A novel crystallization-induced diastereomeric transformation based on a reversible carbon-sulfur bond formation. Application to the synthesis of a γ-secretase inhibitor

Article abstract of DOI:10.1021/jo0705925

(Chemical Equation Presented) This paper describes a remarkably efficient process for the preparation of γ-secretase inhibitor 1. The target is synthesized in only five steps with an overall yield of 58%. The key operation is a highly selective and practical, crystallization-driven transformation for the conversion of a mixture of tertiary benzylic alcohols into the desired sulfide diastereomer with 94:6 dr. This unprecedented process is based upon a reversible carbon-sulfur bond formation under acidic conditions.

Full text of DOI:10.1021/jo0705925

A Novel Crystallization-Induced Diastereomeric Transformation
Based on a Reversible Carbon-Sulfur Bond Formation. Application
to the Synthesis of a γ-Secretase Inhibitor
Antony J. Davies,*,† Jeremy P. Scott,*,† Brian C. Bishop,† Karel M. J. Brands,†
Sarah E. Brewer,† Jimmy O. DaSilva,‡ Peter G. Dormer,‡ Ulf-H. Dolling,‡
Andrew D. Gibb,† Deborah C. Hammond,† David R. Lieberman,† Michael Palucki,‡ and
Joseph F. Payack‡
Department of Process Research, Merck, Sharp & Dohme Research Laboratories, Hertford Road,
Hoddesdon, Hertfordshire, EN11 9BU, United Kingdom, and Department of Process Research, Merck
Research Laboratories, P.O. Box 2000, Rahway, New Jersey 07065
tony_daVies@merck.com; jeremy_scott@merck.com
ReceiVed March 22, 2007
This paper describes a remarkably efficient process for the preparation of γ-secretase inhibitor 1. The
target is synthesized in only five steps with an overall yield of 58%. The key operation is a highly
selective and practical, crystallization-driven transformation for the conversion of a mixture of tertiary
benzylic alcohols into the desired sulfide diastereomer with 94:6 dr. This unprecedented process is based
upon a reversible carbon-sulfur bond formation under acidic conditions.
Introduction
the amyloid-â (Αâ) peptide and threadlike neuronal structures
composed of the hyperphosphorylated TAU protein. Both Αâ
and TAU are thought to be crucial to pathogenesis, but both
genetic and biochemical evidence supports Αâ as the major
causative pathway. It is expected that inhibition or modulation
of γ-secretase activity in the CNS will decrease Αâ formation
and thereby alter the underlying pathophysiology of AD. Thus,
the complex γ-secretase enzyme has emerged as an attractive
drug target. Several inhibitors of this enzyme have been
discovered and shown to lower Aâ levels in preclinical models
of AD.2 Recent research at Merck3 has identified compound 1
as a suitable γ-secretase inhibitor. We recently reported on an
Alzheimer’s disease (AD) is the most common of the
neurodegenerative disorders.1 In the elderly, it represents the
most frequently occurring form of dementia, affecting 10% of
individuals over 65 years of age and nearly half of those over
85. It is a progressive and ultimately fatal neurological disorder
for which there is no effective treatment at present. The disease
is characterized pathologically by cerebral plaques that contain
* Author to whom correspondence should be addressed. Phone: +44 (0)-
1992 452614. Fax: +44 (0)1992 452581.
† Merck Sharp & Dohme Research Laboratories.
‡ Merck Research Laboratories.
(1) Pertinent reviews of Alzheimer’s disease, the â-amyloid hypothesis
and potential disease-modifying therapies can be found in the following
references: (a) Larner, A. Expert Opin. Ther. Pat. 2004, 14, 1403. (b) Citron,
M. Nat. ReV. Neurosci. 2004, 5, 677. (c) Parihar, M. S.; Hemnani, T. J.
Clin. Neurosci. 2004, 11, 456. (d) Bullock, R. Expert Opin. InVestig. Drugs
2004, 13, 303. (e) Esteban, J. A. Trends Neurosci. 2004, 27, 1. (f) Mani,
R. B. Statist. Med. 2004, 23, 305. (g) Kornilova, A. Y.; Wolfe, M. S. Annu.
Rep. Med. Chem. 2003, 38, 41. (h) Aguzzi, A.; Haass, C. Science 2003,
302, 814. (i) Lahiri, D. K.; Farlow, M. R.; Sambamurti, K.; Greig, N. H.;
Giacobini, E.; Schneider, L. S. Curr. Drug Targets 2003, 4, 97. (j) Gurwitz,
D. Drug DeV. Res. 2002, 56, 45. (k) Walter, J.; Haass, C. Drug DeV. Res.
2002, 56, 201. (l) Wolfe, M. S. Nat. ReV. 2002, 859. (m) Selkoe, D. J. J.
Clin. InVest. 2002, 110, 1375. (n) Lahiri, D. K.; Farlow, M. R.; Greig, N.
H.; Sambamurti, K. Drug DeV. Res. 2002, 56, 267.
(2) Selected articles: (a) Prasad, C. V. C.; Vig, S.; Smith, D. W.; Gao,
Q.; Polson, C. T.; Corsa, J. A.; Guss, V. L.; Loo, A.; Barten, D. M.; Zheng,
M.; Felsenstein, K. M.; Roberts, S. B. Bioorg. Med. Chem. Lett. 2004, 14,
3535. (b) Fuwa, H.; Okamura, Y.; Morohashi, Y.; Tomita, T.; Iwatsubo,
T.; Kan, T.; Fukuyama, T.; Natsugari, H. Tetrahedron Lett. 2004, 45, 2323.
(c) Churcher, I.; Williams, S.; Kerrad, S.; Harrison, T.; Castro, J. L.;
Shearman, M. S.; Lewis, H. D.; Clarke, E. E.; Wrigley, J. D. J.; Beher, D.;
Tang, Y. S.; Liu, W. J. Med. Chem. 2003, 46, 2275. (d) Churcher, I.; Ashton,
K.; Butcher, J. W.; Clarke, E. E.; Harrison, T.; Lewis, H. D.; Owens, A.
P.; Teall, M. R.; Williams, S.; Wrigley, J. D. J. Bioorg. Med. Chem. Lett.
2003, 13, 179. (e) Bihel, F.; Quelever, G.; Lelouard, H.; Petit, A.; Alves da
Costa, C.; Pourquie, O.; Checler, F.; Thellend, A.; Pierre, P.; Kraus, J.-L.
Bioorg. Med. Chem. 2003, 11, 3141.
10.1021/jo0705925 CCC: $37.00 © 2007 American Chemical Society
Published on Web 05/24/2007
4864
J. Org. Chem. 2007, 72, 4864-4871

Crystallization-Induced Diastereomeric Transformation
known reduction of the inexpensive 6.6 The key step in this
analysis is the diastereoselective introduction of a sulfur-based
nucleophile. At the outset of our activities we were hopeful that
under acidic conditions, addition of 4-chlorothiophenol (4-CTP)
to incipient cation 7 would favor the formation of the desired
cis isomer 8 for stereoelectronic reasons (Scheme 3).7
SCHEME 1
In the event that the thiol addition proved unselective,
however, we thought that we might be able to develop a novel
crystallization-induced process for this transformation. Two
fundamental requirements need to be satisfied in order to
transform this hypothesis into reality. The desired sulfide 8 needs
to be less soluble than diastereomer 9. If this is the case, then
conditions need to be identified under which 8 can be crystal-
lized while simultaneously, the two diastereomers equilibrate.8
Many crystallization-induced transformations9 have been de-
scribed in the literature, but most of them deal with salts or
other aggregates.10 Cases in which the diastereomeric set of
stereocenters are part of the same molecule are much rarer11-13
as are transformations in which the diastereomers are achiral.
SCHEME 2
(5) SN1 type additions of thiols to cyclohexyl benzylic positions are rare
but have been noted; see: (a) Koenig, W.; Geiger, R.; Siedel, W. Chem.
Ber. 1968, 101, 681-693. Additions to simple tertiary benzylic moieties
are more prevalent; see: (b) Nishino, T.; Nishiyama, Y.; Sonoda, N. Chem.
Lett. 2003, 32, 918.
(6) (a) Adkins, H.; Bowden, E. J. Am. Chem. Soc. 1940, 62, 2422. (b)
Adkins, H.; Harris, E. E.; D’Ianni, J. J. Am. Chem. Soc. 1938, 60, 1467.
(7) By analogy with the preferred axial attack of sterically unencumbered
nucleophiles on conformationally-locked, cyclohexyl centers. See (a)
Greeves, N.; Lyford, L.; Pease, E. J. Tetrahedron Lett. 1994, 35, 285. (b)
Reetz, M. T.; Harmat, N.; Mahrwald, R. Angew. Chem. Int., Ed. Engl. 1992,
31, 342. (c) Keys, B. A.; Eliel, E. L.; Juaristi, E. Isr. J. Chem. 1989, 29,
171. (d) Meakins, G. D.; Percy, K. R.; Richards, E. E.; Young, R. N. J.
Chem. Soc. C 1968, 9, 1106. (e) Houlihan, W. J. J. Org. Chem. 1962, 27,
3860. (f) Jones, W. M. J. Am. Chem. Soc. 1960, 82, 2528. (g) Hennion, G.
F.; O’Shea, F. X. J. Am. Chem. Soc. 1958, 80, 614. (h) Hennion, G. F.;
Barrett, S. O. J. Am. Chem. Soc. 1957, 79, 2146. (i) Stork, G.; White, W.
N. J. Am. Chem. Soc. 1956, 78, 4609.
(8) Tertiary benzylic thiols are labile under acidic conditions: (a) Horton,
J. R.; Stamp, L. M.; Routledge, A. Tetrahedron Lett. 2000, 41, 9181. (b)
Eicher, T.; Lerch, D. Tetrahedron Lett. 1980, 21, 3751.
(9) These processes are sometimes, less accurately, described as crystal-
lization-induced dynamic resolutions. For a review, see Brands, K. J. M.;
Davies, A. J. Chem. ReV. 2006, 106, 2711.
(10) (a) Jacques, J.; Collet, A.; Wilen, S. H. Enantiomers, Racemates
and Resolution; Wiley-Interscience Publication: New York, 1981; pp 369-
377 and references therein. (b) Chen, J. G.; Zhu, J.; Skonezny, P. M.; Rosso,
V.; Venit, J. J. Org. Lett. 2004, 6, 3233. (c) Kiau, S.; Discordia, R. P.;
Madding, G.; Okuniewicz, F. J.; Rosso, V.; Venit, J. J. J. Org. Chem. 2004,
69, 4256. (d) Patrick, B. O.; Scheffer, J. R.; Scott, C. Angew. Chem. Int.
Ed. 2003, 42, 3775. (e) Ko_smrlj, J.; Weigel, L. O.; Evans, D. A.; Downey,
W. C.; Wu, J. J. Am. Chem. Soc. 2003, 125, 3208. (f) Kaptein, B.; Boesten,
W. H. J.; De Lange, B.; Grimbergen, R. F. P.; Hulshof, L. A.; Vries, T. R.;
Seerden, J.-P. G.; Broxterman, Q. B. Chim. Oggi 2002, 77. (g) Brunetto,
G.; Gori, S.; Fiaschi, R.; Napolitano, E. HelV. Chim. Acta 2002, 85, 3785.
(h) Boesten, W. H. J.; Seerden, J.-P. G.; De Lange, B.; Dielemans, H. J.
A.; Elsenberg, H. L. M.; Kaptein, B.; Moody, H. M.; Kellogg, R. M.;
Broxterman, Q. B. Org. Lett. 2001, 3, 1121. (i) Kolarovic, A.; Berkes, D.;
Baran, P.; Povazanec, F. Tetrahedron Lett. 2001, 42, 2579. (j) Lee, S.-K.;
Lee, S. Y.; Park, Y. S. Synlett 2001, 1941. (k) Shi, Y.-J.; Wells, K. M.;
Pye, P. J.; Choi, W.-B.; Churchill, H. R. O.; Lynch, J. E.; Maliakal, A.;
Rossen, K.; Reider, P. J.; Sager, J. W.; Volante, R. P. Tetrahedron 1999,
55, 909. (l) Maryanoff, C. A.; Scott, L.; Shah, R. D.; Villani, F. J., Jr.
Tetrahedron: Asymmetry 1998, 9, 3247. (m) Shieh, W.-C.; Carlson, J. A.;
Zaunius, G. M. J. Org. Chem. 1997, 62, 8271. (n) Caddick, S.; Jenkins, K.
Chem. Soc. ReV. 1996, 447.
(11) Also known as ‘self-controlled’ resolutions. For the evolution of
this term, see: Kanomata, N.; Yoshiharu, O. Tetrahedron Lett. 2001, 42,
1045.
(12) Brands, K. M. J.; Payack, J. F.; Rosen, J. D.; Nelson, T. D.;
Candelario, A.; Huffman, M. A.; Zhao, M. M.; Li, J.; Craig, B.; Song, Z.
J.; Tschaen, D. M.; Hansen, K.; Devine, P. N.; Pye, P. J.; Rossen, K.;
Dormer, P. G.; Reamer, R. A.; Welch, C. J.; Mathre, D. J.; Tsou, N. N.;
McNamara, J. M.; Reider, P. J. J. Am. Chem. Soc. 2003, 125, 2129 and
references therein.
efficient stereoselective synthesis of this clinical candidate.4 In
this paper we report on a new and significantly more practical
synthesis of 1.
Compound 1 is an achiral 1,1,4-trisubstituted cyclohexane
derivative. Crucial to the design of any successful synthesis is
the control of the 1,4-cis relative stereochemistry between the
propionic acid side chain and the 4-chlorophenylsulfonyl moiety.
In our previous synthesis of this compound (Scheme 1), we
developed two alternative strategies which achieved this goal.
The first synthesis was based on the catalyst-controlled hydro-
genation of racemic cyclohexene derivative 2 while the second
used a conjugate hydride reduction of acrylonitrile derivative
3. Unfortunately, we foresaw that both of these syntheses had
limitations for implementation on a manufacturing scale. The
hydrogenation route, although quite convergent (six linear steps),
was not very selective (75:25 dr). The conjugate reduction route
was highly stereoselective (>99.9:0.1 dr) but required a rather
lengthy linear sequence of steps to introduce the propionic acid
side chain.
Our new approach to the synthesis of 1 is based on a
dramatically different analysis of the problem (Scheme 2).
Disconnection of the carbon-sulfur bond at the tertiary stereo-
genic center reveals precursor 4, with undefined stereochemistry
at the benzylic position, as a key intermediate.5 Access to 4
was envisaged to occur by a chemoselective organometallic aryl
addition to the ketone moiety in ketoacid 5. We thought that
the latter compound would be available by improvement of the
(3) Castro Pineiro, J. L.; Churcher, I.; Dinnell, K.; Harrison, T.; Kerrad,
S.; Nadin, A. J.; Oakley, P. J.; Owens, A. P.; Shaw, D. E.; Teall, M. R.;
Williams, B. J.; Williams, S. PCT Int. Pat. Appl. WO 2002081435, Aug
21, 2001; Chem. Abstr. 2002, 137, 310695.
(4) (a) Brands, K. M. J.; Davies, A. J.; Oakley, P. J.; Scott, J. P.; Shaw,
D. E.; Teall, M. R. PCT Pat. Appl. WO 2004013090, February 12, 2004;
Chem. Abstr. 2004, 140, 163493. (b) Scott, J. P.; Lieberman, D. R.; Beureux,
O. M.; Brands, K. M. J.; Davies, A. J.; Gibson, A. W.; Hammond, D. C.
J. Org. Chem. 2007, 72, 3086.
J. Org. Chem, Vol. 72, No. 13, 2007 4865

Davies et al.
SCHEME 3
SCHEME 4 a
SCHEME 5 a
a Conditions: (a) 11.2 bar H2 pressure, 5% Rh/Al2O3, IPAc, 80 °C. (b)
RuCl3, AcOH, 10-13 wt % NaOCl.
a (a) Conditions: i-PrMgCl, 1-bromo-2,5-difluorobenzene, THF, -30 °C
then add solution of Mg salt of 5 generated using i-PrMgCl at <-50 °C.
To the best of our knowledge, reversible cleavage and formation
of carbon-sulfur bonds have not been described in this context.
We therefore developed a practical and high-yielding pro-
cedure which obviated the isolation of the mixture of 10a and
10b. Thus, hydrogenation of 6 in the presence of 5% Rh/Al2O3
in isopropyl acetate (iPAc) at 80 °C under 11.2 bar of hydrogen
gave a 1.2-1.4:1 mixture of 10a and 10b, respectively, in 89%
assay yield. The main impurity observed by GCMS was the
over-reduced product 11 at around 5%; however, this was easily
rejected in the eventual crystallization of ketoacid 5 (Vide infra).
After filtration of the catalyst, the mixture of 10a and 10b was
oxidized with bleach and RuCl3 catalysis under biphasic
conditions17 to afford 5 in 81% yield over the two steps.
Addition of the 1,4-Difluoroaryl Moiety. With the ketoacid
5 in hand, several protocols for the chemoselective addition of
a 1,4-difluorophenyl moiety to the ketone moiety were inves-
tigated.18 We favored a sacrificial protocol, whereby the
carboxylic acid proton of 5 was removed by an 1 equiv of base,
prior to the addition of the 1,4-difluorophenyl organometallic
species. Initially, we prepared 2,5-difluorophenylmagnesium
Grignard from the corresponding bromide by halogen ex-
change19 with i-PrMgCl (Scheme 5). This reagent was then
added to a solution of the magnesium salt of 5 at e-20 °C
which had been generated using another sacrificial equivalent
of i-PrMgCl at <-50 °C. The temperature for addition of the
i-PrMgCl to generate the magnesium carboxylate salt is critical;
above -50 °C, isopropyl addition is observed to the ketone
moiety to give diastereomeric alcohols 13 (Scheme 5).20 In early
attempts, we obtained relatively low yields for carbinols 4 in
this process (50-60%) which we ascribed to competitive ketone
Results and Discussion
Synthesis of Ketoacid 5. A synthesis of 5 which provides
the nine-carbon backbone of 1 has been described in the
literature. Over-reduction of the readily available trans-p-
coumaric acid 6 to diastereomeric alcohols 10a and 10b can be
achieved in >85% yield (Scheme 4).14 However, the reoxidation
of this mixture to 5 is significantly less efficient.15 We hoped,
therefore, that the reduction of 6 could be halted at the stage of
5.16 A large number of heterogeneous catalytic reduction
conditions were screened starting from either 6 or intermediate
phloretic acid 12. Under the best conditions we were able to
identify (12 wt % of 5% Pd/Al2O3 in water containing 0.9 equiv
of NaOH at 70 °C under 6.2 bar of hydrogen) that 6 was reduced
to a 85:12:3 mixture of 5, 10a/10b, and 12, respectively.
Unfortunately, we could not isolate the desired 5 in pure form
from this mixture.
(13) (a) Pye, P. J.; Rossen, K.; Weissman, S. A.; Maliakal, A.; Reamer,
R. A.; Ball, R.; Tsou, N. N.; Volante, R. P.; Reider, P. J. Chem. Eur. J.
2002, 8, 1372. (b) Komatsu, H.; Awano, H. J. Org. Chem. 2002, 67, 5419.
(c) Silverberg, L. J.; Kelly, S.; Vemishetti, P.; Vipond, D. H.; Gibson, F.
S.; Harrison, B.; Spector, R.; Dillon, J. L. Org. Lett. 2000, 2, 3281.
(14) (a) Newman, M. S.; Yu, T. Y. J. Am. Chem. Soc. 1952, 74, 507.
(b) Mastagli, P.; Metayer, M. Compt. Rend. 1947, 224, 1779.
(15) (a) Pletneva, I. D.; Muromova, R. S.; Pervukhina, I. V. Zh. Obshch.
Khim. 1965, 10, 708. (b) Bowden, E.; Adkins, H. J. Am. Chem. Soc. 1940,
62, 2422.
(16) The analogous reduction of phenol to cyclohexanone is a known
process. For some recent articles, see (a) Ohde, H.; Ohde, M.; Wai, C. M.
Chem. Commun. 2004, 8, 930. (b) Shore, S. G.; Ding, E.; Park, C.; Keane,
M. A. J. Mol. Catal. A 2004, 212, 291. (c) He, H.-J.; Suo, Z.-H.; Zou,
X.-H.; An, L.-D. Yantai Daxue Xuebao 2003, 16, 185. (d) Rode, C. V.;
Joshi, U. D.; Sato, O.; Shirai, M. Chem. Commun. 2003, 15, 1960. (e) Basov,
N. L.; Gryaznov, V. M.; Ermilova, M. M. Zh. Fiz. Khim. 1993, 67, 2413.
(f) Dodgson, I.; Griffin, K.; Barberis, G.; Pignataro, F.; Tauszik, G. Chem.
Ind. (London) 1989, 24, 830.
(17) Gonsalvi, L.; Arends, I. W. C. E.; Sheldon, R. A. Org. Lett. 2002,
4, 1659.
(18) The chemoselective addition of an allyl organometallic reagent to
a ketone in the presence of an acid moiety has been reported in Kumar, S.;
Kaur, P.; Chimni, S. S. Synlett 2002, 573 and references therein.
(19) For a recent review, see: Knochel, P.; Dohle, W.; Gommerman,
N.; Kneisel, F. F.; Kopp, F.; Korn, T.; Sapountzis, I.; Vu, V. A. Angew.
Chem. Int. Ed. 2003, 42, 4302.
4866 J. Org. Chem., Vol. 72, No. 13, 2007

Crystallization-Induced Diastereomeric Transformation
enolization.21 However, detailed investigations showed that
ketoacid 5 is hygroscopic. When the protocol was performed
under rigorously dry conditions, excellent conversions (>90%)
were observed. Using a pH-controlled extractive workup,
isopropyl bromide, 1,4-difluorobenzene, unreacted 1-bromo-2,5-
difluorobenzene, and other low-level impurities could be
separated from carbinols 4, which are soluble in aqueous NaOH.
A 3.6:1 mixture of 4a and 4b, respectively, could be isolated
in 87% yield after crystallization (Scheme 5).
TABLE 1. Addition of 2,5-Difluorophenyllithium to the
Magnesium Salt of 5
equiv ArLi
assay yield (%),
isomer ratio,
entry
(HPLC assay)
4a + 4b
4a:4b
1
2
3
4
1.1
1.3
1.6
1.8
67
76
83
85
1:1.4
1:1.4
1:1.3
1:1.3
We were concerned that the 2,5-difluorophenyl Grignard
reagent might be unstable and would generate the corresponding
benzyne.22,23 A study of the formation and stability of this
reagent, however, showed that it was fully formed within 1 h
and is stable for up to 4 h at <-10 °C. While we were
unconcerned with the stereochemical outcome of the addition
reaction, it is interesting to note that transmetalation of the
Grignard reagent with cerium24 resulted in a higher proportion
of equatorial attack, leading to a 1:1.9 ratio of 4a and 4b,
respectively.25
Further work on this transformation focused on attempts to
make it more practical and economical. As mentioned above,
we initially used 1 equiv of i-PrMgCl as the sacrificial base for
carboxylate generation which required a temperature <-50 °C.
Far more convenient was to treat 5 with equimolar amounts of
triethylamine and anhydrous MgCl2 in THF at ambient tem-
perature. Filtration of the insoluble Et3N·HCl salt yielded a
solution of the magnesium salt of 5, which could be directly
used in the addition of 1,4-difluorophenyl Grignard. A practical
metalation of 1,4-difluorobenzene was also seen as desirable
as 1,4-difluorobenzene is 3-fold less expensive per mole than
1-bromo-2,5-difluorobenzene.26 It was established that the
2-lithio species is best prepared (95% assay yield)23 by the
dropwise addition of 1,4-difluorobenzene to an equimolar
solution of 1.1 equiv n-BuLi and TMEDA in THF at -70 °C.27
When the order of addition was reversed (85% assay yield),
the solution became very dark and viscous, which we attributed
to benzyne formation. Other solvents (t-BuOMe, 1,2-dimethox-
yethane) and additives (DABCO, pentamethyldiethylenetri-
amine) were significantly less effective for the metalation
reaction. Above -60 °C the 2-lithio-1,4-difluorobenzene species
begins to degrade, but below this temperature it appears
reasonably stable. When 2.2 equiv of the resulting 2-lithio-1,4-
difluorobenzene solution was added to a cold solution of 5, the
yield of 4a/b was disappointingly low (35-45%). Subsequently,
we found that a similar addition of 2-lithio-1,4-difluorobenzene
to the magnesium salt of 5 (as generated from Et3N/MgCl2)
resulted in much better yields (up to 85%). Interestingly,
increasing the stoichiometry of the aryllithium relative to 5
enhanced the yield (Table 1, entries 1-4), even though in all
cases the excess lithium reagent could be accounted for by
HPLC assay. We speculate that this phenomenon might be the
result of aryl lithium equilibration with the magnesium salt (i.e.,
ArLi + RCO2MgX f RCO2MgAr + LiCl) rather than adding
into the ketone moiety. The stereoselectivity in the addition of
the aryl lithium reagent to the ketone shifted toward more
equatorial attack when compared to the corresponding Grignard
addition (1:1.4 vs 3.6:1 for 4a and 4b, respectively). It is
believed that steric factors might explain this.28
Diastereoselective Synthesis of Sulfide 8. The key trans-
formation of our synthesis entails the diastereoselective reaction
of 4-CTP with a mixture of 4a and 4b to provide the desired
sulfide diastereomer 8 (Scheme 6). A large number29 of Brønsted
and Lewis acids were screened. The best stereoselectivity was
obtained with EtAlCl2 (a disappointing 60:40 dr at best), while
the best chemical conversion was observed with BF3 etherate
(generating a 1:1 mixture of 8 and 9 in 93% isolated yield).
It was noted that styrene derivative 14 was invariably
generated in these reactions. Similarly, we found that exposure
of either pure 8 or 9 to the reaction conditions provided similar
mixtures of 8, 9, 14, and 4-CTP. This provided unequivocal
proof for the reversible formation and cleavage of the critical
carbon-sulfur bond under these conditions. With these results
in hand, we shifted our attention toward the development of a
crystallization-induced process. Solubility experiments30 (Table
2) quickly showed that the desired sulfide 8 was less soluble
than its diastereoisomer 9 in all solvent systems examined.
Several physical properties of 8 and 9 as well as of their 1:1
mixture were determined. The melting points of 8 and 9 are
144 and 127 °C, respectively, which correlates with the solubility
(20) For example, 13 is formed at ca. 1% at -45 °C, 4% at -25 °C,
and 10% at 0 °C. 1H NMR analysis of the magnesium salt (prepared by the
addition of 0.98 equiv of a 2 M THF solution of i-PrMgCl, to a THF solution
of 5 followed by concentration in Vacuo and dissolution in d6-DMSO)
showed it to be stable at ambient temperature for at least 16 h.
(21) Residual ketoacid was always seen in the crude reaction mixtures
after work-up.
(22) Selected references on 4-fluorobenzyne: (a) Liu, H.; Zhang, X.;
Wang, C.; Guo, W.; Wu, Y.; Yang, S. J. Phys. Chem. A 2004, 108, 3356.
(b) Maurin, P.; Ibrahim-Ouali, M.; Parrain, J.-L.; Santelli, M. THEOCHEM
2003, 637, 91. (c) Dkhar, P. G. S.; Lyngdoh, R. H. D. Proc. Indian Acad.
of Sci., Chem. Sci. 2000, 112, 97. (d) Langenaeker, W.; De Proft, F.;
Geerlings, P. J. Phys. Chem. A 1998, 102, 5944.
(23) For a detailed study on the stability of 2-lithio-1,4-difluorobenzene
and the corresponding Grignard reagent, see Scott, J. P.; Brewer, S. E.;
Davies, A. J.; Brands, K. M. J. Synlett 2004, 9, 1646.
(24) The physical form of the cerium chloride used in the transmetalation
step is crucial in ensuring high yields. For a detailed investigation, see
Conlon, D. A.; Kumke, D.; Moeder, C.; Hardiman, M.; Hutson, G.; Sailer,
L. AdV. Synth. Catal. 2004, 346, 1307.
(25) The increasing proportion of equatorial vs axial attack on cyclo-
hexananones, with increasing steric nucleophilic bulk has been noted, see
(a) Eliel, E. L.; Wilen, S. H.; Mander, L. N. Stereochemistry of Organic
Compounds; Wiley-Interscience: New York, 1994; pp 735-736, 880-887.
(b) Reetz, M. T.; Haning, H.; Stanchev, S. Tetrahedron Lett. 1992, 33,
6963. (c) Molander, G. A.; Burkhardt, E. R.; Weinig, P. J. Org. Chem.
1990, 55, 4990. (d) Paquette, L. A.; Ra, C. S. J. Org. Chem. 1988, 53,
4978. (e) Reetz, M. T.; Kyung, S. H.; Hullman, M. Tetrahedron 1986, 42,
2931.
(28) The aryllithium reagent is most likely complexed with TMEDA in
our reaction. Axial addition predominates in the addition of phenyllithium
to 4-tert-butylcyclohexanone, see Ashby, E. C.; Noding, S. R. J. Org. Chem.
1979, 44, 4371.
(29) The following acid catalysts were screened in the reaction of
4-chlorothiophenol with 4a and 4b in dichloromethane: Sc(OTf)3, Ti(OEt)4,
BF3·Et2O, Et2AlCl, Zn(OTf)2, ZnI2, SnCl2, SnCl4, TiCl4, InCl2, InCl3,
LiClO4, B(OMe)3, BEt3, EtAlCl2, MeAlCl2, AlCl3, AlBr3, MeSO3H, CF3-
SO3H, (CF3SO2)2NH. Phenylthio ethers have been prepared previously from
tertiary alcohols using ZnI2 as the Lewis acid, see Guindon, Y.; Frenette,
R.; Fortin, R.; Rokach, J. J. Org. Chem. 1983, 48, 1357.
(26) Based on survey of potential commercial suppliers.
(27) 2-Lithio-1,4-difluorobenzene can also be prepared by halogen-metal
exchange using n-BuLi according to the procedure of Butler, I. R.; Lindsell,
W. E.; Preston, P. N. J. Chem. Res. Synop. 1981, 7, 185.
(30) A total of 47 solvents and solvent combinations were screened.
J. Org. Chem, Vol. 72, No. 13, 2007 4867

Davies et al.
a
SCHEME 6
a Conditions: 1.4 equiv of 4-CTP, 1.3 equiv of BF3‚Et2O, CH2Cl2, 0 °C.
TABLE 2. Solubility of Sulfide Diastereomers 8 and 9 in Various
Solvents
after stirring overnight the ratio of 8 and 9 had increased to
6.7:1, respectively (entry 1). In addition, approximately 11%
of 14 was also formed. Warming the mixture to 40 °C and
addition of more 4-CTP (2 mol equiv) shifted the equilibrium
further toward 8, resulting in a 7.5:1 ratio of 8 to 9 (entry 2).
Changing the solvent to n-heptane reduced the solubility further
and enabled a reduction in the amount of acid required,
providing an excellent 96:4 ratio for 8 to 9 (entry 3).
solubility, mg/mLa
solubility ratio
entry
solvent
n-heptane
methylcyclohexane
2,2,2-trifluoroethanol
hexafluorobenzene
R,R,R-trifluorotoluene
1,1,1,3,3,3-hexafluoro-
2-propanol
9
8
9:8
1
2
3
4
5
6
0.58
1.7
2.5
5.5
0.46
1.3
1.6
4.4
8.1
1.26:1
1.31:1
1.56:1
1.25:1
1.32:1
1.30:1
10.7
In spite of these encouraging results it was apparent during
scale-up that uncontrolled crystallization of undesired 9 was
occurring, thus lowering our selectivity. Methanesulfonic acid
was identified as the likely culprit, as this acid is virtually
immiscible in this system resulting in some phase separation.
Switching to triflic acid overcame this limitation to some extent.
This stronger acid is not necessarily more miscible, but only
0.1 mol equiv was required to effect comparable conversions,
while also enabling a reduction in the amount of added 4-CTP
(Table 3, entries 4 and 5). Nevertheless, scale-up still proved
difficult, as significant gumming of the solids occurred under
these conditions. To gain more insight into the system, the
solubilities of 8 and 9 were determined as a function of the
temperature, both in the absence and presence of 4-CTP (Figure
1). As expected, the addition of 4-CTP increased the absolute
solubilities of 8 and 9, but intriguingly the relative solubilities
begin to diverge significantly above 60 °C. With these data in
hand, the 1:1 mixture of 8 and 9 and 2 equiv of 4-CTP in
heptane was warmed to 63 °C before 10 mol % triflic acid was
added. When the resulting suspension was stirred at this
temperature overnight and then allowed to cool to ambient
temperature, a 97:3 dr of 8 and 9, respectively, was obtained.
The system was further optimized via addition of 5 vol % of
toluene, to effect a slight increase in solubility of all components.
The resulting process proved scaleable and afforded 8 in a
reproducible 90% isolated yield with >99:1 dr.
The presence of 14 in all the reaction mixtures alerted us to
the fact that it might actually serve as the initial reactant in this
process. Dehydration of a mixture of 4a and 4b with excess
TFA in CH2Cl2 provided pure 14 in an unoptimized 65%
isolated yield. A similar screening process, as described above,
identified a second generation process whereby a 1:1 mixture
of 1,1,1,3,3,3-hexafluoro-2-propanol and perfluorinated bu-
tyltetrahydrofuran (FC-75)32 was optimal for the crystallization-
induced transformation. Thus, 14 and 1.1 equiv of 4-CTP were
dissolved in the mixed solvent, and 10 mol % of triflic acid
was added at 0 °C. Sulfides 8 and 9 are rapidly formed as a 1:1
mixture of solids within minutes. Further stirring (16 h) of this
mixture at ambient temperature led to the desired turnover, and
8 was produced in 95% yield with 98.5:1.5 dr.
13.6
22.8
10.5
7
acetonitrile
15.0
1.52:1
a Determined by HPLC assay.
TABLE 3. First Generation Crystallization-induced Process
HPLC assay yield, %
solvent and
entry
acid (equiv)
additive (equiv)
14
9
8
1
2
MeSO3H (2)
MeSO3H (2)
c-C6H11Me
c-C6H11Me,
4-ClPhSH (2)
n-heptane,
11
4
10
10
67
75
3
4
5
MeSO3H (0.6)
2
2
1
4
4
3
96
93
95
4-ClPhSH (2)
CF3SO3H (0.1) n-heptane,
4-ClPhSH (2)
CF3SO3H (0.1) n-heptane,
4-ClPhSH (0.5)
data above. The 1:1 mixture of 8 and 9 has a melting point of
109 °C. Pure 8 and 9 show distinct XRPD patterns, while the
XRPD of the 1:1 mixture can be interpreted as the sum of the
two components (even though the peaks are relatively weak).
The combination of these data strongly suggests that the mixture
of these diastereomers can be expected to behave in an
analogous manner to a racemic conglomerate.10a With suitable
crystallization and reversible carbon-sulfur bond-formation
conditions identified, the challenge became how to integrate
these into a single vessel, viable process.
It was established that relatively small amounts of a strong
Brønsted acid (e.g., methanesulfonic and trifluoromethane-
sulfonic acid) in apolar solvents such as n-heptane, methylcy-
clohexane, hexafluorobenzene, and trifluorotoluene were suf-
ficient for an effective equilibration.8 In more polar, Lewis basic
solvents like acetonitrile and trifluoroethanol, no reaction took
place. However, in the extremely polar and non-nucleophilic
1,1,1,3,3,3-hexafluoro-2-propanol, the cleavage was very fac-
ile.31 On the basis of these observations, we developed a first-
generation process, and salient results of our extensive screen
of conditions are summarized in Table 3. A 1:1 mixture of 8
and 9 (prepared in a separate step using BF3 etherate in
dichloromethane, as mentioned above) suspended in methyl-
cyclohexane was treated with 2 mol equiv of methanesulfonic
acid at ambient temperature. We were delighted to find that
While we were delighted with the discovery of these
unprecedented crystallization-induced processes, we were also
keenly aware that an extra step was required, in the first instance
to prepare the 1:1 mixture of 8 and 9, and in the second instance
(31) A solvent quoted as possessing a higher polarity than water, see
Reichardt, C. SolVents and SolVent Effects in Organic Chemistry; VCH (UK)
Ltd.: Cambridge, 1990; p 366.
(32) This solvent was chosen because it is miscible with 1,1,1,3,3,3-
hexafluoro-2-propanol, without enhancing the solubility of 8 and 9.
4868 J. Org. Chem., Vol. 72, No. 13, 2007

Crystallization-Induced Diastereomeric Transformation
a
SCHEME 7
a Conditions: 1.2 equiv of 4-CTP, 9.1 vol. of MeSO3H, 0.9 vol. of H2O,
40 °C.
a
SCHEME 8
FIGURE 1. Variable temperature solubility curve of sulfides 8 and 9
with and without added 4-CTP in n-heptane.
a Conditions: 3.1 equiv of 27.5 wt % H2O2, AcOH, 50 °C.
Oxidation of 8 to 1. With all the key bonds of our target in
place, we were left with an oxidation of a sulfide to sulfone. In
our previous synthesis of 1, we used a sodium tungstate-
catalyzed hydrogen peroxide oxidation for a similar reaction.4
Application of these conditions to the oxidation of 8 was less
successful, resulting in poor reaction profiles and incomplete
conversion. However, oxidation of 8 to 1 could be achieved
much more cleanly, utilizing 27% hydrogen peroxide in acetic
acid.35 The temperature at which this oxidation sequence is
carried out proved critical. As expected, the oxidation to the
corresponding sulfoxide intermediate 16 is rapid (∼1 h at 50
°C), whereas the second oxidation to the sulfone requires an
overnight age. When we attempted to accelerate the reaction
by performing it above 60 °C, we obtained a relatively low yield
(84%) for isolated 1. A series of control experiments showed
that at these temperatures, 16 is relatively unstable and
eliminates 4-chlorophenylsulfenic acid with concomitant forma-
tion of styrene 14. The latter is rapidly oxidized to a series of
unidentified products which are rejected in the crystallization
of 1. When the temperature is maintained in the 45-50 °C
range, this decomposition pathway is suppressed and 1 can be
isolated in 94% yield after crystallization (Scheme 8).
the dehydration to obtain 14. It seemed obvious that the ultimate
synthesis would use hydroxyacids 4a and 4b directly and
generate 14 in situ. Indeed, development of such a third
generation process proved to be feasible. Dissolution of 4a and
4b in MeSO3H resulted in rapid dehydration to 14. Addition of
a slight excess (1.2 equiv) of 4-CTP rapidly provided a 1:1
mixture of 8 and 9. When an equal volume of water was added
to the homogeneous reaction mixture, solids were produced
immediately which, upon filtration, proved to be a mixture of
8 and 9 in a 53:47 dr with no appreciable styrene 14 content.33
It was quickly discovered that the quantity of added water is
critical. Insufficient water makes the medium too solubilizing,
thus thwarting a productive crystallization. A surfeit of water
reduces the solubility of 8 and 9 too much, thus preventing a
productive turnover. Within a narrow range of optimum water
amounts (9-13 vol %), equilibration can occur in the solution
phase while most of the products are out of solution. Under
these conditions it is just a matter of time before thermodynamic
equilibrium is reached, producing predominantly crystalline 8.
Temperature proved another critical parameter for similar
reasons, allowing further optimization of the process.
In a direct comparison on multigram scale it became apparent
that the mode of agitation is also important in this process.
Magnetic stirring provided better selectivity than overhead
agitation (97:3 vs 94:6 dr). Obviously, on larger scale magnetic
stirring is not practical, but the slightly lower selectivity resulting
from overhead stirring can easily be corrected by a recrystal-
lization of the isolated solids (Vide infra). Thus, in the fully
optimized third generation procedure, 4a and 4b are added to a
solution of 1.1 equiv of 4-CTP in MeSO3H containing 9 vol %
of water. After overnight stirring at 40 °C, dilution with 9
volumes of water, and filtration, the desired sulfide 8 can be
obtained in 95% yield and 94:6 dr. The solid typically also
contains 1% of thioester 15 (Scheme 7). This process proved
to be scaleable and reproducible. In order to reduce the undesired
diastereomer 9 as well as the other impurities to an insignificant
level, the product was recrystallized from acetonitrile, which
provided pure 8 (99.9:0.1 dr) with 92% yield recovery.34
Conclusions
The γ-secretase inhibitor 1 can be synthesized starting from
6 in 58% overall yield using only five chemical steps. This
remarkably efficient synthesis does not require the use of any
protecting groups. The cornerstone of this new approach to 1
is a practical crystallization-driven transformation of a mixture
of tertiary benzylic alcohols into the desired sulfide diastereomer
with 94:6 dr. This unprecedented process is based upon a
reversible, acid-catalyzed carbon-sulfur bond formation. In
comparison to our two previously reported syntheses of 1,4b the
synthesis reported herein is significantly shorter, more efficient,
and represents the potential basis for the manufacture of this
compound.
Experimental Section
3-(4-Oxocyclohexyl)propanoic Acid (5).15a Step A. To a 40 L
autoclave were charged p-coumaric acid 6 (3.33 kg, 20.0 mol),
isopropyl acetate (iPAc) (30 L), and 5% Rh/Al2O3 (0.248 kg). The
(33) Thioethers have been prepared using dodecylbenzenesulfonic acid-
catalyzed dehydration of alcohols, see Manabe, K.; Iimura, S.; Sun, X.-M.;
Kobayashi, S. J. Am. Chem. Soc. 2002, 124, 11971.
(34) The structures of 8 and 9 were confirmed by single-crystal X-ray
crystallography. See Supporting Information for details.
(35) ComprehensiVe Organic Synthesis; Trost, B. M.; Fleming, I.; Ley,
S. V., Eds.; Pergamon Press: Oxford, 1993; vol. 7, p 766 and references
therein.
J. Org. Chem, Vol. 72, No. 13, 2007 4869

Davies et al.
1
vessel was placed under 11.2 bar of hydrogen and heated at 80 °C
for 18 h. The reaction mixture was cooled to ambient temperature
and filtered through solka floc (500 g) using iPAc (2 × 3 L) as a
rinse. The combined filtrates (containing a 1.2-1.4-1 mixture of
10a and 10b, respectively, at 89% yield according to GC analysis)
were used directly in the next step.
rophenyl)-4-hydroxycyclohexyl]propanoic Acid (4b): H NMR
(500 MHz, CD3OD): δ 7.32 (ddd, J ) 9.9, 6.7, 3.6 Hz, 1H), 7.04
(ddd, J ) 11.3, 8.7, 4.5 Hz, 1H), 6.96 (dddd, J ) 8.8, 7.3, 3.4, 3.4
Hz, 1H), 2.36 (m, 2H), 2.33 (t, J ) 7.6 Hz, 2H), 2.00 (m, 2H),
1.76 (q, J ) 7.6 Hz, 2H), 1.68 (m, 1H), 1.50 (dt, J ) 14.1, 4.5 Hz,
2H), 1.36 (dq, J ) 13.7, 4.5 Hz, 2H). 1H NMR (500 MHz,
CDCl3): δ 7.24 (ddd, J ) 9.7, 6.5, 3.3 Hz, 1H), 7.00 (ddd, J )
11.5, 9.0, 4.7 Hz, 1H), 6.92 (dddd, J ) 8.8, 7.3, 3.6, 3.6 Hz, 1H),
2.39 (t, J ) 7.6 Hz, 2H), 2.34 (dd, J ) 10.3, 3.7 Hz, 2H), 1.95 (m,
2H), 1.74 (q, J ) 7.4 Hz, 2H), 1.68 (m, 1H), 1.62 (m, 2H), 1.33
(m, 2H); 13C NMR (125 MHz, CDCl3): δ 180.0, 158.9 (dd, J )
243.1, 4.0 Hz), 156.8 (d, J ) 239.1 Hz), 136.3 (dd, J ) 12.9, 6.4
Hz), 117.6 (dd, J ) 27.1, 8.6 Hz), 115.1 (dd, J ) 24.0, 9.8 Hz),
114.6 (dd, J ) 25.3, 5.5 Hz), 73.0 (d, J ) 3.0 Hz), 33.6, 33.3,
32.5, 27.5, 26.9; IR ν (cm-1) 2924, 2872, 1692, 1486, 1405, 1181,
1162,; mp 138-139 °C; Anal. Calcd for C15H18F2O3; C, 63.37; H,
6.38; F, 13.37. Found: C, 63.31; H, 6.34; F, 13.27.
3-[trans-4-(2,5-Difluorophenyl)-4-hydroxycyclohexyl]propano-
ic Acid (4a). 1H NMR (500 MHz, CD3OD): δ 7.35 (ddd, J ) 9.9,
6.4, 3.1 Hz, 1H), 7.01 (ddd, J ) 11.3, 8.9, 4.6 Hz, 1H), 6.94 (dddd,
J ) 8.8, 7.4, 3.6, 3.6 Hz, 1H), 2.35 (t, J ) 7.6 Hz, 2H), 2.21 (td,
J ) 13.5, 3.8 Hz, 2H), 1.70 (br d, J ) 13.7 Hz, 2H), 1.63 (m, 2H),
1.59 (q, J ) 7.4 Hz, 1H), 1.51 (m, 2H), 1.41 (m, 2H); 13C NMR
(125 MHz, CD3OD): δ 178.0, 160.4 (dd, J ) 239.6, 1.9 Hz), 157.1
(dd, J ) 239.8, 2.2 Hz), 140.3 (dd, J ) 14.6, 6.2 Hz), 118.5 (dd,
J ) 28.0, 8.5 Hz), 115.5 (dd, J ) 24.3, 9.2 Hz), 115.1 (dd, J )
26.2, 5.3 Hz), 73.3 (d, J ) 5.3 Hz), 37.7, 37.0 (d, J ) 4.5 Hz),
33.5, 32.7, 29.2; IR ν (cm-1) 3391, 2937, 2868, 1707, 1481, 1416,
1277, 1266, 1183; mp 113-114 °C; Anal. Calcd for C15H18F2O3;
C, 63.37; H, 6.38; F, 13.37. Found: C, 63.36; H, 6.37; F, 13.29.
Method B: The Grignard reagent is prepared in the same way
as for method A. The magnesium salt solution is prepared as
follows: 3-(4-oxo-cyclohexyl)propanoic acid 5 (7 g, 41.1 mmol)
was dissolved in THF (56 mL) and degassed with N2 (3×) before
cooling to 4 °C. Magnesium chloride (3.92 g, 41.1 mmol) was added
in one portion. The resulting solution was aged for 5 min followed
by the dropwise addition of triethylamine (5.73 mL, 41.1 mmol)
over 15 min. The resulting suspension was aged for 1 h at ambient
temperature, cooled to 1 °C, and filtered under a nitrogen
atmosphere. The filter cake was washed with THF (20 mL). The
combined filtrates were used as in method A.
Method C: TMEDA (13.7 mL, 90.5 mmol) was dissolved in
THF (100 mL) and degassed (3×) with N2. The solution was cooled
to -65 °C, and n-butyl lithium (36.2 mL, 90.5 mmol) was added
over 10 min while maintaining the internal temperature below
-60 °C. The resulting pale yellow solution was aged for 10 min
below -60 °C. 1,4-Difluorobenzene (8.5 mL, 82.3 mmol) was
added over 15 min while maintaining the internal temperature below
-55 °C. The resulting solution was aged for 1 h at -60 °C. A
solution of the magnesium salt of 5 (prepared as in method B) was
added to the organolithium solution over 20 min while maintaining
the temperature below -55 °C. The reaction mixture was allowed
to warm to -30 °C over 1 h and then quenched with acetic acid
(9.5 mL) keeping the internal temperature below -20 °C. After
warming to 10 °C, 2 M HCl (190 mL) and toluene (190 mL) were
added to the reaction mixture. After stirring for 10 min and warming
to ambient temperature, the organic phase was separated and washed
with water (90 mL) to give a 1:1.4 mixture of 4a and 4b,
respectively, in 78% combined yield based on HPLC assay.
3-[4-(2,5-Difluorophenyl)cyclohex-3-en-1-yl]propanoic Acid
(14). Diastereomeric alcohols 4a and 4b (2.5 g, 8.79 mmol) were
suspended in dichloromethane (39 mL), and the resulting solution
was cooled to -2 °C. Trifluoroacetic acid (10.03 g, 87.9 mmol)
was added over 1 h while maintaining internal temperature below
5 °C. The resulting suspension was allowed to warm to ambient
temperature, whereupon it gradually became homogeneous. The
reaction was quenched by the addition of water (50 mL). The
solution was then extracted with iPAc (150 mL). The organic layer
was washed with water (50 mL) and brine (50 mL), dried with
Step B: To the iPAc solution from step A were charged RuCl3
(42 g) and acetic acid (5.6 kg), and the reaction mixture was cooled
to 5 °C. Sodium hypochlorite (10-13 w/v%, 26 kg, 23.76 mol)
was added to this solution below the surface over 1.5 h, while
maintaining the internal temperature in the 2-8 °C range. The
resulting biphasic mixture was stirred in the 2-5 °C temperature
range for 3 h until all of 10a and 10b had been consumed. The
reaction mixture was quenched by the addition of a sodium bisulfite
solution (1.0 kg in 5 L water) over 5 min and stirred for a further
10 min, and then 3 M hydrochloric acid (6.6 L) was added. The
organic layer was separated, and the aqueous layer was extracted
with iPAc (2 × 30L). The combined organic layers were washed
with brine (5 L), dried over MgSO4, and then concentrated to a
total volume of 20 L while flushing with fresh iPAc (2 × 25 L).
The resulting solution was then solvent switched to heptane (6 ×
18 L), resulting in the crystallization of 5. The product was filtered,
washed with heptane (2 × 5 L) at ambient temperature, and dried
to afford 3.07 kg of 5 (90%). 1H NMR (500 MHz, CDCl3): δ 2.42
(t, J ) 7.5 Hz, 2H), 2.43-2.29 (m, 4H), 2.06 (m, 2H), 1.77 (m,
1H), 1.67 (q, J ) 7.5 Hz, 2H), 1.41 (m, 2H); 13C NMR (125 MHz,
CDCl3): δ 212.4, 179.4, 40.7, 35.5, 32.42, 32.0, 30.4; IR ν (cm-1
)
2932, 2859, 1692, 1417, 1326, 1273, 1210, 1110; mp 72-73 °C
(lit. mp 55-58 °C); Anal. Calcd for C9H14O3; C, 63.51; H, 8.29.
Found: C, 63.49; H, 8.30.
3-[cis-4-(2,5-Difluorophenyl)-4-hydroxycyclohexyl]propano-
ic Acid (4b) and 3-[trans-4-(2,5-Difluorophenyl)-4-hydroxycy-
clohexyl]propanoic Acid (4a). Method A: 1-Bromo-2,5-difluo-
robenzene (157.3 g, 0.815 mol) was dissolved in THF (500 mL),
degassed with N2 (3×), and then cooled to -30 °C. A 2 M solution
of isopropylmagnesium chloride in THF (400 mL, 0.80 mol) was
then added dropwise over 35 min, while maintaining the temperature
below -10 °C. The resultant opaque solution was aged for 1 h at
-10 °C (the Grignard reagent is stable for around 4 h when the
temperature is maintained below -10 °C). In a separate flask, 5
(85.1 g, 0.50 mol) was dissolved in THF (1000 mL) and degassed
with N2 (3×) before cooling to -58 °C. A 2 M solution of
isopropylmagnesium chloride in THF (240 mL, 0.48 mol) was
added dropwise over 40 min while maintaining the temperature
below -55 °C. The resulting solution was aged for 30 min between
-55 and -50 °C, warmed to -30 °C over 15 min, and held at this
temperature until use (this magnesium salt solution is stable
1
according to H NMR analysis when stored oVernight at ambient
temperature). The solution of the magnesium salt was added to
the solution of the Grignard reagent via cannula over 15 min while
maintaining the internal temperature below -20 °C, followed by a
rinse with THF (50 mL). The resulting thick slurry was aged for
45 min, while maintaining the temperature below -10 °C. The
mixture was quenched by the addition of acetic acid over 5 min
(86 mL, 1.50 mol) while maintaining the temperature below 5 °C.
The mixture was aged for 10 min before warming to 10 °C,
whereupon 1 M HCl (750 mL) was charged to the reaction mixture
followed by toluene (750 mL). The organic layer was separated,
washed with water (1.25 L), and then extracted with 1 M NaOH
(1.25 L). Toluene (1.5 L) and 2 M HCl (700 mL) were added to
the aqueous extract. The mixture was agitated for 5 min. The
organic layer was separated, washed with water (2 × 1 L), and
then concentrated to dryness in Vacuo. The residue was recrystal-
lized from toluene/heptane to afford 124 g of 4a and 4b (87%) as
a 3.6:1.0 mixture, respectively. Pure diastereomers were obtained
by preparative SFC separation: (Chiral Technologies 2.1 cm
Chiralpak AD 20 µm column. 100 bar; 30% MeOH/CO2 @ 70.00
mL/min; 35 °C and 215 nm. Feedstock concn. ∼200 mg/mL in
IPA with 800 µL injections every 120 s). 3-[cis-4-(2,5-Difluo-
4870 J. Org. Chem., Vol. 72, No. 13, 2007

Crystallization-Induced Diastereomeric Transformation
Na2SO4, and concentrated to dryness in Vacuo. The crude product
was recrystallized from iPAc to afford 1.39 g of 14 (65%). 1H NMR
(400 MHz, CDCl3) δ 11.1 (br.s, 1H), 7.05-6.75 (m, 3H), 2.55-
2.25 (m, 5H), 1.95-1.65 (m, 5H), 1.45-1.35 (1H, m). 13C NMR
(100 MHz, CD2Cl2) δ 180.6, 158.5 (d, J ) 260 Hz), 155.9 (d, J )
241 Hz), 132.7, 132.3 (dd, J ) 7.7, 16.3 Hz), 128.1, 116.6 (dd, J
) 8.7, 26.1 Hz), 115.4 (dd, J ) 5.1, 23.8 Hz), 114.1 (dd, J ) 8.7,
23.8 Hz), 32.4, 32.0, 31.7, 31.0, 28.7, 28.2 (d, J ) 3.2 Hz); IR ν
(cm-1) 2915, 2859, 1699, 1584, 1493, 1479, 1415, 1324, 1279,
1232, 1207, 1170; mp 78-79 °C; Anal. Calcd for C15H16F2O2; C,
67.66; H, 6.06; F, 14.27. Found: C, 67.48; H, 6.03; F, 14.15.
3-[cis-4-[(4-Chlorophenyl)sulfanyl]-4-(2,5-difluorophenyl)cy-
clohexyl]propanoic Acid (8). 4-Chlorothiophenol was slurried in
methanesulfonic acid (1070 mL) at ambient temperature. The solids
were solubilized by warming to 38 °C. Water (118 mL) was added
over 45 min while maintaining the temperature below 40 °C. A
mixture of 4a and 4b (106.8 g, 0.376 mol) was added to the reaction
mixture in one portion. After 10 min the mixture was seeded with
sulfide 8 (1 g) and stirred between 36 and 42 °C overnight. After
cooling to ambient temperature, water (1.07 L) was added over 50
min while maintaining the temperature below 40 °C. The suspension
was then cooled to ambient temperature, aged for 30 min, and
filtered. The filter cake was washed with water (5 L) and dried in
vacuo at 40 °C overnight to afford 152 g of 8 (95%, containing
6% of the trans diastereomer 9). Recrystallization from CH3CN
affords >99.9:0.1 dr. 3-[cis-4-[(4-Chlorophenyl)sulfanyl]-4-(2,5-
min run time). 3-[trans-4-[(4-chlorophenyl)sulfanyl]-4-(2,5-dif-
luorophenyl)cyclohexyl]propanoic Acid (9): 1H NMR (500 MHz,
C6D6): δ 12.28 (br s, 1H), 6.86 (m, 2H), 6.82 (m, 2H), 6.54-6.49
(m, 2H), 6.43 (m, 1H), 2.48 (br s, 2H), 1.93 (t, J ) 7.5 Hz, 2H),
1.57 (td, J ) 13.4, 2.5 Hz, 2H), 1.30 (d, J ) 13.4 Hz, 2H), 1.08 (q,
J ) 7.5 Hz, 2H), 0.98 (m, 1H), 0.63 (m, 2H). 13C NMR (125 MHz,
C6D6): δ 181.2, 159.0 (dd, J ) 241.3, 1.8 Hz), 158.3 (dd, J )
247.6, 2.4 Hz), 139.2, 136.3, 131.5 (dd, J ) 11.1, 6.5 Hz), 130.4,
129.1, 118.5 (dd, J ) 28.2, 8.5 Hz), 117.0 (dd, J ) 24.9, 5.1 Hz),
115.4 (dd, J ) 23.8, 9.6 Hz), 54.7 (d, J ) 3.9 Hz), 39.2 (br s),
36.8, 32.0, 31.4, 30.0; IR ν (cm-1) 2923, 2850, 1697, 1571, 1492,
1474, 1412, 1181; mp 126-127 °C; Anal. Calcd for C21H21-
ClF2O2S; C, 61.38; H, 5.15. Found: C, 61.37; H, 4.91. The
structures of 8 and 9 were confirmed by single-crystal X-ray
crystallography; see Supporting Information for details.
3-[cis-4-[(4-Chlorophenyl)sulfonyl]-4-(2,5-difluorophenyl)cy-
clohexyl]propanoic Acid (1). Sulfide 8 (119 g) was dissolved in
acetic acid (965 mL) at 50 °C. A hydrogen peroxide solution (27.5
wt %, 110 mL) was added dropwise over 10 min. The reaction
mixture was maintained at this temperature overnight. The solution
was cooled to ambient temperature, and water (220 mL) was added
dropwise over 10 min until the solution became turbid. At this point,
authentic 1 (8.6 g) was added as a seed and the resulting slurry
was aged for 20 min, before additional water (1710 mL) was added
over 40 min. After stirring at ambient temperature for 1 h, the
crystals were collected by filtration, washed with water (1930 mL),
and dried in Vacuo at 40 °C to afford 129.5 g of 1 (94% yield). 1H
and 13C NMR and HPLC data for 1 were in full accord with those
that have been previously reported.4b
1
difluorophenyl)cyclohexyl]propanoic Acid (8): H NMR (500
MHz, C6D6): δ 6.80 (m, 2H), 6.74 (m, 2H), 6.58 (m, 1H), 6.49-
6.43 (m, 2H), 2.18-2.14 (m, 4H), 1.57 (m, 2H), 1.51 (q, J ) 7.5
Hz, 2H), 1.33-1.27 (m, 4H), 0.95 (m, 1H); 13C NMR (125 MHz,
C6D6): δ 181.0, 158.8 (dd, J ) 241.6, 2.0 Hz), 158.0 (dd, J )
247.6, 2.4 Hz), 138.3, 136.6 (dd, J ) 10.6, 6.5 Hz), 135.9, 131.1,
129.2, 118.1 (dd, J ) 27.5, 8.7 Hz), 115.4 (dd, J ) 25.2, 4.9 Hz),
115.0 (dd, J ) 23.8, 9.4 Hz), 54.7 (d, J ) 4.1 Hz), 36.6, 34.7 (d,
J ) 4.6 Hz), 32.0, 32.0, 28.4; IR ν (cm-1) 2924, 2855, 1698, 1584,
1484, 1474, 1411, 1290, 1187, 1091; mp 144-145 °C; Anal. Calcd
for C21H21ClF2O2S; C, 61.38; H, 5.15; F, 9.25; Cl, 8.63; S, 7.80.
Found: C, 61.29; H, 5.17; F, 9.20; Cl, 8.69; S, 7.62. Pure 9 was
obtained by preparative SFC separation: (Chiralpak AD column.
200 Bar; 4% MeOH for 4 min then ramp to 40% MeOH @ 2%
per min with a 3 min hold at 40% MeOH; 1.5 mL/min; 35 °C; 25
Acknowledgment. We thank Andrew Kirtley for performing
XRPD and DSC analyses, Alexia Bertrand for IR spectra
determination and Jennifer Chilenski for obtaining X-ray
crystallographic data.
Supporting Information Available: X-ray crystallographic
(CIF) files and ORTEP plots for 8 and 9. 1H and 13C NMR spectra
for all new compounds. This material is available free of charge
via the Internet at http://pubs.acs.org.
JO0705925
J. Org. Chem, Vol. 72, No. 13, 2007 4871