.
(15) Liu, C.; Ng, J. S.; Behling, J. R.; Yen, C. H.; Campbell, A. L.; Fuzail,
K. S.; Yonan, E. E.; Mehrotra, D. V. Org. Process Res. DeV. 1997, 1,
45.
(8) The preparation of dianion-8 was described in the subsequent text in
the Scheme 4 approach.
(16) The ee of the aldehyde 12 was determined at alcohol 13 by reducing
an analytical amount of 12 to 13 with NaBH4 in MeOH. See
Experimental Section for the ee methods for alcohol 13.
(17) Lower yield was due to the recovery of aldehyde starting material 12,
which may be caused by the aldehyde enolization during the dianion
8 addition. The hypothesis was that the extra diisopropylamine might
help suppress the enolization of the aldehyde.
(9) (a) Kaldor, S. W.; Hammond, M.; Dressman, B. A.; Fritz, J. E.;
Crowell, T. A.; Hermann, R. A Bioorg. Med. Chem. Lett. 1994, 4,
1385. (b) Bohnstedt, A. C.; Prasad, J. V. N. V.; Rich, D. H.
Tetrahedron Lett. 1993, 34, 5217. (c) Seki, M.; Matsumoto, K.
Tetrahedron Lett. 1996, 37, 3165.
(10) Jurczak, J.; Golebiowski, A. Chem. ReV. 1989, 89, 149.
(11) Kempf, D. J. J. Org. Chem. 1986, 51, 3921.
(18) (a) Seebach, D. Angew. Chem., Int. Ed. Engl. 1988, 27, 1624. (b) Sun,
X.; Collum, D. B. J. Am. Chem. Soc. 2000, 122, 2459. (c) McNeil,
A. J.; Collum, D. B. J. Am. Chem. Soc. 2005, 127, 5655. (d) Xu, F.;
Reamer, R. A.; Tillyer, R.; Cummins, J. M.; Grabowski, E. J. J.;
Reider, P. J.; Collum, D. B.; Huffman, J. C. J. Am. Chem. Soc. 2000,
122, 11212.
(12) (a) Reetz, M. T.; Drewes, M. W.; Schmitz, A. Angew. Chem., Int.
Ed. Engl. 1987, 26, 1141. (b) Mikami, K.; Kaneko, M.; Loh, T.;
Masahiro, T.; Nakai, T. Tetrahedron Lett. 1990, 31, 3909. (c) Reetz,
M. T. Angew. Chem., Int. Ed. Engl. 1991, 30, 1531. (d) Diederich,
A. M.; Ryckman, D. M. Tetrahedron Lett. 1993, 34, 6169. (e)
DeCamp, A. E.; Kawaguchi, A. T.; Volante, R. P.; Shinkai, I.
Tetrahedron Lett. 1991, 32, 1867. (f) Barluenga, J.; Baragana, B.;
Concellon, J. M. J. Org. Chem. 1995, 60, 6696. (g) Beaulieu, P. L.;
Wrnic, D.; Jean-Simon, D.; Yvan, G. Tetrahedron Lett. 1995, 36, 3317.
(19) The FTIR instrument is a Mettler-Toledo ReactIR-4000 equippe with
a DiComp diamond sensor, a liquid nitrogen cooled MCT detector.
Software version ReactIR 3.0 is used.
(20) Ng, J. S.; Przybyla, C. A; Liu, C.; Yen, J. C.; Muellner, F. W.; Weyker,
C. L. Tetrahedron 1995, 51, 6397.
220
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development

Scheme 3. Retrosynthetic analysis of LY2497282 (1) using aldehyde approach
Scheme 4. Synthesis of LY2497282 (1) from amino ester 14
amino ester derivative available through an asymmetric hydro-
genation or enzyme resolution approach (Scheme 3).13
To protect free amine 15 with the dibenzyl protecting group,
acetonitrile was first used as the reaction solvent and potassium
carbonate as the base. The reaction would not go to completion
unless a large excess of benzyl bromide was used. Addition of
the phase transfer reagent Bu4NBr promoted the formation of
∼5% of benzyl alcohol as the hydrolysis product of benzyl
bromide. To facilitate the solubility of potassium carbonate,
water was next used as the cosolvent. Under acetonitrile/water
(10:1), the reaction proceeded to completion with 2.1 equiv of
benzyl bromide at 78 °C overnight. Upon completion of the
reaction, aqueous work up was applied to remove residual
potassium carbonate, potassium bromide and potassium bicar-
bonate. The next step was to reduce ester 16 to alcohol 13 with
1 N LiAlH4in THF. To telescope the steps, 2-methyltetrahy-
drofuran was chosen as the reaction solvent for the LiAlH4
reduction. Following an aqueous work up at the end of the
reaction, alcohol 13 was directly solvent exchanged into heptane,
the final crystallizing solvent for 13. The residual benzyl
bromide, a mutagen used in the benzyl protection step, was
also completely consumed during the LiAlH4 reduction process.
The three-step telescoped procedure afforded alcohol 13 with
Since bulk suppliers of nicotinamide 814 and N-acetyl-
protected amino ester 14 had been identified, they were chosen
as the starting materials for our work. The conversion of amino
ester 14 to final compound LY2497282 (1) was accomplished
in six steps, as depicted in Scheme 4. The synthesis started from
the deprotection of N-acetyl protected amino ester 14. The acetyl
deprotection was first attempted with anhydrous HCl in reflux-
ing MeOH, however a significant amount of ester hydrolysis
was observed under the condition. The deprotection was much
more effective when 14 was refluxed under concentrated H2SO4
in MeOH, affording amine 15 as its H2SO4 salt with >97% in
situ yield. Since neither the H2SO4 salt nor the free base was a
crystalline solid, amine 15 was used in the next step without
isolation. Thus, upon completion of the reaction, water and
MTBE were added to the reaction mixture to promote layer
separation. The aqueous layer, with the desired H2SO4 salt of
15 in it, was basified with ammonium hydroxide to give free
base 15 in MTBE. During the pilot-plant operation, the organic
extracts were solvent exchanged to acetonitrile, one of the
reaction solvents for the subsequent reaction.
Vol. 12, No. 2, 2008 / Organic Process Research & Development
•
221

Scheme 5. FTIR absorbance profiles generated during the formation of dilithium-8 and addition of aldehyde 12 to dilithium-8
>99% ee and 75% yield from methyl ester 14. The process
was carried out at 10 kg scale during the pilot-plant operation.
Following a modification of a literature procedure,15 alcohol
13 was converted to aldehyde 12 using sulfur trioxide pyridine
complex as the oxidation reagent and DMSO as the reaction
solvent. The highly volatile dimethyl sulfide byproduct was
produced in this reaction. To control the dimethyl sulfide
emission, the off gas was scrubbed with 5% NaOCl during scale
up. Initially, ethyl acetate was used to extract the product from
the aqueous solution. Since the residual ethyl acetate was
difficult to remove and would interfere with the next reaction,
it was later replaced with MTBE. Removal of residual DMSO
through 5% citric acid washes was also more efficient when
MTBE was used as the extraction solvent.
We collected chiral stability data on aldehyde 12, both as a
solid and as a MTBE solution, because we were concerned
about our ability to control thermal racemization on scale.
Starting with a representative lot of aldehyde 12 with 99.5%
ee, the enantiomeric purity was monitored over time under
different conditions. The results indicated that little racemization
occurred if aldehyde 12 was stored as a solid at room
temperature for 24 h. In 20 volumes of MTBE, the enantiomeric
excess of 12 would stay at >98% for more than 24 h if the
temperature was controlled below 35 °C. During the scale up
of aldehyde 12, the reaction and solvent distillation temperature
was controlled below 15 °C. The reaction proceeded smoothly
and afforded aldehyde 12 in >95% yield and >99% ee.16
The preparation of amino alcohol 11 from aldehyde 12 was
the key step of the entire synthetic sequence. The reaction
involved two sequential steps. The first step was to deprotonate
8 to form the dianion of 8. The second step was to couple
dianion-8 with aldehyde 12. A NaH/LDA combination proce-
dure was first developed to form the dianion of 8. Thus, the
NH proton of 8 was first deprotonated with 1.0 equiv of NaH;
subsequently the desired C-H proton was deprotonated with
LDA, which was preformed under 1.3 equiv of diisopropyl-
amine and 1.2 equiv of n-BuLi in THF. Higher product yield
was observed when diisopropylamine was used in slight excess
of n-BuLi.17 The dianion-8 was then reacted with 1.0 equiv of
aldehyde 12 at –20 °C to give the two amino alcohols in ∼3:1
ratio in favor of the desired Felkin-Anh nonchelation controlled
anti-amino alcohol 11. There was no significant difference in
the diastereoselectivity if the reaction was carried out at lower
temperature. The rejection of wrong diastereomer through
heptane/THF crystallization turned out to be extremely efficient.
Despite the moderate diastereoselectivity, the desired amino
alcohol 11 was isolated in 99.4% HPLC area percentage in 62%
yield after one crystallization, with the undesired amino alcohol
consistently controlled below 0.1%.
Because of the long-term liability of using NaH, we also
developed an alternative procedure by using LDA as the only
deprotonating reagent. Thus, addition of a THF solution of
nicotinamide 8 to LDA, which was formed in situ by adding
2.2 equiv of n-BuLi to 2.4 equiv of diisopropylamine in THF,
afforded dilithium-8 as a red solution. One equivalent of
aldehyde was then introduced to the dilithium-8 solution at -20
°C to give the two diastereomers in 2.6:1.0 ratio. Under this
condition, the desired diastereomer was isolated in 56% yield
with a impurity profile comparable with that of the one isolated
from the NaH/LDA procedure.
Initial results indicated that THF was the best reaction solvent
for the LDA-promoted dilithium-coupling reaction. We there-
fore further investigated the reaction using THF as the solvent.
One observation we made during the development was that the
reaction was very sensitive to THF concentrations. At -20 °C,
no reaction occurred at 0.05 M THF; better aldehyde conversion
was observed if the reaction was preformed at 0.4 M rather
than 0.2 M. It was well documented that several aggregation
states could be present in organolithium dianion reactions and
the mixed aggregates might influence reactivity.18 In an attempt
to ensure reproducibility, we decided to monitor the reaction
using in situ FTIR19 to better understand the formation of the
dianion and the subsequent aldehyde addition step (Scheme 5).
LDA was formed under the standard condition by addition of
n-BuLi to diisopropylamine in THF at -20 °C. Upon addition
of nicotinamide 8 to LDA, the IR absorbance profiles showed
the consumption of LDA band (1304 cm-1) and production of
a band at 1555 cm-1, which was believed to correspond to
222
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development

Scheme 6. Structures of des\-fluorinated impurities
aggregate I for dilithium-8. Upon 50% of nicotinamide 8
addition, the IR profiles showed that the concentration of
dilithium-8 aggregate I reached its maximum. At the same time,
dilithium-8 aggregate II started to grow. The formation of
aggregate II was completed at the end of nicotinamide 8
addition, indicating the dianion formation was addition rate
controlled. Upon the addition of aldehyde 12 to dilithium-8,
the IR profiles showed the consumption of the 1543 cm-1 band
and the production of a product band at 1495 cm-1. Again, the
formation of product 11 was completed at the end of aldehyde
12 addition, indicating the coupling reaction was also addition
rate controlled. On the basis of the results of these experiments,
we were able to shorten the reaction time significantly with no
concerns of reproducibility due to an incomplete reaction. The
observation of the two dilithium-8 aggregation states by in situ
IR was intriguing. The details on these mixed aggregates and
how they influence reactivity will be described in a separate
report.
To complete the synthesis of LY2497282 (1), the last step
was to remove the dibenzyl protecting group of amino alcohol
11. The deprotection was first attempted under the classical
hydrogenation conditions using MeOH or EtOH as the solvent
and Pd/C or Pd(OH)2 as the catalyst. The deprotection under
these hydrogenation conditions gave highly inconsistent results.
For example, for one lot of starting material 11, the reaction
proceeded to completion when it was hydrogenated with 10%
water wet Pd/C in EtOH at 60 psi and 65 °C overnight. For
another lot of starting material 11, only 35% conversion was
observed when the same lot of catalyst and reaction conditions
were applied. When acetic acid was used as the solvent, the
reaction proceeded to completion. However, a significant
amount of impurity was formed as a result of reduction of the
pyridine ring in 11. Hydrogenation using Pearlman’s catalyst
(20% palladium hydroxide on carbon) in MeOH20 also produced
inconsistent results. Presumably, the inconsistency was a result
of catalyst poisoning caused by residual sulfur carried over from
the aldehyde oxidation step. Since the catalyst poisoning was
difficult to control, we next turned our attention to transfer
hydrogenation conditions for the dibenzyl deprotection. Under
the treatment of 10% Pd on carbon, ammonium formate in
MeOH and water (10:1) at 65 °C, the reaction consistently
proceeded to completion within 2 h to give the target compound
LY2497282 (1). However, 4% des-fluorinated impurities 17,
18 and 19 (Scheme 6) were formed under this condition. As
expected, the rejection of these impurities were inefficient under
a variety of crystallization conditions. To overcome this
problem, we further optimized our transfer hydrogenation
conditions by replacing MeOH with EtOH as the reaction
solvent and reducing the reaction temperature from 65 to 40
°C. The modifications consistently afforded LY2497282 (1) in
>97% assay yield and controlled the formation of the des-
fluorinated impurities below the desired level. Following transfer
hydrogenation, the catalyst was removed by filtration and the
filtrate went through an aqueous work up to remove the
inorganic salts. Final crystallization afforded LY2497282 (1)
in >99.0% purity and >99.5% ee. In addition, the Pd level
was determined to be <1 ppm.
Conclusion
In conclusion, we have developed an efficient and chroma-
tography-free route for the preparation of LY2497282 (1), a
potent and selective DPP IV inhibitor targeted for the treatment
of diabetes. The compound was prepared in 39% overall yield
in six steps from N-acetyl protected R-amino ester 14, a chiral
intermediate available through an asymmetric hydrogenation
or enzyme resolution approach. The second chiral center was
established by a stereoselective addition of nucleophile 8 to
N-dibenzyl protected R-amino aldehyde 12, which was derived
from 14 without epimerization. The reactions described were
successfully scaled up from 12 L to pilot-plant scale.
Experimental Section
General. Melting points were obtained using a Thomas-
1
Hoover capillary melting apparatus and are uncorrected. H
NMR and 13C NMR spectra were recorded as specified for each
experiment with chemical shift recorded as parts per million.
Elemental analysis and high-resolution mass spectrometry data
was provided by the Physical Chemistry group of Lilly Research
Laboratories. Commercially available reagents and solvents
were used without further purification. Reaction progress was
monitored using an Agilent 1100 series instrument equipped
with a UV using the following conditions: Mobile phase: (A)
0.1% TFA in water; (B) 0.1% TFA in CH3CN. Gradient: T )
0 min 60% A 40% B, T ) 20 min 30% A 70% B; flow rate:
1.0 mL/min; column temperature: 30 °C; column: Zorbax SB-
C8, 4.6 mm × 250 mm, 5 µm; detector: 254 nm. Chiral assays
were performed using an Agilent 1100 series HPLC equipped
with a UV.
2-Dibenzylamino-3-(2,5-difluoro-phenyl)-propan-1-ol (13).
N-acetyl-protected methyl ester 14 (2.0 kg, 7.78 mol) was
charged to 16 L of MeOH through a ventilated addition funnel.
To the resulting solution was added concentrated sulfuric acid
(1.9 kg, 19.4 mol), while maintaining the temperature below
60 °C. The reaction mixture was refluxed at 65 °C for 24 h,
and HPLC analysis indicated that <0.5% of methyl ester 14
remained. The reaction mixture was allowed to cool to rt and
was then quenched by slow addition of 19.2 L of water followed
by 18 L of MTBE. The layers were separated, and the aqueous
layer, which contained the salt of 15, was recharged to the
vessel. Eighteen liters of MTBE was added, and the two phases
Vol. 12, No. 2, 2008 / Organic Process Research & Development
•
223

were allowed to settle. The two phases were separated, and the
organic layer was discarded. An additional 18 L of MTBE was
added to the aqueous layer, and the pH was adjusted to
approximately 9.5 by slow addition of ammonium hydroxide
(2.1 kg). The aqueous layer was back extracted with 9 L of
MTBE. The organic layers were combined and were assayed
to contain 1.6 kg of amine 15 (7.4 mol, 95% yield from 14).
The organic portion was concentrated by atmospheric
distillation to a total volume of 8 L. To the still warm solution,
20 L of acetonitrile was added, and the mixture was concen-
trated to 8 L. To the still warm solution, 20 L of acetonitrile
was added again, and the mixture was concentrated to 18 L. A
quantity of 1.8 L of water was added, and the resulting
acetonitrile/water solution was charged with potassium carbon-
ate (2.6 kg, 18.5 mol) and benzyl bromide (2.66 kg, 15.5 mol).
The mixture was heated to 78 °C and stirred at 78 °C overnight.
The reaction mixture was cooled and then quenched with 16 L
of water followed by 16 L of EtOAc. The layers were separated,
and the organic layer was washed with 16 L of 10% brine
solution. The organic layer was assayed to contain 2.8 kg of
dibenzyl-protected methyl ester 16 (7.08 mol, 91% yield
from 14).
116.6, 116.5, 116.3, 114.6, 114.4, 114.3, 60.3, 59.7, 53.2, 25.2.
IR (KBr, cm–1) ν 3422, 3079, 3027, 2955, 2938, 2857. MS
(EI) m/z (rel intensity) 368 (M+, 100); HRMS (ES+) exact mass
calcd for C23H24F2NO 368.1820, found 368.1816. Anal. Calcd
for C23H23F2NO C, 75.19, H, 6.31, N, 3.81, found C, 75.11, H,
6.36, N, 4.01.
N-tert-Butyl-2-[3-dibenzylamino-4-(2,5-difluoro-phenyl)-
2-hydroxy-butyl]-6-trifluoromethyl-nicotinamide (11). To a
12-L three neck round-bottom flask equipped with an overhead
stirrer apparatus, a cooling bath, a thermometer/thermocouple,
a nitrogen inlet, and a 2-L addition funnel was charged alcohol
13 (510 g, 1.4 mol), triethylamine (568 g, 5.62 mol), and 100
mL of DMSO. The resulting solution was cooled in an ice/
acetone bath to -9 °C. A solution of sulfur trioxide pyridine
complex (443 g, 2.73 mol) in 1.96 L of DMSO was added to
the reaction mixture at such a rate as to maintain the temperature
below 10 °C. The reaction mixture was stirred at 0–10 °C for
an additional 30 min. Five liters of water was added over 1 h,
and the temperature was again controlled below 10 °C. The
cooling bath was removed, and 5 L of MTBE was added. After
the layer separation, the aqueous layer was extracted with
another 5 L of MTBE. The MTBE layers were combined and
washed with 3 × 5 L of aqueous citric acid (5%), followed by
4 L of brine, and dried over 756 g of Na2SO4. The solution
was filtered and concentrated to give aldehyde 12 (491 g, 1.3
mol, 96% yield) as a yellow semisolid. The crude mixture was
used in the next step without further purification.
Method A. To a 2-L round-bottom flask equipped with a
condenser, a nitrogen inlet, magnetic stirring, and a thermo-
couple was charged nicotinamide 8 (42.7 g, 0.164 mmol),
followed by 250 mL of THF. NaH (6.56 g, 0.164 mol, 60%
dispersion in mineral oil) was added to the nicotinamide solution
in portions. The reaction mixture was heated at 65 °C for 1 h,
and then it was allowed to cool to rt. To a separate 500-mL
round-bottom flask equipped with magnetic stirring, an addition
funnel, a nitrogen inlet, and a thermocouple was charged
diisopropylamine (21.6 g, 0.213 mol) and 60 mL of THF. The
solution was cooled to -30 °C in a dry ice/acetone bath. n-BuLi
(79.2 mL, 0.198 mol, 2.5 M solution in hexanes) was charged
to the addition funnel via syringe and was added to the flask
over ∼30 min. The addition funnel was rinsed with 10 mL of
THF. The sodium-8 solution above was cooled in a dry ice/
acetone bath to -30 °C, and the LDA solution was added to
the sodium-8 solution dropwise over 30 min while maintaining
the temperature below -20 °C. The resulting dark-red reaction
mixture was stirred for 30 min. A solution of aldehyde 12 (60.0
g, 0.164 mol) in 120 mL of THF was charged to the dark-red
dianion-8 solution dropwise while maintaining the maximum
temperature below -20 °C. The color stayed red throughout
the addition. After the completion of the addition, the reaction
mixture was stirred at -20 °C for an additional hour. Water
(400 mL) was added to the reaction mixture dropwise. The
cooling bath was removed, and the pH was adjusted to
approximately 9 by addition of 120 mL of 5 N HCl. The
mixture was transferred to a separatory funnel, and layers were
separated. The aqueous layer was extracted with 300 mL of
EtOAc. The combined organic layers were washed with 300
mL of water and then solvent switched to 1080 mL of heptane
The organic portion was concentrated by atmospheric
distillation to a volume of 10 L. To the still warm solution, 22
L of 2-methyltetrahydrofuran was added, and the mixture was
concentrated to 10 L. To the still warm solution, an additional
22 L of 2-methyltetrahydrofuran was added, and the mixture
was concentrated to 8 L. The total volume of 2-methyltetrahy-
drofuran was adjusted to 24 L, and the solution was cooled to
–10 °C. Lithium aluminum hydride (4.7 L, 4.7 mol, 1 M THF
solution) was added to the above solution over 150 min, while
maintaining the pot temperature below 0 °C. The reaction
mixture was stirred at 0 °C for 5 h. HPLC analysis indicated
that <1% of methyl ester 16 remained. To a separate reactor
was added 30 L of water and 2.6 L of hydrochloride acid, and
the solution was cooled to 0 °C. The reaction mixture was
transferred to the cold hydrochloride solution over 60 min. The
reaction mixture was rinsed with 2 L of 2-methyltetrahydro-
furan. The layers were separated, and the organic layer was
washed with 18 L of 0.1 N HCl solution, followed by 18 L of
10% brine solution. The organic portion was concentrated
through atmospheric distillation to a total volume of 10 L. To
the still warm solution was added 20 L of heptane. The mixture
was concentrated to 8 L. To the still warm solution was again
added 20 L of heptane, and the mixture was concentrated to
16 L. The slurry was cooled to rt and stirred at rt for 4 h. The
slurry was filtered, and the solid was washed with 8 L of heptane
twice. The solid was dried in vacuo at 40 °C to give alcohol
13 as a white solid (2.14 kg, 5.84 mol, 75% yield from 14).
Chiral HPLC analysis indicated that the enantiomeric purity
was >99.5%. (ChiralPak AD 250 mm × 4.6 mm, 85% hexanes,
15% EtOH, 0.2% DEA, 1.0 mL/min, 30 °C, 225 nm, tR ) 5.4
1
min (undesired), 8.5 min (desired)). H NMR (500 MHz,
CDCl3) δ 7.35–7.26 (m, 10H), 7.00–6.96 (m, 1H), 6.91–6.86
(m, 1H), 6.81–6.78 (m, 1H), 3.93 (d, J ) 13.2 Hz, 2H),
3.60–3.53 (m, 3H), 3.37 (m, 1H), 3.14 –3.10 (m, 2H), 2.90 (m,
1H), 2.51 (t, J ) 10.4 Hz, 1H). 13C (400 MHz, CDCl3) δ 159.7,
158.2, 157.3, 155.8, 138.9, 129.0, 128.6, 127.4, 117.6, 117.4,
224
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development

and 60 mL of THF. The mixture was allowed to cool to rt
slowly. The resulting slurry was stirred at rt for 15 h and then
filtered to collect the solid. The solid was washed with 200 mL
of heptane and dried in vacuo at 45–50 °C to afford alcohol 11
as a white solid (64.6 g, 0.102 mol, 62% yield).
2-[3-Amino-4-(2,5-difluoro-phenyl)-2-hydroxy-butyl]-N-
tert-butyl-6-trifluoromethyl-nicotinamide (1). To a 12-L three
neck round-bottom flask equipped with a condenser, nitrogen
inlet, mechanical stirring, and thermocouple was charged water
wet 10% Pd/C (94 g), 5.9 L of EtOH, dibenzyl protected amino
alcohol 11 (590 g, 0.94 mol), and a solution of ammonium
formate (236 g, 3.7 mol) in 700 mL of water. The reaction
mixture was heated to 40 °C and held for 3 h or until HPLC
analysis indicated >99.5% conversion. The reaction mixture
was allowed to cool to 20 °C and then filtered through water-
wet Celite. The residual catalyst on the Celite was washed with
1 L of EtOH. The filtrate was then transferred to a 20-L
evaporatory flask and concentrated in vacuo to remove EtOH.
The concentrate was dissolved in 5.9 L of MTBE and was then
transferred to a separatory funnel; 2.4 L of water was added,
and layers were separated. Saturated aqueous NH4OH (200 mL)
was added to the aqueous layer, and the resulting aqueous
layer was extracted with 2.9 L of MTBE. The combined organic
layer was then washed with 1 L of water. The organic portion
was then solvent switched into 3 L of IPA. The solution was
stirred at 50 °C, and 2.2 L of water was added dropwise. The
solution became cloudy during the addition, and the mixture
was allowed to cool to rt. A slurry was formed upon cooling.
The resulting slurry was stirred at rt for 3 h and then filtered to
collect the solid. The solid was washed with 1.5 L of water
and dried in vacuo at 45 °C to a constant weight. The title
compound was obtained as a white solid (368 g, 0.83 mol, 88%
yield). The optical purity was assayed to be >99.5% ee.
(ChiralPak AD 250 mm × 4.6 mm, 85% hexanes, 15% IPA,
0.2% DEA, 1.0 mL/min, 30 °C, 225 nm, tR ) 5.2 min (desired),
6.5 min (undesired)). mp (DSC) (10 °C/min) onset 127.40 °C,
peak 128.41 °C. 1H NMR (CDCl3, 500 MHz) δ 7.98 (d, J )
7.7 Hz, 1 H), 7.58 (d, J ) 7.7 Hz, 1 H), 7.45 (s, 1H), 7.03–6.89
(m, 3H), 4.08–4.05 (m, 1 H), 3.38–3.33 (m, 1 H), 3.14–3.08
(m, 3 H), 2.67–2.62 (m, 1 H), 1.47 (s, 9 H). 13C (CDCl3, 400
MHz) δ 166.5, 159.7, 158.5, 157.3, 156.9, 156.1, 148.5, 148.1,
147.8, 147.4, 137.9, 136.6, 127.8, 127.7, 127.6, 127.5, 125.2,
122.4, 119.7, 118.0, 117.8, 117.7, 117.5, 117.0, 116.5, 116.3,
116.2, 114.7, 114.6, 114.4, 114.3, 75.1, 55.8, 52.3, 37.9, 31.9,
28.5. Anal. Calcd for C21H24F5N3O2: C, 56.63; H, 5.43; N, 9.43.
Found: C, 56.72; H, 5.42; N, 9.40.
Method B. Diisopropylamine (53.2 g, 0.526 mol) and 140
mL of THF were charged to a 5-L round-bottom flask equipped
with mechanical stirring, a nitrogen inlet, an addition funnel,
and a thermocouple. The solution was cooled to -25 °C in a
dry ice/acetone bath, and n-BuLi (193 mL, 0.482 mol, 2.5 M
solution in hexanes) was added over 30 min through an addition
funnel. The addition funnel was rinsed with 10 mL of THF.
The reaction was stirred for additional 5 min at -25 °C. A
solution of nicotinamide 8 (57.0 g, 0.219 mol) dissolved in 350
mL of THF was charged to an addition funnel and was added
to the reaction mixture over 30 min, while maintaining the
reaction temperature below -20 °C. A red solution was formed
upon the addition of nicotinamide 8. The reaction mixture was
stirred at -20 °C for 30 min. A solution of aldehyde 12 (80.0
g, 0.219 mol) dissolved in 150 mL of THF was added to the
addition funnel and was added to the reaction over 30 min,
again maintaining the reaction temperature below -20 °C. The
solution was rinsed in with an additional 10 mL of THF. The
reaction was stirred at -20 °C for 30 min and was quenched
by slow addition of 500 mL of water. The cooling bath was
removed, and the pH was adjusted to approximately 9 by adding
160 mL of 5 N HCl. After the layer separation, the aqueous
layer was extracted with 500 mL of EtOAc. The organic layers
were combined and washed with 500 mL of water. The organic
portion was solvent switched into 1440 mL of heptane and 80
mL of THF. The mixture was allowed to cool slowly to 23 °C,
and the resulting slurry was filtered. The solid was washed with
220 mL of heptane and dried in vacuo at 40–45 °C to a constant
weight. The title compound was isolated as a white solid (77.0
g, 0.123 mol, 56% yield). The optical purity was assayed to be
>99.5% ee. (ChiralPak AD-H 150 mm × 4.6 mm, 94%
hexanes, 6% IPA, 1.0 mL/min, 25 °C, 230 nm, tR ) 3.9 min
(undesired), 6.3 min (desired)). 1H NMR (500 MHz, CDCl3) δ
7.87 (d, J ) 8.0 Hz, 1H), 7.54 (d, J ) 8.0 Hz, 1H), 7.26–7.18
(m, 10H), 6.97–6.87 (m, 3H), 6.48 (s, 1H), 4.50 (bs, 1H),
3.78–3.69 (m, 5H), 3.31 (m, 1H), 3.17 (m, 1H), 3.04 (m, 1H),
2.99–2.93 (m, 2H), 1.34 (s, 9H). 13C (400 MHz, DMSO-d6) δ
166.6, 159.6, 158.7, 157.3, 156.3, 146.6, 146.3, 140.1, 138.1,
137.5, 130.8, 130.6, 128.7, 128.3, 127.0, 123.2, 120.4, 118.7,
118.4, 116.4, 116.2, 116.1, 114.2, 114.0, 69.7, 61.5, 53.8, 51.6,
42.1, 39.5, 28.7, 24.5. IR (KBr, cm –1) ν3432, 3244, 3071,
3031, 2168, 2848. MS (EI) m/z (rel intensity) 626 (M+, 100).
HRMS (ES+) exact mass calcd for C35H37F5N3O2 626.2800,
found 626.2801. Anal. Calcd for C35H36F5N3O2 C, 67.19, H,
5.80, N, 6.72, found C, 66.91, H, 5.81, N, 6.81.
Acknowledgement
We thank our Drug Discovery colleagues Scott M. Sheehan,
Chin Liu, and Eugene F. Kogut for useful discussions. We are
also thankful to Dr. Nicholas Magnus for helpful scientific
discussions and Mr. Timothy Woolsey for analytical support.
Received for review October 22, 2007.
OP700235C
Vol. 12, No. 2, 2008 / Organic Process Research & Development
•
225