Asymmetric conjugate addition of glycine derivatives under copper catalysis

Article abstract of DOI:10.1002/anie.201105258

Coppertunity knocks: An inexpensive, practical, and environmentally benign procedure for the enantioselective addition of glycine derivatives to α,β-unsaturated ketones has been developed (see scheme). By modifying the workup, the conjugate addition produ

Full text of DOI:10.1002/anie.201105258

DOI: 10.1002/anie.201105258
Asymmetric Catalysis
Asymmetric Conjugate Addition of Glycine Derivatives under Copper
Catalysis**
Mark Strohmeier, Kevin Leach, and Matthew A. Zajac*
Pyrrolidines are ubiquitous substructures which are incorpo-
rated into varied pharmaceutical and natural product targets,
and significant efforts have surrounded their syntheses.[1]
Although many synthetic approaches have been developed
for their formation,[1,2] we became interested in an inexpen-
sive, scalable, asymmetric, and sustainable synthesis that
would allow access to medicinally relevant derivatives of 1
(Scheme 1). Cycloaddition is a well-established method of
The conditions required for the conjugate addition of 5 to
4 (Table 1) were examined. Ketone 4a was chosen as a test
substrate,[6] along with the imine counterpart 5a. Metal–
ligand complexes were screened to obtain a selective reaction
favoring the conjugate addition product 6a over cycloaddition
product 7a. Reactions of 4 and 5 catalyzed by a calcium
bisoxazoline (BOX) complex[4h–j] containing ligand B1 pro-
duced mixtures of 6a and 7a (Table 1, entry 1).[7] Silver and
nickel-based systems with 2,2’-bis(diphenylphosphino)-1,1’-
binaphthyl (BINAP) 8 as the ligand (Table 1, entries 2 and 3)
afforded complex mixtures, which included products 6a and
7a. However, switching to a copper catalyst gave exceptional
results, with only trace amounts of 7a obtained (Table 1,
entry 4). The reaction could be run at lower catalyst loadings
(1 mol%) and lower temperatures resulting in a cleaner
reaction and increased selectivity for 6a (Table 1, entry 5).
Table 1: Selected screening conditions.
Scheme 1. Conjugate addition approach to 1.
forming pyrrolidines with great efficiency, but few processes
are capable of preparing 3,4-unsubstituted systems easily.[2]
The approach highlighted in Scheme 1 was of particular
interest because of the commercial availability of the glycine
derivatives 5. Standard hydrogenations of cyclic imines 2/3,
which have the desirable reaction features discussed above,
are also known to produce cis-1 with high selectively and in
high yield.[3] However, the existing asymmetric methods[4] for
the conjugate addition of 5 to 4 rely on expensive addi-
tives,[4b–e] high catalyst loadings (10 mol% or more),[4e–j,m,o] or
the use of cesium salts,[4g,n] which are difficult to use in fixed
equipment and have a high cost of disposal.[5] With this in
mind, a more convenient and more economic alternative is
desirable. Herein, we report a method for the reaction of 4
and 5 based on an environmentally innocuous chiral copper
complex, and present evidence gained through NMR spec-
troscopy for the solution structure of the associated inter-
mediates.
Entry[a]
Metal salt
Ligand
Temp. [8C]
6a:7a[b]
1
2
3
4
5
Ca(OiPr)2
B1
À20
À20
À20
À20
À78
3:2[d]
1:2
[c]
AgOAc
NiCl2
[Cu(MeCN)4]PF6
[Cu(MeCN)4]PF6
(Æ)-8
(Æ)-8
(Æ)-8
(Æ)-8
1:1[e]
18:1
>20:1
[f]
[a] Reactions were run at 0.2m with 5 mol% of the catalyst for 12–18 h.
[b] Based on approximate ratios of the corresponding signals in the NMR
spectrum of the crude reaction mixture. [c] 10 mol% catalyst loading was
used. [d] No DBU and 4 ꢀ molecular sieves were used according to the
published method.[4h–j] [e] Only a trace amount of product was detected
along with multiple by-products. [f] 1 mol% catalyst loading was used.
[*] Dr. M. Strohmeier, K. Leach, Dr. M. A. Zajac
API Chemistry and Analysis,
Twenty-four commercially available chiral ligands were
then screened at À208C and positive hits were identified on
the basis of enantioselectivity versus cost. Selected ligands
from the Josiphos and ferrocenyloxazolinylphosphines
(FOXAP) families[8] which were used in this screen are
depicted in Scheme 2.[9] The best selectivity was obtained by
using J1. It is important to note that most of the ligands
GlaxoSmithKline Pharmaceuticals, King of Prussia, PA (USA)
E-mail: Matthew.A.Zajac@gsk.com
[**] We thank Alan Freyer for help with NMR studies, J. J. Conde and
Huan Wang for helpful discussions, and Doug Fuerst and Chris
Morgan for help with HPLC data.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201105258.
Angew. Chem. Int. Ed. 2011, 50, 12335 –12338
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
12335

Communications
Table 2: Olefin substrates used in the conjugate addition.
Entry[a] Olefin
R’
Product
Yield [%](e.r.)[b]
1
2
3
4
5
6
7
8
4a R=p-BnOC6H4 Et
2a
2a
81 (99:1)
84 (94:6)[c]
79 (96:4)
85 (98:2)
88 (94:6)
78 (97:3)
77 (95:5)
70 (96:4)
71 (95:5)
78 (96:4)
88 (99:1)
81 (99:1)
Et
tBu 3a
Et 2b
tBu 3b
Et 2c
tBu 3c
Et 2d
tBu 3d
4b R=Ph
4c R=p-CF3C6H4
4d R=Me
9
10
11
12
4e R=tBu
4a
4e R=tBu
Et
Et
Et
2e
6a
6e
Scheme 2. Selected chiral ligands with the enantiomer ratios of 6a
produced in parentheses.
13
Et
82 (99:1)[d]
belonging to these families that were tested in this reaction
gave acceptable levels of enantioselectivity for our purposes,
especially since running the reaction at À788C improved the
selectivity. Ligands J1 (87% yield of isolated 6a, 99:1
enantiomer ratio at À208C) and F1 (86% yield of isolated
6a, 94:6 enantiomer ratio at À208C and a 99:1 enantiomer
ratio at À788C) were the most attractive based on cost and
performance. However, F1 was selected for further develop-
ment work, based on its availability on a larger scale. Ethyl
acetate, acetonitrile, toluene, methyl tert-butyl ether,
dichloromethane, and hexanes, along with NEt3 or tetra-
methylpiperdine were screened as solvent and base combi-
nations with F1, but nothing improved the selectivity of the
reaction or the reaction profile over THF and DBU.
We determined that 6a could be converted in situ into 2a
by using an acid/base workup. Treatment of a reaction
mixture containing 6a with aqueous 1m H2SO4 cleaved the
protecting group efficiently. The majority of the ligand and
benzophenone was removed by extraction with tert-butyl
methyl ether at this stage. After neutralization with aqueous
sodium bicarbonate, a dehydrative cyclization occurred to
afford 2a, with less than 2% epimerization of the stereocen-
ter. The S configuration of the stereocenter was established
by comparing the 2a obtained from the copper-catalyzed
reaction with a sample of 2a with known stereochemistry,
which was prepared from pyroglutamic acid.[3c]
[a] Reactions were run under the standard conditions (5a at À788C, 5b
at À208C) using the appropriate workup to produce either 2/3 or 6.
[b] The enantiomer ratio (e.r.) was determined by HPLC analysis. [c] The
reaction was run at À208C. [d] A 56:44 ratio of diastereomers was
obtained, both diastereomers were 99:1 e.r.[9]
to play little role in the outcome of the reaction, although
more electron-rich systems gave a higher selectivity. In
addition, increasing the size of alkyl substituents on the
ketone did not affect the overall rate of the reaction or the
selectivity (Table 2, entries 8–10). It is important to note that
although E- and Z-crotonyl systems based on 4a did not react,
the reaction with ketone 4 f furnished 6 f effectively (Table 2,
entry 13). Attempts to use acrylonitrile, methylacrylate, or
phenylvinyl sulfone dramatically altered the reaction pathway
and afforded cycloaddition products, similar to 7a, almost
exclusively.
The conjugate addition method also appears to be
scalable. Starting with 10 grams of 4a, product 2a was
isolated as a crystalline solid in 80% yield and greater than
99:1 enantiomer ratio from a reaction (Scheme 3). On this
scale, we found that only 0.5 mol% of catalyst was necessary
because of lower levels of adventitious water and oxygen. The
resulting imine product was then hydrogenated with Pt/C into
the corresponding pyrrolidine 1a. Acyclic products 6, which
do not contain the benzyloxy group, were converted directly
into 1 by Pd/C hydrogenation.[3d]
Alkyl- and arylvinyl ketones both proved to be viable
substrates for the conjugate addition/dehydrative cyclization
sequence (Table 2). Although both the ethyl (5a) and tert-
butyl (5b) esters were used, 5b required a higher temperature
(À208C) because of solubility issues and a slower rate of
reaction. The higher temperature resulted in a lower selec-
tivity for 6 when compared with the reaction with 5a, which
was run at À788C. Electronic differences between aryl rings
on the a,b-unsaturated ketone (Table 2, entries 1–7) seemed
To gain more understanding about ligand efficiency and
the structure of the associated copper complexes, intermedi-
ate species were studied by 1H NMR and 31P NMR spectros-
copy. As a result of the poor solubility of [Cu(MeCN)4]PF6 in
[D8]-THF and the fact that copper ligation is a fast process,[10]
formation of F1–Cu (Scheme 4) should drive the dissolution
of [Cu(MeCN)4]PF6. This appeared to be the case, since
12336 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 12335 –12338

A single new peak in the 31P NMR spectrum was detected at
d = À20.5 ppm. The formation of 10a instead of 10b was
confirmed by the detection of a nOe interaction between the
tert-butyl group and the lower ferrocene ring (Scheme 4). This
interaction provides strong experimental evidence for the top
face orientation of the imine phenyl substituents relative to
the ligand–metal complex.
Only one diamagnetic species was detected by 31P NMR
spectroscopy, which indicated that 10a exists as the major
complex in solution (approximately 9:1 or greater). This
corresponds to an estimated energy difference of approx-
imately 1.3 kcalmolÀ1[12] or more, between 10a and 10b.
Similar results from the NMR spectrum of a mixture of J1,
5b, and [Cu(MeCN)4]PF6 revealed that complex 11 forms,
which corresponds to the analogous binding of 5b to F1–Cu.
Density functional theory (DFT) full structure optimiza-
tions were performed for 10a and 10b by using the
B3LYP functional with the LANL2DZ basis set for all
transition metals with effective core potentials and 6-31G*
for all other elements.[13] Both optimized structures in
Figure 1 have a distorted tetrahedral geometry around the
copper center. The phosphine and imine phenyl groups in 10a
assume an offset, approximately coplanar conformation, with
the tert-butyl group below the complex. The tert-butyl group
in 10b tips the substrate into a more angled position relative
to the ferrocene moiety, with the imine and phosphine phenyl
planes perpendicular. The electronic energy difference
between the two optimized structures was calculated to be
À0.42 and À0.31 kcalmolÀ1 at the HF and B3LYP level with
the above basis set, respectively. However, the difference was
calculated to be À2.39 kcalmolÀ1 using Møller-Plesset per-
turbation theory (MP2), which is in accord with the exper-
imental observations. It is known that HF and B3LYP are not
suitable to account for attractive van der Waals interactions,
as encountered in p stacking, and the observed difference
between HF, DFT, and perturbation theory is in accord with
p stacking energies reported in the literature.[14] Hence, the
differences between HF, B3LYP, and MP2 calculations
provide evidence that van der Waals interactions have a
Scheme 3. Conversion of 4a into 1a.
Scheme 4. Intermediate complexes.
F1–Cu formed quantitatively upon dissolution of both F1 and
[Cu(MeCN)4]PF6, whereas [Cu(MeCN)4]PF6 would not com-
pletely dissolve without the ligand present. The protons of the
MeCN moieties (integration = 12) were detected at the same
1
chemical shift as unbound MeCN in the H NMR spectrum,
which indicated some amount of fast exchange, or nearly
complete [D8]-THF ligation of the complex. Two signals were
present in the 31P NMR spectrum. One septet at d =
À144.3 ppm corresponded to PF6,[11] and one singlet at d =
À19.8 ppm corresponded to complex F1–Cu. The singlet at
d = À19.8 ppm had shifted from the
signal detected for free F1 at d =
À17.6 ppm. It appears that a 1:1
complex of F1:Cu forms, since
models (Maruzen HGS Stereochem-
istry) indicated that severe steric
interactions would preclude dimer
formation, and all of the [Cu-
(MeCN)4]PF6 dissolved. Furthermore
the addition of other chelating ligands
to F1–Cu resulted in facile exchange
and produced one species in each
case, which was detected by 31P NMR
spectroscopy (Scheme 4).
Our attention turned to the imine
complex 10a (Scheme 4). Upon dis-
solution of 5b in a solution of F1–Cu,
the initial copper complex (F1–Cu)
could no longer be detected by
Figure 1. B3LYP 6-31G*/LANL2DZ geometry optimized structures. Fe blue sphere, Cu pink,
1H NMR or 31P NMR spectroscopy. P orange, O red, N blue.
Angew. Chem. Int. Ed. 2011, 50, 12335 –12338
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 12337

Communications
Bhagavatula, M. G. Fickes, D. Pireh, L. R. Solomon, M. R. Lake,
R. Edalji, E. H. Fry, H. L. Sham, J. M. Trevillyan, J. Med Chem.
2006, 49, 3520; c) G. Alvaro, D. Amantini, M. Bergauer, F.
Bonetti, R. Profeta, (Glaxo Group Ltd.), WO2007042240A1,
2007; d) A. Hutchinson, B. Chenard, G. Li, M. Ghosh, J. Tarrant,
T. Yoon, G. Luke, K. Lee, M. M. O’Donnell, W. Pringle, J.
Peterson, K. Hodgetts, C. Steenstra, D. Doller, (Cantor Colburn
LLP), US20060009456A1, 2006.
significant stabilizing effect in 10a. This effect is likely to be
caused by the coplanar phenyl p stacking that is absent in 10b.
Based on molecular orbital calculations and NMR experi-
ments, binding of the imine is favored for a coplanar
orientation of the imine and phosphine phenyl groups,
allowing for p stacking (as drawn for 10a in Scheme 4 and
Figure 1). Alternate binding results in the higher energy
conformer 10b, because of the lack of attractive p stacking
interactions. These results emphasize the importance of the
phosphine groups for complex stability and, hence, the
selectivity of the reaction.[9] Taking this binding geometry
and the product stereocenters into account, 4 must approach
the associated anion from the si-face and eclipse the Ph
groups.
In summary, we have described an inexpensive and
practical procedure for the enantioselective reaction between
glycine derivatives 5 and a,b-unsaturated ketones 4 under
copper catalysis. The products from these reactions can be
directly isolated (6 with a basic workup or 2 with an acidic
workup) and transformed into the pyrrolidine 1. This method
uses low catalyst loadings (down to 0.5 mol%) and both
enantiomers of the catalyst are commercially available.
Additionally, the insights gained from NMR spectroscopy
and molecular orbital calculations, which supports the binding
orientation of 5b to the chiral copper catalyst, should be
useful for other catalytic systems that utilize rigid chiral
ligands, such as FOXAP and Josiphos.
[4] a) A. Lꢀpez, R. Pleixats, Tetrahedron: Asymmetry 1998, 9, 1967;
b) M. J. Oꢁ Donnell, F. Delgado, E. Domꢂnguez, J. de Blas, W. L.
Scott, Tetrahedron: Asymmetry 2001, 12, 821; c) S. Arai, K.
Tokumaru, T. Aoyama, Chem. Pharm. Bull. 2004, 52, 646; d) B.
Lygo, B. Allbutt, E. H. M. Kirton, Tetrahedron Lett. 2005, 46,
4461; e) T. Ishikawa, Y. Araki, T. Kumamoto, H. Seki, K.
Fukuda, T. Isobe, Chem. Commun. 2001, 245; f) T. Akiyama, M.
Hara, K. Fuchibe, S. Sakamoto, K. Yamaguchi, Chem. Commun.
2003, 1734; g) E. J. Corey, M. C. Noe, F. Xu, Tetrahedron Lett.
1998, 39, 5347; h) S. Saito, T. Tsubogo, S. Kobayashi, J. Am.
Chem. Soc. 2007, 129, 5364; i) S. Kobayashi, T. Tsubogo, S. Saito,
Y. Yamashita, Org. Lett. 2008, 10, 807; j) T. Tsubogo, S. Saito, K.
Seki, Y. Yamashita, S. Kobayashi, J. Am. Chem. Soc. 2008, 130,
13321. For asymmetric additions of glycine derivatives to imines,
see k) X. X. Yan, Q. Peng, Q. Li, K. Zhang, J. Yao, X. L. Hou,
Y. D. Wu, J. Am. Chem. Soc. 2008, 130, 14362. For SN2’ additions,
see l) C. G. Chen, X. L. Hou, L. Pu, Org. Lett. 2009, 11, 2073;
m) B. Lygo, C. Beynon, C. Lumley, M. C. McLeod, C. E. Wade,
Tetrahedron Lett. 2009, 50, 3363; n) T. Ma, X. Fu, C. W. Kee, L.
Zong, Y. Pan, K. Huang, C. J. Tan, J. Am. Chem. Soc. 2011, 133,
2828; o) X. Hou, Q. Li, J. Yao, L. Dai, Chin. J. Chem. 2010, 28,
1761.
[5] Efficient mixing of CsOH on a manufacturing scale is difficult
because of its density. Also, waste disposal is more than 500-
times more expensive when the waste contains cesium.
[6] Ketone 4a is a bench-stable crystalline solid. See the Supporting
Information for a preparative procedure.
Received: July 26, 2011
Revised: September 28, 2011
Published online: October 26, 2011
[7] The ratios obtained by Kobayashi et al. for conjugate addition
versus [3+2] cycloaddition were heavily dependent upon the
structure of the substrate. To favor conjugate addition, the
structure of 5 was modified (see Ref. [4i]), but for our purposes
this is prohibitively expensive, and still requires 10 mol% of
catalyst to proceed efficiently.
[8] All of these ligands were purchased from Solvias. For FOXAP
ligands, see Y. Miyake, Y. Nishibayashi, S. Uemura, Synlett 2008,
1747, and references therein.
Keywords: 1,4-addition · asymmetric catalysis · ferrocenes ·
glycines · pyrrolidines
.
[1] a) A. van der Werf, R. M. Kellogg, Tetrahedron Lett. 1991, 32,
3727; b) J. Sisko, J. R. Henry, S. M. Weinreb, J. Org. Chem. 1993,
58, 4945; c) L. E. Overman, J. E. Tellew, J. Org. Chem. 1996, 61,
8338; d) W. H. Pearson in Studies in Natural Product Chemistry,
Vol. 1 (Ed.: Atta-Ur-Rahman), Elsevier, New York, 1998, p. 323;
e) D. J. Denhart, D. A. Griffith, C. H. Heathcock, J. Org. Chem.
1998, 63, 9616; f) T. Shibuguchi, H. Mihara, A. Kuramochi, S.
Sakuraba, T. Ohshima, M. Shibasaki, Angew. Chem. 2006, 118,
4751; Angew. Chem. Int. Ed. 2006, 45, 4635; g) T. Arai, A.
Mishiro, N. Yokoyama, K. Suzuki, H. Sato, J. Am. Chem. Soc.
2010, 132, 5338.
[2] a) K. V. Gothelf, K. A. Jørgensen, Chem. Rev. 1998, 98, 863; b) I.
Coldham, R. Hufton, Chem. Rev. 2005, 105, 2765; c) G. Pandey,
P. Banerjee, S. R. Gadre, Chem. Rev. 2006, 106, 4484; d) L.
Stanley, M. P. Sibi, Chem. Rev. 2008, 108, 2887.
[3] a) J. S. Petersen, G. Fels, H. Rapoport, J. Am. Chem. Soc. 1984,
106, 4539; b) Z. Pei, X. Li, K. Longenecker, T. W. von Geldern,
P. E. Wiedeman, T. H. Lubben, B. A. Zinker, K. Stewart, S. J.
Ballaron, M. A. Stashko, A. K. Mika, D. W. A. Beno, M. Long,
H. Wells, A. J. Kempf-Grote, D. J. Madar, T. S. McDermott, L.
[9] See the Supporting Information for more details.
[10] a) T. Osako, Y. Tachi, M. Doe, M. Shiro, K. Ohkubo, S.
Fukuzumi, S. Itoh, Chem. Eur. J. 2004, 10, 237; b) X. Ribas, R.
Xifra, T. Parella, A. Poater, M. Solꢃ, A. Llobet, Angew. Chem.
2006, 118, 3007; Angew. Chem. Int. Ed. 2006, 45, 2941.
[11] C. Dꢄaz, Polyhedron 1997, 16, 999.
[12] DG = ÀRTlnK = À(0.001986 kcalmolÀ1 KÀ1)(298 K)ln(1/9) =
À1.3.
[13] All calculation were performed with Gaussian 03 (Gaussian 03,
revision D.02, M. J. Frisch et al.). See the Supporting Informa-
tion.
[14] a) X. Ye, Z.-H Li, W. Wang, K. Fan, W. Xu, Z. Hua, Chem. Phys.
Lett. 2004, 397, 56; b) X. X. Yan, Q. Peng, Y. Zhang, K. Zhang,
W. Hong, X. L. Hou, Y. D. Wu, Angew. Chem. 2006, 118, 2013;
Angew. Chem. Int. Ed. 2006, 45, 1979.
12338 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 12335 –12338