Design, development, and scale-up of a selective meso-epoxide desymmetrization process

Article abstract of DOI:10.1021/op060225f

A pilot-plant scale desymmetrization of the cyclic meso-epoxide 4b, using a chiral lithium amide prepared from symmetrical diamine 17, was designed and implemented to provide allylic alcohol 3b in high yield and greater than 99% ee. This chiral alcohol was converted to ketone 2b, a key intermediate in a new asymmetric synthesis of LY459477. Chiral diamine 17 was prepared from a readily available chiral precursor, (R)-α-methylbenzylamine, and could be recovered from the reaction mixture and reused. Studies performed to probe the mechanism of the rearrangement reaction of epoxide 4b showed that diamine 17 provided an optimal combination of selectivity and scaleability for this process.

Full text of DOI:10.1021/op060225f

Organic Process Research & Development 2007, 11, 546−559
Design, Development, and Scale-Up of a Selective meso-Epoxide
Desymmetrization Process
David L. Varie,* Christopher Beck, Sandra K. Borders, Molly D. Brady, Jason S. Cronin, Tracy K. Ditsworth,
David A. Hay, David W. Hoard, Richard C. Hoying, Ryan J. Linder, Richard D. Miller, Eric D. Moher,
Jacob R. Remacle, and John A. Rieck, III
Chemical Product Research and DeVelopment, Eli Lilly and Co., Indianapolis, Indiana 46285, U.S.A.
David D. Anderson, Paul N. Dodson, Mindy B. Forst, Duane A. Pierson, and Joseph A. Turpin
Analytical Sciences Research and DeVelopment, Eli Lilly and Co., Indianapolis, Indiana 46285, U.S.A.
Abstract:
subsequent diastereoselective reactions to convert 2 to the
drug substance 1 will be described in a separate report.
A pilot-plant scale desymmetrization of the cyclic meso-epoxide
4b, using a chiral lithium amide prepared from symmetrical
diamine 17, was designed and implemented to provide allylic
alcohol 3b in high yield and greater than 99% ee. This chiral
alcohol was converted to ketone 2b, a key intermediate in a
new asymmetric synthesis of LY459477. Chiral diamine 17 was
prepared from a readily available chiral precursor, (R)-r-
methylbenzylamine, and could be recovered from the reaction
mixture and reused. Studies performed to probe the mechanism
of the rearrangement reaction of epoxide 4b showed that
diamine 17 provided an optimal combination of selectivity and
scaleability for this process.
Results and Discussion
At the outset of this work we were encouraged by the
report of Hodgson3 on the conversion of cyclopentene
epoxide 4a to alcohol 3a with 33% ee using n-BuLi/(-)-
sparteine, albeit in low yield (Scheme 2). Several reports of
the preparation of enantioenriched cyclopentenols using
lithium amides derived from chiral diamines led us to initiate
screening studies with epoxide 4a.4 To evaluate the re-
arrangement in a racemic case, we treated epoxide 4a with
3 equiv of LDA in THF at 0 °C/2 h. The desired allylic
alcohol 3a was isolated in 37% yield. Amide 6, resulting
from reaction of the ester with LDA, was also obtained in
25% yield (Scheme 3). To limit this undesired displacement
reaction we prepared tert-butyl ester 4b (vide infra) and again
treated it with LDA under the same conditions. Alcohol 3b
was obtained in 81% crystallized yield; no formation of
amide 6 was observed. Thus, epoxide 4b, with a tert-butoxy
carbonyl (Boc) amine protecting group and the tert-butyl
ester, was used in subsequent studies.
An initial screen of 25 chiral amines, 1,2-diamines, and
1,2-amino alcohols, was performed as follows. Epoxide 4b
in THF was added to a 0 °C THF solution containing three
equivalents of the lithium amide, diamide, or amide/alkoxide
(prepared by treating the chiral amine with one or two
equivalents of n-BuLi/equivalent of amine). The reaction
mixtures were allowed to warm to room temperature and
were stirred 18 h. Alcohol 3b was isolated by an aqueous
HCl/MTBE workup and purified via chromatography on
Introduction
The preparation of enantiomerically enriched allylic
alcohols via desymmetrization of meso-epoxides with chiral
lithium amide reagents is a well-studied asymmetric rear-
rangement.1 In the course of designing an asymmetric
synthesis of the metabatropic glutamate (mGlu2/3) receptor
agonist LY459477 (1),2 this transformation was evaluated
as a key element of a potential commercial synthesis for this
stereochemically dense glutamic acid analogue. Shown
retrosynthetically in Scheme 1, we envisioned the synthesis
of cyclopentene meso-epoxide 4 from a protected glycine
equivalent followed by a desymmetrization of 4 with a chiral
lithium amide. Simple oxidation would then provide the key
intermediate, cyclopentenone 2.
This report describes the pilot-scale synthesis of epoxide
4, design and scale-up of the process for the conversion of
4 to enantiomerically enriched allylic alcohol 3, and the
preparation of the key intermediate cyclopentenone 2. The
(3) Hodgson, D. M.; Thompson, A. J.; Wadman, S.; Keats, C. J. Tetrahedron
1999, 55, 10815.
* To whom correspondence should be addressed. E-mail: varie@lilly.com.
(1) For reviews see: (a) O’Brien, P. J. Chem. Soc., Perkin Trans. 1 1998, 1439.
(b) Cox, P. J.; Simpkins, N. S. Tetrahedron: Asymmetry 1991, 2, 1. (c)
Eames, J. Eur. J. Org. Chem. 2002, 393. (d) Hodgson, D. M.; Gibbs, A.
R.; Lee, G. P. Tetrahedron 1996, 52, 14361. (e) Magnus, A.; Bertilsson, S.
K.; Andersson, P. G. Chem. Soc. ReV. 2002, 31, 223. For general reviews
see also: (f) Crandall, J. K.; Apparu, M. Org. React. 1983, 29, 245. (g)
Satoh, T. Chem ReV. 1996, 96, 3303.
(4) For example see: (a) Asami, M.; Ogawa, M.; Inoue, S. Tetrahedron Lett.
1999, 40, 1563. (b) O’Brien, P.; Towers, T. T.; Voith, M. Tetrahedron Lett.
1998, 39, 8175. (c) Hodgson, D. M.; Gibbs, A. R. Tetrahedron: Asymmetry
1996, 7, 407. (d) Bhuniya, D.; DattaGupta, A.; Singh V. K. J. Org. Chem.
1996, 61, 6108. (e) Hodgson, D. M.; Gibbs, A. R.; Drew, M. G. B. J. Chem.
Soc., Perkin Trans. I 1999, 3579. (f) Milne, M.; Murphy, P. J. J. Chem.
Soc., Chem. Commun. 1993, 884. (g) Hodgson, D. M.; Witherington, J.
Moloney, B. A. J. Chem. Soc., Perkin Trans. I 1994, 3373. (h) Asami, M.;
Takahashi, J.; Inoue, S. Tetrahedron: Asymmetry 1994, 5, 1649. (i)
Sodergen, M. K.; Bertilsson, S. K.; Andersson, P. G. J. Am. Chem. Soc.
2000, 122, 6610.
(2) (a) Moher, E. D.; Monn, J. A.; Pedregal-Tercero, C.; Collado, C. I.; Blanco-
Urgoiti, J. G. PCT Int. Appl. WO 2003104217, 2003. (b) Massey, S. M.;
Monn, J. A.; Valli, M. J. Eur. Pat. Appl. 878463, 1998.
546
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development
10.1021/op060225f CCC: $37.00 © 2007 American Chemical Society
Published on Web 02/01/2007

Scheme 1
Scheme 2. Rearrangement of epoxide 4a with n-BuLi/
5). Diamine 13 dihydrochloride salt was recovered from the
reaction mixture by filtration following the quench of the
reaction with aqueous HCl. This salt was converted to the
free base and crystallized from ethanol to yield diamine 13
in 80% overall yield.
(-)-sparteine3
This process provided intermediates for the study of
downstream chemistry, but it was not initially viewed as ideal
for commercial development. Although diamine 13 could
be recovered from the reaction and reused, its preparation
required a low-yielding, cryogenic Grignard reaction (Scheme
4).9 In addition the rearrangement reaction required a
relatively large mass of diamine 13 (3.5 kg/kg of epoxide
4b). However, attempts to reduce the amount of 13 needed,
potentially to catalytic amounts,10 provided valuable insight
into the reaction. Data to accompany the following discussion
is presented in Table 2. In principle, only one equivalent of
lithium diamide should be required to both deprotonate the
NHBoc group and effect the rearrangement of 4b. In the
event, adding epoxide 4b to 1.2 equiv of dilithium-13 at 0
°C gave incomplete conversion (after 21 h) and produced
alcohol 3b with diminished (83%) enantiomeric excess (entry
2). Pretreatment of epoxide 4b with 1.2 equiv of LDA at
-10 °C followed by reaction with 1.2 equiv of dilithium-13
did consume all epoxide starting material but still provided
3b with decreased enantiomeric purity (89% ee). This was
presumably due to a competing background rearrangement
reaction with LDA. Attempts to deprotonate the NHBoc of
epoxide 4b with one equivalent of lithium reagents (n-BuLi,
s-BuLi, or PhLi) at -78 °C resulted in substantial formation
of the expected ketone and carbinol products resulting from
alkyl/aryl lithium addition to the tert-butyl ester. However,
reaction of 3.3 equiv of 2-lithio-1,3-dimethoxybenzene11 with
1.1 equiv of diamine 13, followed by addition of a THF
solution of epoxide 4b provided a 70% yield of alcohol 3b
with 90% ee (entry 5). No products resulting from addition
of the aryl lithium to the ester were observed. An attempted
catalytic reaction, in which epoxide 4b was treated with 0.2
equiv of dilithium-13 followed by 2.2 equiv of LDA, gave
essentially racemic 3b (entry 4). Again, this is presumably
Scheme 3. Rearrangement of epoxides 4a,b with LDA
silica gel. A subset of the screening results is shown in Table
1.
Notably, the best enantioselectivities were obtained with
the symmetrical diamines 13 and 14, which gave alcohol
3b with 93 and 78% ee, respectively. Both diamines were
prepared from readily available (R)-R-methylbenzylamine
using modified literature procedures (Scheme 4).7,8
Thus, further optimization work was focused on this series
of symmetrical diamines. While the simpler diamine 14 was
easier to prepare, it is an oil at room temperature. Diamine
13 is highly crystalline, provides higher enantioselectivity,
and was therefore studied first. Reducing the amount of
dilithium-13 to 2.2 equiv and maintaining the reaction
temperature at 0 °C provided crude alcohol 3b with 97%
ee. Crystallization of the crude product from heptane/MTBE
gave 3b of 99.8% ee in 64% yield at 250-g scale (Scheme
(5) Commercially available from Sigma-Aldrich Chemical Company, Milwau-
kee, WI.
(6) (a) Reference 4i. (b) Modin, S. A.; Andersson, P. G. J. Org. Chem. 2000,
65, 6736. (c) Andersson, P. G.; Sodergen, M. K.; Bertilsson, S. K.;
Andersson, P. G. J. Org. Chem. 2002, 67, 1567.
(7) Bambridge, K.; Begley, M. J.; Simpkins, N. S. Tetrahedron Lett. 1994, 35,
3391. For an example of asymmetric deprotonation of a sulfonium salt using
the lithium amide of 13, see: McComas, C. C.; Van Vranken, D. L.
Tetrahedron Lett. 2003, 44, 8203.
(8) (a) Mimoun, H.; de saint Laumer, J. Y.; Giannini, L.; Scopelliti, R.; Floriani,
C. J. Am. Chem. Soc. 1999, 121, 6158. (b) Horner, L.; Dickerhof, K. Liebigs
Ann. Chem. 1984, 1240.
(9) The estimated cost to prepare diamine 13 on large scale was >$365/kg.
(10) Several systems for the asymmetric rearrangement of meso-epoxides using
a catalytic amount of a chiral lithium amide and an achiral stoichiometric
base (e.g., LDA with DBU or lithiated imidazoles) have been reported. For
leading references, see: (a) Reference 1c. (b) Oxenford, S. J.; Wright, J.
M.; O’Brien, P.; Panday, N.; Shipton, M. R. Tetrahedron Lett. 2005, 46,
8315 and references therein.
(11) Prepared by reaction of 1,3-dimethoxybenzene with n-BuLi in diethyl ether.
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
547

Table 1. Screening results for conversion of epoxide 4b to alcohol 3b
a Reaction performed at 0 °C with 2 equiv of Li-12.
Scheme 4. Preparation of diamines 13 and 14
due to a substantially faster background reaction rate of
epoxide 4b with LDA, relative to the rate of reaction with
dilithium-13.
Another outcome of these studies was the finding that
crude alcohol 3b of <93% ee could not be reliably upgraded
by crystallization using the standard heptane/MTBE crystal-
lization process.12 With this target in mind, attention was
focused on optimizing the selectivity of the simpler diamine,
14. Using diamine 14 and decreasing the reaction temperature
to -55 °C provided alcohol 3b with 97% ee, the same
selectivity achieved with diamine 13 (entry 9). However, the
low solubility of dilithium-14 at this temperature resulted in
very thick, poorly agitated mixtures at desirable reaction
volumes.13
As observed with diamine 13, maximum selectivity
required two equivalents of the lithium diamide of 14
(compare entries 6 and 7).14 Entry 12 of Table 2 is also
illustrative. In this experiment, 2.2 equiv of 14 were reacted
with 3.3 (rather than the standard 4.4) equiv of n-BuLi to
presumably produce 1.1 equiv of dilithium-14 and 1.1 equiv
(12) Racemic and enantiomerically pure alcohol 3b have different crystal forms,
based on X-ray powder diffraction patterns and melting points (122-124.
°C and 112-112. °C, respectively). In addition, the racemate is significantly
less soluble in the 65:35 heptane/MTBE crystallization solvent: 3 mg/mL
vs 15 mg/mL for the single enantiomer (at 22 °C).
(13) On laboratory scale, using a mechanical paddle stirrer, the maximum final
concentration of 4b that enabled agitation of the lithium diamide was 0.25 M.
548
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development

Table 2. Comparison studies of rearrangement reactions using diamines 13 and 14
a Measured on sample of 3b obtained after chromatography. b Yield after chromatography on silica gel. c 1.1 equiv diamine 13 was added to 3.3 equiv 2-lithio-
1,3-dimethoxybenzene d Epoxide 4b treated with 1.5 equiv NaH in THF at 22 °C for 1 h. e 2.2 equiv of diamine 14 treated with 3.3 equiv n-BuLi rather than the
standard 4.4 equiv n-BuLi.
Scheme 5. Epoxide rearrangement process using diamine 13
of monolithium-14. Epoxide 4b was added, and the reaction
mixture was stirred 7 h at 0 °C. Alcohol 3b was obtained
with significantly reduced enantiomeric purity (62% ee),
compared to the analogous reaction utilizing 2.2 equiv of
dilithium-14 (86% ee, entry 6).
4b with 1.1 equiv of dilithium-14 was also studied. Unfor-
tunately, this protocol gave 3b in 61% yield and only 32%
ee (entry 11). Assuming that the poor enantioselectivity might
be related to the presence of sodium counterions, LiH was
also evaluated. However, deprotonation of the carbamate
nitrogen of 4b with LiH in THF was not successful.15
One additional observation from these data is worth
noting. Total accountable mass balance (% product + %
recovered starting material) appeared to increase with
The use of inexpensive NaH for deprotonation of the
carbamate nitrogen followed by reaction of the resulting Na-
(14) Aliquots of the reaction of 4b with 1.1 equiv dilthium-14 (entry 7) were
quenched and assayed by NMR for conversion. After the 2 h, the ratio of
starting material/product was 40:60 and remained constant thereafter. This
suggests that the reaction may have occurred substantially during the addition
of the epoxide solution, when the epoxide would be exposed to excess
dilithium-14. After addition of one-half of the epoxide solution, only
monolithium-14 would remain.
(15) For example, reaction of 4b with 1.5 equiv of LiH in THF for 76 h at
22 °C, followed by quenching with CD3CO2D gave 4b (89% recovery) with
no measurable incorporation of deuterium, based on integration of the NH
signal in the NMR spectrum.
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
549

Scheme 6. Proposed decomposition pathway for alcohol 3
participate in bidentate coordination with the carbamate and
epoxide oxygens. Epoxide 4b was added to the resulting
solution at 0 °C, and the reaction mixture was stirred for
6 h. Alcohol 3b was isolated in 21% yield (along with 58%
recovered epoxide 4b) and was essentially racemic (4% ee).
Further studies will certainly be required to establish the
kinetically relevant amide species in the rearrangement
reaction. The simple rationale proposed does not account for
the potential impact of reactive lithium amide dimers or
aggregates. In addition to solvent effects, the lithium amide
environment may be substantially impacted by the buildup
of lithium alkoxide and the monolithium amides as the
reaction proceeds. However, this model led to the study of
additional diamines to optimize the tether length between
the chiral amine fragments. Longer carbon chain tethers in
the diamine were attractive as they might provide for higher
selectivity and possibly more soluble lithium diamides, thus
solving a critical processing problem.
Results of comparative rearrangement reactions performed
with diamines having modified carbon chain tethers and
substituents (diamines 17-21) are shown in Table 3.20,21 In
summary, diamine 17 having a three carbon linker, provided
3b with the highest enantiomeric excess. The findings that
the enantioselectivity could be optimized by changing the
size of the carbon chain linker and that the replacement of
one secondary amine of the diamine with a tertiary amine
(diamine 19) gave essentially racemic 3b are consistent with
the bidentate chelation model shown in Scheme 7.
decreasing reaction temperature (compare entries 6 and 8-10
in Table 2.) No other organic solvent-soluble byproducts
were isolated via extraction of the aqueous layer (even after
adjusting the pH to 13), suggesting a decomposition pathway
of the product and/or starting material to water-soluble
byproducts. Indeed, when alcohol 3b was treated with 2.2
equiv of dilithium-14 at 23 °C for 18 h, 3b was recovered
in only 45% yield and was the only identifiable product-
like fragment observed. The mass loss in this reaction may
be attributed to the proposed pathway shown in Scheme 6,
resulting in the presumably highly water-soluble amino acid
16, derived from the cyclic oxazolidinone 15.16
In total, the studies with diamines 13 and 14 led to the
hypothesis that chelation of the dilithium amide to both the
carbamate oxygen and the epoxide may be necessary for
optimum selectivity.17 This resulted in a rationale for the
stereochemical outcome of the reaction as shown in Scheme
7.
Selectivity is achieved via coordination of the dilithium-
14 with the epoxide oxygen and the carbamate oxygen,
followed by the known removal of a syn-â-hydrogen.18 In
this model, the minor enantiomer is derived from chelated
structure B, in which a methyl group of the chiral diamine
occupies a space above the five-membered ring. The major
enantiomer is derived from chelated structure A in which a
hydrogen can occupy this space above the ring. The reaction
of epoxide 4b with the magnesium amide of 14 may offer
support for this chelate model. Diamine 14 was treated with
Bu2Mg in refluxing THF19 presumably to provide the cyclic
Mg-14 amide (Scheme 8), which would be less likely to
With project timing dictating pilot-plant processing, we
focused on developing the process using the simplest
(18) Examples of epoxide rearrangements via removal of syn-â, R-, and anti-â
hydrogen have been demonstrated via deuterium labeling and other studies.
See for example (a) Thummel, R. P.; Rickborn, B. J. Am. Chem. Soc. 1970,
92, 2064 (syn-â elimination pathway). (b) Morgan, K. M.; Gajewski, J. J.
J. Org. Chem. 1996, 61, 820. (c) Hodgson, D. M.; Gibbs, A. R. Tetrahedron
Lett. 1997, 51, 8907. (d) Hodgson, D. M.; Gibbs, A. R. J. Chem. Soc.,
Perkin Trans. 1 1999, 3679. (e) Ramirez, A.; Collum, D. B. J. Am. Chem.
Soc. 1999, 121, 11114. (f) Morgan, K. M.; Gronert, S. J. Org. Chem. 2000,
65, 1461 (anti-â elimination pathway). Both solvent and conformational
effects have been reported to impact the site of deprotonation. 3-Substituted
cyclopentene oxides have been reported to show a preference for the syn-
â-deprotonation pathway in the presence of chiral lithium amides. Based
on the retention of deuterium labels in the R-positions of epoxide 4b (below),
there is indication that the often observed syn-â elimination mechanism
may be operating in this substrate.
(16) Amino acid 16 has been prepared, as the trifluoroacetate salt, via reaction
of 3b with neat trifluoroacetic acid. Further studies to confirm the presence
of 16 in the aqueous layers of quenched reaction mixtures are in progress.
No amide products, resulting from the reaction of chiral lithium amides
with the ester of 4b, have been observed.
(17) The reaction of epoxide diastereomer 4c with dilithium-13 provided alcohol
3c with only 61% ee.
In this substrate simultaneous chelation with the epoxide and carbamate
oxygens is not possible. A carbamate anion may be not be essential for
bidentate coordination, however. Reaction of the N-Me epoxide i with
dilithium-13 yielded alcohol ii with 85% ee (see Supporting Information).
(19) For an asymmetric deprotonation of meso ketones with chiral magnesium
amides, see for example: Anderson, J. D.; Garcia Garcia, P.; Hayes, D.;
Henderson, K. W.; Kerr, W. J.; Moir, J. H.; Fondekar, K. P. Tetrahedron
Lett. 2001, 42, 7111.
(20) Diamine 17 was originally prepared from dibromopropane as described in
reference 8b. For a recent use of catalytic dilithium-17 in the rearrangement
of cyclcohexene oxide see: Equuey, O.; Alexakis, A. Tetrahedron Asym-
metry 2004, 15, 1069.
(21) Diamine 18: Kobayashi, Y.; Hayashi, N.; Kishi, Y. Org. Lett. 2002, 4,
411. The preparations of diamines 19-21 are described in the Supporting
Information.
550
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development

Scheme 7. Rationale for enantioselectivity of the rearrangement reaction
Scheme 8. Reaction of diamine 14 with Bu2Mg
isolated as shown in Scheme 9. Commercially available (R)-
R-methylbenzylamine of 98.2% ee provided the dihydro-
chloride salt of 17 in 85% yield (35-kg scale) with an isomer
ratio R,R/meso of 98.1:1.9. The S,S-enantiomer was not
detected by capillary electrophoresis assay.
(although noncrystalline) diamine, 17. As shown in Table
4, reducing the reaction temperature below -25 °C did
provide 3b with >95% ee. Importantly, dilithium-17 did not
precipitate at the lower reaction temperatures. For pilot-plant
processing, -45 °C was selected as the reaction temperature.
One added challenge to pilot-plant processing was the use
of diamine 17 as an oil. The free base was conveniently
prepared by adding 1,3-dibromopropane to 4 equiv of (R)-
R-methylbenzylamine at 100 °C. Upon reaction completion,
the mixture was partitioned between aqueous NaOH and
MTBE. MTBE was removed, and excess (R)-R-methylben-
zylamine was recovered via vacuum distillation. The product
could be easily separated from higher-boiling byproducts22
by further vacuum distillation. A more practical purification
process was designed in which crude 17 was treated with
aq HCl in ethanol, and the dihydrochloride salt of 17 was
It was most desirable to integrate the free basing of 17-
dihydrochloride and the rearrangement reaction into a
telescoped process. Thus, we evaluated the impact of
potential water-immiscible organic solvents for free basing
that would be carried into the rearrangement reaction. Under
model laboratory reaction conditions, the final reaction
mixture solvent contained 6:1 (v:v) THF/hexanes (the latter
derived from n-BuLi solution). As shown in Table 5,
performing the reaction in 6:1 MTBE/hexanes reduced the
% ee of the crude product 3b by approximately 10%. In
toluene the enantioselectivity was reversed, providing the
undesired enantiomer of 3b with 35% ee. However, adding
2 or 10 equiv of THF/equivalent of diamine to either MTBE
or toluene nearly restored the selectivity to that observed
with THF as solvent. Thus, MTBE solutions of the free base
17 were used for pilot-plant processing.
Two additional studies were performed prior to scale-up
of the rearrangement reaction. One set of experiments probed
the stability of dilithium-17. This study was prompted by
the observation of acetophenone in the crude organic extracts
from the epoxide rearrangements using diamines 13, 14, or
17. Acetophenone presumably is formed via hydrolysis of
imine 22 resulting from â-hydride elimination of the lithium
amide (Scheme 10). This reaction was of potential concern
for larger-scale processing in which the likely longer addition
times would require longer solution lifetimes for the diamide.
Laboratory studies were conducted in which the dilithium-
17 solution was aged 30 min, 2 and 5 h prior to addition of
the epoxide solution. These studies showed no change in
yield or enantiomeric purity of isolated 3b. The amount of
acetophenone observed in the crude product was not
substantially different (<10%, relative to 3b) in these three
reactions.
Table 3. Survey results for second generation diamines
A second screening study examined the impact of
inorganic cations or anions that might be carried into the
rearrangement reaction from the free basing of diamine 17-
(22) Crude reaction mixtures typically contained a 96:4 ratio of 17 and tertiary
amine iii. The latter was removed by isolation of 17-dihydrochloride salt.
a Isolated yield after column chromatography.
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
551

Table 4. Temperature study using diamine 17
total %
isolated yield
of 3b
temp (°C)/
reaction time (h)
% ee of
crude 3b
% crystallizeda
yield of 3b (% ee)
% 3b isolated from
% recovered
mother liquorb (% ee)
epoxide 4bb
-78/7
-45/6
-25/3
-5/1
99.2
97.5
95.7
93.2
-
-
80b
85
84
78
9
5
-
75 (99.8)c
74 (99.7)c
57 (97.6 )d
10 (89.7)
10 (76.2)
21 (86.2)
a Crude 3b was crystallized from 13 mL of 65:35 hexanes/MTBE/g. b Isolated by silica gel chromatography. c Unseeded crystallizations. d Crystallization seeded
with 99% ee 3b.
Scheme 9. Preparation of diamine 17-dihydrochloride salt
Scheme 10. Possible decomposition pathway of
dilithium-17
equivalents of LiCl resulted in an 8% ee decrease, while the
more soluble lithium tert-butoxide23 caused a 17% reduction
in ee, relative to the control reaction. Addition of 5 equiv of
sodium tert-butoxide gave essentially racemic product (12%
ee). As seen previously with diamine 14, soluble sodium
cations have a substantial deleterious impact on the enanti-
oselectivity of the reaction.
Pilot-Scale Preparation of Epoxide 4b. The synthesis
of the ethyl ester epoxide 4a from diethyl malonate was
previously reported by Hodgson.3 Syntheses of epoxide 4b
were demonstrated in the laboratory using this strategy and
a variety of strategies based on the alkylation of tert-butyl
glycine imines24 with 1,4-disubstituted-cis-2-butene deriva-
tives to form the cyclopentene ring. Due to ready commercial
dihydrochloride (sodium and/or chloride) or from the pres-
ence of adventitious water during processing (e.g., LiOH).
As shown in Table 6, the presence five equivalents of LiOH
did not impact the yield or enantioselectivity of the reaction
(performed -5 °C in this screening study). Addition of five
Table 5. Impact of solvent on selectivity of the
rearrangement reaction
Table 6. Impact of additives on the epoxide rearrangement
reaction
equiv THF added/ % ee 3b % yield % recovered
solvent
equiv diamine
(HPLC)
3ba
4ba
additive
% ee 3b (HPLC) % yield 3ba % recovered 4ba
THF
-
-
2
10
-
2
94
84
88
88
-35
90
64
73
73
77
39
53
65
-
5
1
4
20
3
MTBE
MTBE
MTBE
PhCH3
PhCH3
PhCH3
none
LiOH
LiCl
LiOt-Bu
NaOt-Bu
92
91
84
75
12
75
77
76
69
39
1
2
1
9
21
10
92
-
a Isolated yields after purification by column chromatography.
a Isolated yields after purification by column chromatography.
552
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development

Scheme 11. Pilot-plant preparation of epoxide 4b
access, tert-butyl methylmalonate25 was selected as the
starting material in preference to tert-butyl glycine. cis-1,4-
Dichloro-2-butene was selected as the alkylating agent.
The pilot-plant process used to prepare epoxide 4b
(Scheme 11) began with the alkylation of tert-butyl methyl-
malonate with cis-1,4-dichloro-2-butene26 using LiH in THF/
DMPU.27 The resulting solution of cyclopentene diester 23
was hydrolyzed directly with aqueous NaOH to give the
mono ester acid 24 in 79% overall yield (5-kg scale).
Acid 24 was converted to the Boc-protected amine 5b
via a four-step sequence of acid chloride (25) formation with
thionyl chloride, phase transfer-catalyzed preparation of the
acylazide 26, Curtius rearrangement of 26, and trapping of
the isocyanate 27 with potassium tert-butoxide/tert-BuOH.
Crystallization from MTBE/heptane provided cyclopentene
5b in 71% overall yield from 24 on 6-kg scale.
Under a variety of oxidizing conditions, epoxidation of
5b preferentially gave epoxide diastereomer 4b in which the
epoxide oxygen occupies the same face of the ring as the
NHBoc group.28 Magnesium monoperoxyphthalate, available
in commercial quantities, was used in a biphasic water/ethyl
acetate mixture with n-Bu4NHSO4 phase transfer catalyst to
afford a 93:7 mixture of epoxide diastereomers (4b:4c). The
crude product was crystallized from EtOAc/heptane to give
a 71% yield of 4b, containing 3% of the minor diastereomer
(on 10-kg scale).
Scale-Up of the Epoxide Rearrangement. The rear-
rangement of epoxide 4b was performed in pilot-plant
equipment using diamine 17 via the process shown in
Scheme 12. The free base diamine 17 was prepared from
the dihydrochloride salt in a mixture of aqueous NaOH and
MTBE. The MTBE layer was azeotropically dried and
diluted with THF. The solution was cooled to -45 °C, and
2.5 M n-BuLi in hexanes was added to form the dilithium
amide. A solution of epoxide 4b was then added, and the
reaction was stirred at -45 °C. Upon reaction completion
(approximately 16 h),29 the reaction was quenched with
aqueous sulfuric acid.30 After layer separation, a brine wash,
and concentration of the organic layer, alcohol 3b was
(23) For a study of achiral rearrangements of epoxides with lithium tert-butoxide
and lithium amide bases see: Saravanan, P.; DattaGupta, A.; Bhuniya, D.;
Singh, V. K. Tetrahedron 1997, 53, 1855.
(24) Park, K. H.; Olmstead, M. M.; Kurth, M. J. J. Org. Chem. 1998, 63,
113.
(25) tert-Butyl methylmalonate was purchased in bulk quantities from Tateyama
Kasei Co., Ltd, Toyama, Japan. A synthesis of 5b starting with tert-butyl
acetoacetate has been reported: Larionov, O. L.; Kozhushkov, S. I.; de
Meijere, A. Synthesis 2005, 158.
(26) The cis-1,4-dichloro-2-butene used contained 3.9% trans-1,4-dichloro-2-
butene and 1.0% 3,4-dicihloro-1-butene. In the preparation of 24, trans-
1,4,dichloro-2-butene reacts with tert-butyl methylmalonate to give vinyl
cyclopropane diastereomers iv and subsequently acids v, which are removed
in the crystallization of 24.
(28) Characterization data for major/minor epoxide diastereomers 4b and 4c were
consistent with that reported for the analogous ethyl ester epoxides (ref 3).
The observed epoxidation facial preference has been reported for several
other 3-amide cyclopentene substrates. See, for example, refs 4a,b and
O’Brien, P. O.; Childs, A. C.; Ensor, G. J.; Hill, C. L.; Kirby, J. P.; Dearden,
M. J.; Oxenford, S. J.; Rosser, C. M. Org. Lett. 2003, 26, 4955.
(27) (a) Depres, J. P.; Greene, A. E. Org. Synth. 1997, 75, 195. (b) Depres, J.
P.; Greene, A. E. J. Org. Chem. 1984, 49, 928.
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
553

Scheme 12. Pilot-plant process for the preparation of alcohol 3b
Scheme 13. Comparison of rearrangement reaction of epoxide diastereomers
crystallized form heptane/MTBE. In two runs at 3-kg scale,
alcohol 3b was obtained in 79 and 74% (assay-corrected)
yields, each lot with 99.6% ee. Diamine 17-dihydrochloride
was subsequently recovered from the combined aqueous acid
extracts in 80% overall yield via free-basing and re-forming
the HCl salt (see Experimental Section). Recovered diamine
17-dihydrochloride showed the same stereoisomeric purity
as the starting lot and was successfully reused in the
rearrangement reaction.31
Both of the above lots of 3b contained approximately
0.5% of alcohol 3c, derived from the rearrangement of the
minor epoxide diastereomer 4c (Scheme 13). By virtue of
the subsequent oxidation step, this alcohol is also converted
to the desired ketone 2b. However, the presence of alcohol
3c could potentially have a minor impact on the enantiomeric
purity of ketone 2b if the rearrangement of epoxide 4c
proceeds with opposite or less enantioselectivity relative to
that observed for epoxide 4b. In a laboratory reaction
comparison, epoxides 4b and 4c were reacted with lithium
dilithium-17 under standard pilot-plant reaction conditions,
and the reactions were quenched after 1.25 h. Interestingly,
the rearrangement reaction of the minor epoxide 4c is much
slower (3% yield vs 72% yield) and less selective (48% ee
vs 99.8% ee product) than the reaction of epoxide 4b.
Oxidation of alcohols 3b and 3c provided samples of ketone
2b with the same sign of optical rotation, which indicates
that the deprotonation occurs with the same relative enan-
tiomeric preference for both epoxides.32
Oxidation Reaction. The final conversion to the key
intermediate 2b was accomplished in the pilot plant in a
biphasic mixture of MTBE and aqueous NaOCl containing
8 mol % of the TEMPO radical.33 This process provided
the key API intermediate, ketone 2b, in 81% assay corrected
yield on 3-kg scale (Scheme 14).
Conclusion
The desymmetrization of epoxide 4b was accomplished
on pilot-plant scale to provide high yields of allylic alcohol
(32) See data shown below.
(29) The reaction was typically >80% complete 1 h after addition of the epoxide
solution. For pilot-plant processing, end of reaction was defined as less
than 2% epoxide 4b remaining. Levels of 4b greater than 2% were not
substantially reduced in the crystallization of 3b or the subsequent
intermediate 2b. As the reaction proceeds, it is possible that the effective
concentration of epoxide is further reduced by nonproductive coordination
with monolithium-17, which is present at substantially higher concentration
than dilithium-17.
(30) Sulfuric acid was used in preference to aqueous HCl, as the former salt
does not precipitate from the reaction mixture, resulting in simpler layer
separations.
(31) Recovered 17-dihydrochloride used in a laboratory test provided alcohol
3b in 72% yield and 99.9% ee.
(33) Anneli, P. L.; Montanari, F.; Quici, S. Org. Synth. 1990, 69, 212 and
references therein.
554
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development

Scheme 14. Oxidation of alcohol 3b
For small-scale and screening epoxide rearrangement reac-
tions, compounds 3a,b were purified via column chroma-
tography using silica gel (EMD silica gel 60), eluting with
1:1 hexanes/MTBE. (Rf for compound 3b ) 0.15; compound
4b ) 0.33). Magnesium monoperoxy phthalate was obtained
from DeGussa, Peroxid-Chemie GmbH & Co., Germany.
tert-Butyl methylmalonate was obtained from Tateyama
Kasei Co., Ltd, Toyama, Japan. cis-1,4-Dichloro-2-butene
was obtained form Alfa Aesar, Ward Hill, MA. (R)-R-
Methylbenzylamine was obtained from Zeeland Chemicals,
Zeeland, MI.
3b, with >99% ee. This provided a key chiral intermediate,
ketone 2b, for the synthesis of LY459477. The chiral diamine
17 used is easily prepared from commercially available (R)-
R-methyl benzylamine and was recovered in 80% yield from
the pilot-plant reaction mixtures. This reagent was designed
by balancing ease of preparation, enantioselectivity, and
operational requirements in the rearrangement reaction. Study
of a variety of diamine analogues provided insight into the
possible mechanism of the reaction but more importantly
led to the design of a scaleable reaction. Results of studies
to make this transformation more practical (e.g., reducing
the amount of diamine required) will be reported in due
course.
3-Cyclopenetene-1,1-dicarboxylic Acid, Mono(1,1-di-
methylethyl) Ester (24). To a 30-gal glass-lined reactor
under N2, were added N,N-dimethylpropyleneurea (DMPU,
16 kg) followed by a slurry of LiH (0.605 kg, 75.5 mol) in
10 L of THF. The slurry was heated to 40-45 °C and (neat)
tert-butyl methylmalonate (5.0 kg, 28.7 mol) was added over
30 min, maintaining the temperature at 40-45 °C. (Hydrogen
was evolved.) A solution of cis-1,4-dichloro-2-butene (4.10 kg,
31.8 mol) in 10 L of THF was then added over 3.8 h,
maintaining the temperature at 40 °C. Hydrogen was evolved
throughout the addition. The reaction was stirred at 40 °C
and monitored by GC assay. After 1.5 h, the reaction mixture
was cooled to room temperature and added over 25 min to
40 L of water in a 50-gal glass-lined reactor. The temperature
was maintained below 30 °C; hydrogen was evolved.
Aqueous NaOH (5 M, 10 L, 50 mol, 1.75 equiv) was added
over 5 min, and the resulting homogeneous mixture was
stirred at 25 °C for 2 h, at which time HPLC assay showed
no 23 remained. MTBE (25 L) was added to the reaction
mixture, and the layers were separated. The aqueous layer
was transferred to a 50-gal Hastelloy C reactor containing
25 L of MTBE. The pH of the aqueous layer was adjusted
to 2.8 by the addition of 46 L of 2 M NaHSO4 over 16 min.
The layers were separated, and the aqueous layer was
extracted with 25 L of MTBE. The combined organic layers
were extracted with 2 × 50 L of 5% LiCl solution (to remove
DMPU), and filtered through a 5 micron cartridge filter into
a 30-gal glass-lined tank. The solution was concentrated
under vacuum (160 mmHg, solution temperature of 20-
25 °C) to approximately 20 L. Heptane (70 L) was added,
and the solution was concentrated to 30-35 L under vacuum;
crystals formed. The slurry was stirred 1.5 h at 20-25 °C
and filtered. The solid was washed with 20 L of heptane
and vacuum-dried 24 h at 45 °C to obtain 4.98 kg (81%
yield) of 625222 as a white solid, mp 119-122 °C. HPLC
assay showed 0.7% of the cyclopenetene dicarboxylic acid
and 1.45% residual DMPU. Corrected yield ) 79%. 1H NMR
(300 MHz, CDCl3): δ 5.61 (s, 2H), 3.00 (s, 4H), 1.46 (s,
9H). 13C NMR (125 MHz, CDCl3): δ 178.8, 178.8, 127.8,
82.1, 59.6, 40.9, 27.7. IR (CHCl3): 3800-3000 (br, COOH),
1741, 1705, 1283, and 1149 cm-1.
Experimental Section
General. Melting points were obtained using a Thomas-
Hoover capillary melting point apparatus and are uncorrected.
NMR data were obtained using a Varian instrument at 300
or 500 MHz for 1H and 125 MHz for 13C. Unless otherwise
stated, reagents were commercially available and used
without further purification. Reaction completion for com-
pounds 23, 25, 26, and 27 was monitored using an Agilent
5890 GC equipped with a 15 m × 0.25 mm DB1701 column
and a flame ionization detector using the following condi-
tions: flow, 1.5 mL/min; temperature gradient, 60 °C for
2 min, 18 °C/min to 280 °C, 280 °C for 5 min. Reaction
completion for compounds 24, 3b, 4b, 5b, and 2b was
monitored by HPLC using an Agilent 1100 series instrument
equipped with a UV detector and a Zorbax SB-Phenyl 25 cm
× 4.5 mm column using the following conditions: eluents,
acetonitrile and 0.1% aqueous H3PO4; eluent gradient, 40:60
acetonitrile/0.1% aq H3PO4 to 80:20 acetonitrile/0.1% H3-
PO4 from 1 to 21 min; flow ) 1 mL/min; detector ) 200 nm;
column temperature ) 25 °C. Chiral assays for compounds
4a-c were performed using an Agilent 1100 series HPLC
equipped with a Chiralpak 5 µm, 4.6 mm × 250 mm column
and a UV detector using the following conditions: column
temperature, 30 °C; eluent, isocratic 90:10 hexane/isopropyl
alcohol; flow rate, 1 mL/min; detector, 210 nm. Reaction
completion for compound 17 was monitored using an Agilent
5890 GC equipped with a 30 m × 0.25 mm DB-1 column
and a flame ionization detector using the following condi-
tions: flow ) 1 mL/min, temperature gradient, 60 °C for
2 min, 18 °C/min to 300 °C. Chiral assays for diamine 17
were obtained with a Hewlett-Packard 3D capillary electro-
phoresis instrument equipped with an uncoated capillary and
a UV detector using the following conditions: buffer, 25 mM
NaH2PO4 (pH 7.0) + 15 mM SBE-B-CD; preconditioning,
0.1 N H3PO4 for 3 min; voltage, +25 kV; column temper-
ature, 15 °C; detection: 195 nM; pressure injection (50 mbar,
4 s); capillary: 50 µm i.d. × 64.5 cm (56 cm to detector).
1-[[(1,1)-Dimethylethoxy]carbonyl]amino]-3-cyclopen-
tene-1-carboxylic Acid, 1,1-Dimethylethyl Ester (5b).
Preparation of acid chloride 25. Compound 24 (6.0 kg, 28.27
mol) was dissolved in 100 L of 70:30 toluene/MTBE in a
200-gal glass-lined tank. Thionyl chloride (4.0 kg, 33.6 mol,
1.2 equiv) was added to the stirred reaction mixture
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
555

(maintaining the temperature at 23 °C), followed by a toluene
rinse (1-2 L) of the SOCl2 container. The reaction solution
was cooled to 0-10 °C and a solution of triethylamine
(3.83 kg, 37.85 mol, 1.34 equiv) in 15 L of toluene was added
over 1.75 h. The reaction mixture was warmed to 20-23 °C
over 10 min. GC assay of the reaction showed that 0.85%
of the starting material remained. The reaction mixture was
transferred over 25 min to a 200-gal glass-lined tank
containing deionized water (75 L), while maintaining the
temperature at 20-25 °C. The tank containing the reaction
mixture was rinsed with toluene (20 L), and the rinse was
added to the aqueous reaction mixture. The solution was
stirred for 10 min, and the layers were separated. To the
organic layer was added 1 M NaHCO3 (60 L), and the
mixture was stirred for 10 min. The layers were allowed to
settle over 10 min and then were separated. The acid chloride
25 contained in the organic (toluene/MTBE) layer was taken
directly into the next reaction. Acid chloride 25 could be
Hz, 2H), 1.50 (s, 9H). IR (film): 2258 (-NCO), 1732, 1157
cm-1.
Preparation of 5b. Potassium tert-butoxide (44 kg, 1 M
solution in THF, 48.8 mol, 1.7 equiv) was added to a 200-
gal glass-lined tank. To the KOt-Bu solution was added tert-
butanol (5.2 L), and the transfer lines were washed with
toluene (2 L). The reaction solution was cooled to 0-5 °C,
and the toluene solution containing isocyanate 27 was added
over 60 min, maintaining the temperature at 0-10 °C. The
reaction was warmed to 23 °C, stirred for 40 min, and
assayed by GC for the disappearance of isocyanate. The
reaction mixture was then added to a mixture of deionized
H2O (140 L) and MTBE (140 L) over 15 min, maintaining
the temperature between 20 and 30 °C. The solution was
stirred for 20 min, and the layers were separated. The
aqueous layer was washed with MTBE (70 L), and the
combined organic layers were extracted with a 20% brine
solution (60 L), stirred for 10 min, and separated. The organic
layer was transferred to 100-gal glass-lined tank and
concentrated via distillation (45-50 °C, 550 mmHg) to
approximately 25 L. A solvent exchange to heptane was
accomplished by repeating the following process four
times: heptane (50 L) was then added to the distillation tank,
and the resulting solution was distilled at 40-50 °C, 60-
90 mmHg, until the total volume was 20-30 L. The final
distillation concentrated the solution to approximately 10 L.
Heptane (20 L) was added and the solution cooled to 20 °C
over 3 h. The solution was further cooled to -10 °C over
60 min and stirred for 3.5 h to precipitate the product. The
solid was filtered, washed with 20 L of cold heptane, and
vacuum-dried at 50 °C for 12 h to yield 5.92 kg (73% yield)
of compound 5b as a white solid, mp 87-89 °C. HPLC assay
showed 97.8 area % compound 5b. Corrected yield ) 71.4%.
1H NMR (500 MHz, CDCl3): δ 5.63 (s, 2H), 5.1 (bs, 1H,
NH), 2.99 (d, J ) 17.2 Hz, 2H), 2.57 (d, J ) 16.0 Hz, 2H),
1.46 (s, 9H), 1.44 (s, 9H). 13C NMR (CDCl3): δ 173.3, 154.9,
127.7, 81.1, 64.5, 44.8, 28.3, 27.8. IR (KBr): 3451, 2981,
2932, 1712, 1489, 1369, 1154 cm-1. MS (ES) m/e (% relative
intensity): 284.2 (M+ + 1, 56), 228.2 (73), 172.1 (97), 128.0
(100).
(1R,3â,5R)-3-[[(1,1-Dimethylethoxy)carbonyl]amino]-
6-oxabicyclo[3.1.0]hexane-3-carboxylic Acid, 1,1-Dimeth-
ylethyl Ester (4b). Compound 5b (11.02 kg, 38.89 mol) was
dissolved in EtOAc (55 L) in a 100-gal glass-lined carbon
steel tank. To this solution was added 60 L of 0.1 N H2SO4.
The mixture was stirred for 5 min, and the layers were
allowed to separate over 25 min34 The lower aqueous layer
was removed and neutralized. Tetrabutylammonium hydro-
gen sulfate (2.20 kg, 6.48 mol, 0.17 equiv) was then added
to the EtOAc solution, followed by a solution of magnesium
monoperoxyphthalate hydrate (MMPA, 28.80 kg, 4.76%
active oxygen ) 1.1 equiv) in deionized H2O (32 L). The
tank containing the MMPA solution was rinsed with deion-
ized H2O (20 L), and the rinse was added to the reaction
mixture. The reaction was stirred at 22 °C for 22 h. A sample
was analyzed, and the starting material was absent by HPLC
1
isolated as an oil by concentration of the organic layer. H
NMR (300 MHz, CDCl3): δ 5.61 (s, 2H), 3.04 (quartet, JAB
) 15.1 Hz, 4H), 1.49 (s, 9H). IR (film): 1796, 1743, 1274,
1233, 1158 cm-1.
Preparation of Azide 26. A solution of tetrabutyl
ammonium hydrogen sulfate (0.105 kg, 0.31 mol, 0.01 equiv)
in deionized water (0.5 L) was added to a solution of sodium
azide (2.25 kg, 34.61 mol, 1.22 equiv) in deionized H2O
(75 L) in a 100-gal glass-lined tank. The solution containing
acid chloride 25 in MTBE/toluene was added to the azide
solution over 50 min. The tank containing the acid chloride
was rinsed with 10 L of toluene, and the rinse was added to
the reaction solution. The reaction mixture was stirred for
2.75 h at 23 °C. GC analysis of the reaction solution showed
that the acid chloride was consumed. The layers were
separated, and the organic layer was washed with 1 M
NaHCO3 (60 L) and deionized H2O (60 L, then 30 L). The
organic layer was filtered through a 14 kg sodium sulfate
filter and collected in a 200-gal glass-lined tank. This solu-
tion of acyl azide 26 in MTBE/toluene was used in the
next step. Azide 26 could be isolated as an oil by evaporation
1
of the solvent under a stream of nitrogen. H NMR (300
MHz, CDCl3): δ 5.58 (s, 2H), 2.96 (t, J ) 2.3 Hz, 4H),
1.46 (s, 9H). IR (film): 2137 (CON3), 1720, 1246, 1185,
1136 cm-1.
Preparation of Isocyanate 27. Toluene (30 L) was added
to a 100-gal glass-lined tank and was heated to 90-100 °C.
The acyl azide 26 solution was transferred slowly over
45 min to the heated toluene solution, maintaining the
temperature at 90-100 °C. (Evolution of nitrogen gas was
addition-rate controlled under these conditions.) Additionally,
MTBE was distilled from the reaction during the addition.
The reaction was stirred for 15 min at 95 °C. GC assay
showed that 0.8% of the acyl azide remained. The reaction
mixture was cooled to 20-25 °C over 30 min. The
isocyanate 27 solution was carried forward to the next step.
Isocyanate 27 could be isolated as an oil by evaporation of
the solvent under vacuum. 1H NMR (300 MHz, CDCl3): δ
5.67 (s, 2H), 3.01 (d, J ) 15.6 Hz, 2H), 2.61 (d, J ) 15.6
(34) The extraction of the solution of 5b with dilute acid prevented the formation
of small amounts of two, yet unidentified, non-polar impurities during the
reaction.
556
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development

analysis. A solution of sodium sulfite (9.9 kg) in deionized
H2O (100 L) was added to the reaction mixture over 18 min.
The reaction was stirred for 20 min at 22 °C at which point
a KI/starch paper test showed no peroxide present. Ethyl
acetate (60 L) was added to the reaction, and the mixture
was stirred for 5 min. The layers were separated, and the
organic layer was washed with deionized H2O (58 L), 2 N
NaOH (60 L), and deionized H2O (2 × 60 L). The organic
layer was transferred to a clean tank, and the solution was
distilled at 75-95 °C at atmospheric pressure until the
volume reached 20-25 L. The solution was cooled to 75 °C,
and heptane (65 L), heated to 70-80 °C, was added over
13 min, maintaining the reaction solution temperature at 70-
80 °C. The solution was cooled to 65-70 °C, and a small
portion of the reaction solution was drawn into a tube,
allowed to cool and crystallize, and was added back to the
reaction mixture as seed crystals. The reaction was stirred
at 67 °C for 20 min, cooled to 64 °C over 1 h, cooled to
41 °C over 67 min, cooled to -5 °C over 1.5 h, and stirred
for 60 min. The precipitate was filtered, washed with a cold
(-5 °C) mixture of heptane (35 L) and EtOAc (10 L), and
dried under vacuum for 5 h at 50 °C to give 8.70 kg of
epoxide 4b with assay (potency) of 95.9% (HPLC vs a
reference standard) and contained 3% of epoxide 4c. A
recrystallized sample of 4b (free of diastereomer 4c) had a
away from the product. The solution was cooled under N2
to give 5.689 kg (94.2% uncorrected yield) of diamine 17
(94.9 area % by GC; R-methylbenzylamine not detec-
ted).
Preparation of 17-Dihydrochloride Salt. Crude diamine
17 (32.22 kg, 114.2 mol) was added to ethanol (495 L,
denatured with toluene) in a 200-gal, glass-lined stainless
steel tank. The containers were rinsed with ethanol (10 L),
and the rinse was added to the reaction tank. Aqueous HCl
(34 wt %, 29.74 kg, 277.4 mol) was added to the reaction
mixture, while maintaining a temperature of 20-30 °C. The
HCl container was rinsed with ethanol (10 L), and the rinse
was added to the reaction solution. The solution was cooled
and stirred for 1 h at -8 °C. The precipitate was filtered,
washed with cold (-10 °C) ethanol (67 L), and vacuum-
dried for 4 days at 70 °C to give 36.83 kg (85.7% yield) of
diamine 17-dihydrochloride as a white solid, mp 285 °C.
Capillary electrophoresis assay showed this material to
contain 98.1% of the (R,R)-isomer and 1.9% of the meso-
isomer. (tR ) 25.7, 26.1, and 26.8 min for (R,R)-, meso-,
and (S,S)-isomers, respectively.) 1H NMR (500 MHz,
DMSO-d6): δ 9.93 (bs, 2H), 9.47 (bs, 2H), 7.56 (d, J )
7.0 Hz, 4H), 7.37 (m, 6H), 4.27 (m, 2H), 2.84 (m, 2H), 2.48
(m, 2H), 2.1 (m, 2H), 1.56 (d, J ) 6.7 Hz, 6H).
(1S,4R)-1-[[(1,1-Dimethylethoxy)carbonyl]amino]-4-
hydroxy-2-cyclopentene-1-carboxylic acid, 1,1-Dimethyl-
ethyl Ester (3b). Diamine 17-dihydrochloride (13.70 kg,
38.55 mol, 2.4 equiv) was added to deionized H2O (50 L)
in a 100-gal, glass-lined stainless steel tank. The solids
dissolved after 6 min of stirring. A solution of 2.4 N NaOH
(42 L, 102 mol) was added to the reaction solution and stirred
for 10 min followed by the addition of MTBE (75 L). The
reaction mixture was stirred for 10 min, and the layers were
separated. The aqueous layer was extracted with MTBE
(25 L, 10 min stir time), and the layers were separated. The
organic layers were combined and stirred for 10 min with a
20% NaCl solution (40 L). The layers were allowed to settle
for 1 h and then were separated. The organic layer was
refluxed for a total of 4 h, and water was periodically drained
from the distillate take-off valve. The solution was then
concentrated to 30 L by atmospheric distillation of the
MTBE. (Karl Fischer titration assay of the solution indicated
0.14% H2O.) The solution was cooled to 25 °C and diluted
with 50 L of MTBE. The reflux and atmospheric distillation
operations were repeated, and the solution was concentrated
to a volume of 30 L. (KF assay of the solution showed
0.035% water.) The organic solution was transferred to a
50-gal, glass-lined carbon steel tank, the previous tank was
rinsed with THF (10 L), and the rinse was added to the 50-
gal tank. The reaction solution was cooled to -35 to -55 °C
over 30 min. A solution of n-BuLi (2.5 M in hexanes,
19.8 kg, 71.1 mol, 4.4 equiv relative to 4b)35 was added over
70 min, while maintaining the reaction temperature at -35
to -55 °C. The solution was stirred for 2.2 h at -45 °C,
and then a solution of epoxide 4b (4.80 kg, 16.05 mol) in
1
melting point of 131-33 °C. H NMR (500 MHz, CDCl3):
δ 5.0 (bs, 1H), 3.58 (s, 2H), 2.41 (d, J ) 15.3 Hz, 2H), 2.20
(d, J ) 15.2 Hz, 2H), 1.43 (s, 9H), 1.40 (s, 9H). 13C NMR
(CDCl3): δ 171.3, 154.3, 81.2, 79.5, 62.8, 57.0, 38.6, 28.3,
27.7. IR (KBr): 3453, 2982, 2932, 1726, 1708, 1489, 1369,
1293, 1156, 840 cm-1. MS (ES) m/e (% relative intensity):
300.3 (M+ + 1, 65), 244.4 (68), 188.2 (100), 144.1 (99).
N,N′-Bis[(1R)-1-phenethyl-1,3-propanediamine]dihy-
drochloride (17-diHCl). Preparation of Free Base Di-
amine 17. (R)-R-methylbenzylamine (5.36 kg, 44.23 mol,
3.8 equiv) was heated to 98-105 °C in a 22-L flask, and
1,3-dibromopropane (2.25 kg, 11.14 mol) was added over
110 min, while maintaining the temperature at 100-120 °C.
The solution was stirred for 45 min at approximately 100 °C,
at which time GC assay of the reaction mixture showed that
no 1,3-dibromopropane remained. The solution was cooled
to 80 °C, and a 50% NaOH solution (1.4 L) was added over
11 min. The solution was cooled to <50 °C, MTBE (4.45
L) was added, and the mixture was stirred for 15 min.
Deionized water (5 L) was added to the reaction, and the
layers were separated. To the organic layer was added brine
(3 L). The layers were separated, and the organic layer was
dried with Na2SO4 (1 kg). The solid was removed by
filtration, and the filtrate was concentrated under vacuum.
This above reaction was duplicated to give two batches of
crude diamine 17 (5.61 and 5.21 kg). The oils were combined
in a 22-L flask equipped with a mechanical stirrer and a
short-path distillation head. The flasks containing the oils
were rinsed with 0.1 L of MTBE, and the rinses were added
to the oils. Vacuum was slowly applied, and the solution
was heated to 35 °C to distill off MTBE. The remaining oil
was slowly heated (at 4 mmHg), until the vapor temperature
peaked at 75-80 °C, to distill the R-methylbenzylamine
(35) The yield of the free basing was determined to be 93%, based on gravimetric
assay of a sample of the free base 17 solution. This yield was consistent
over multiple pilot-plant runs. The stoichiometry of the n-BuLi and epoxide
4b were calculated based on this yield.
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
557

THF (23 L) was added over 1.75 h, while maintaining the
reaction solution at -35 to -55 °C. The tank containing
the 4b solution was rinsed with THF (2 L), and the rinse
was added to the reaction solution. The reaction solution was
stirred at -45 °C for 15 h at which point HPLC assay
showed 1.1% of epoxide 4b remained. The reaction solution
was then transferred over 1.5 h to a 100-gal, glass-lined
stainless steel tank containing a 4 N H2SO4 solution (45 L)
that was maintained at a temperature below 20 °C. The
mixture was stirred for 20 min, and the layers were separated.
The aqueous layer was extracted with MTBE (15 L), and
the layers were separated. The combined organic layers were
washed with water (20 L) and a 20% NaCl solution (25 L).
The organic layer was concentrated by distillation at 34-41
°C at 350-550 mmHg to a volume of 15 L. The organic
layer was then heated at atmospheric pressure to 45-55 °C,
and heptane (35 L) was added slowly over 1 h, while
maintaining a temperature of 44-55 °C. The solution was
cooled to 0-5 °C over 3 h and stirred at 2 °C for 3 h. The
solid was filtered, and the tank was rinsed with a cold (5 °C)
3:1 heptane/MTBE solution (20 L). The rinse was filtered
over the product cake, and the cake was dried dry under N2
for 4 h. The cake was transferred to a vacuum dryer and
dried for 19 h at 50 °C to give 3.61 kg (76.7% yield) of
578242 as white crystals (97.7% potency). Corrected yield
) 74.9%. HPLC assay showed this lot contained 0.59%
alcohol 3c and 0.42% epoxide 4b. Chiral HPLC assay:
99.6% ee. (tR ) 10.07 and 7.10 min for desired and undesired
enantiomer, respectively.) An analytically pure sample of
phoresis assay showed this material to contain 98.5% of the
(R,R)-isomer and 1.5% of the meso-isomer. The (S,S)-isomer
was not detected.
(1S)-1-[[(1,1-Dimethylethoxy)carbonyl]amino]-4-oxo-
2-cyclopentene-1-carboxylic Acid, 1,1-Dimethylethyl Ester
(2b). Compound 3b (3.10 kg, 10.36 mol) was added to
MTBE (31 L) in a 50-gal, glass-lined carbon steel tank. A
solution of KBr (100 g, 0.84 mol, 0.08 equiv) in deionized
H2O (0.15 L) was added to the reaction solution followed
by the addition of TEMPO (0.13 kg, 0.83 mol, 0.08 equiv).
The solution was cooled to 0 °C, and a premixed solution
of NaOCl (41.8 kg, 3 wt % solution, 561.5 mol, 1.6 equiv)
and NaHCO3 (1.6 kg, 19.04 mol, 1.84 equiv) was added over
50 min, while maintaining the reaction mixture solution at
-5 to 5 °C. The reaction was stirred for 2 h at 0 °C, at
which time HPLC assay showed no alcohol 3b remained.
The reaction mixture was warmed to 20-25 °C over 20 min,
and the layers were separated. The aqueous layer was
extracted with MTBE (2 × 15 L), and the MTBE layers
were combined with the original organic layer. A solution
of 1 N HCl (30 L) containing KI (0.35 kg, 2.11 mol) was
added to the organic layer and stirred. The layers were
separated, and the organic layer was washed with Na2S2O3
(3.2 kg dissolved in 30 L of H2O) and deionized H2O (2 ×
32 L). The organic layer tested negative for hypochlorite
with the use of starch/KI paper. The organic layer was
concentrated by vacuum distillation at 50 °C at 550-650
mmHg to a volume of 14 L. Heptane (20 L) was added, and
the solution was again concentrated to 14 L. The solution
was cooled to 40-50 °C, and heptane (30 L) was added
over 60 min, while maintaining the temperature at 45-
55 °C. The solution was cooled to 0 °C over 3 h and stirred
for 2.5 h. The precipitate was filtered, the tank was rinsed
with a cold 3.3:1 heptane/MTBE (13 L) solution, and the
rinse was filtered over the product cake. The cake was blown
dry under N2 for 2.75 h and then dried under vacuum for
15.5 h at 50 °C to give 2.502 kg (81.2% yield) of ketone 2b
compound 3b had a melting point of 111-112 °C. [R]25
)
D
1
+114 (c 1, MeOH). H NMR (500 MHz, CDCl3): δ 6.1
(bs, 1H), 5.9 (bs, 1H), 5.55 (d, J ) 5.0 Hz, 1H), 4.8 (m,
1H), 4.44 (d, J ) 10.5 Hz, 1H), 2.87 (dd, J ) 14.5, 7.5 Hz,
1H), 2.00 (d, J ) 14.5 Hz, 1H), 1.45 (s, 9H), 1.42 (s, 9H).
IR (KBr): 3413, 2983, 1703, 1491, 1370, 1309, 1255, 1155,
1055 cm-1. MS (ES) m/e (% relative intensity): 300.3 (M+
+ 1, 15), 226.2 (29), 170.1 (100), 126.1 (89), 108.3 (20).
Anal. Calcd for C15H25NO5: C, 60.18; H, 8.41; N, 4.68.
Found: C, 60.59; H, 8.58; N, 4.78.
as white crystals, mp 116-118 °C. HPLC potency ) 99.7%.
1
Corrected yield ) 80.9%. [R]25 ) +123 (c 1, MeOH). H
D
The filtrate was concentrated to approximately 10 L under
vacuum to yield a slurry. The solids were filtered, washed
with 1 L of hexanes, and vacuum dried at 45 °C to obtain
431 g (12% yield) of 3b, 86.2% ee by chiral HPLC assay.
Recovery of Diamine 17-Dihydrochloride. To a 100-
gal glass reactor was charged 150 L of an aqueous sulfuric
acid solution containing a maximum of 68.6 mol diamine 17.
The pH was adjusted to 13 with 50% NaOH solution. The
mixture was extracted with 2 × 115 L of MTBE. The com-
bined organic layers were washed with 60 L of deionized
water. The organic layer was distilled under atmospheric pres-
sure to a volume of 35 L and subsequently diluted with 250 L
of ethanol (denatured with toluene). Hydrochloric acid (32
wt %, 15.5 kg, 136 mol) was then added over 50 min at
15-30 °C. The resulting slurry was cooled to -8 °C, stirred
1 h, and filtered. The filter cake was rinsed with 45 L of
ethanol, dried with nitrogen flow for 3 h, and then vacuum-
dried at 70 °C for 4 days to yield 19.55 kg (55.0 mol, 80%
overall recovery) of 17-dihydrochloride. Capillary electro-
NMR (500 MHz, CDCl3): δ 7.4 (bs, 1H), 6.32 (d, J ) 5.5
Hz, 1H), 5.6 (bs, 1H), 2.87 (d, J ) 18.2 Hz, 1H), 2.70 (d, J
) 18.2 Hz, 1H), 1.43 (s, 18H). 13C NMR (CDCl3): δ 206.1,
170.2, 160.8, 154.9, 136.1, 84.5, 66.1, 46.6, 28.9, 28.4. IR
(KBr): 3419, 2983, 1722, 1487, 1730, 1300, 1259, 1151,
1012 cm-1. MS (ES) m/e (% relative intensity): 254.2 (M+
+ 1, 11), 242.3 (18), 228.2 (13), 186.1 (76), 143.2 (11), 242.3
(100). Anal. Calcd for C15H23NO5: C, 60.59; H, 7.79; N,
4.71. Found: C, 60.57; H, 7.85; N, 4.81.
N,N′-Bis-[(R)-1-phenethyl]-(S,S)-1,2-diamino-1,2-diphen-
ylethane (13). This compound was prepared via modification
of the literature method.7 Sodium sulfate (500 g, 3.5 mol)
was stirred in heptane (1.8 L) in a 5-L three-neck flask
equipped with mechanical stirrer, nitrogen inlet, and ther-
mocouple. Glyoxal (40 wt/wt % in water, 115 mL, 0.96 mol)
was added followed by formic acid (6.0 mL, 0.15 mol).
R-(+)-R-Methylbenzylamine (274 mL, 2.15 mol) was then
added, resulting in a temperature rise to 33 °C. Sodium
sulfate (625 g, 4.4 mol) was added, and the reaction mixture
558
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development

was stirred for 3.5 h. The reaction mixture was filtered, and
the filter cake was rinsed with heptane (600 mL). The filtrate
was concentrated under reduced pressure to obtain N,N′-bis-
[(R)-1-phenethyl]ethane diimine as a clear, dark-orange oil
(270.4 g, 106% yield) which was used without further
purification. H NMR (500 MHz, CDCl3): δ 1.58 (d, J )
6.5 Hz, 6H), 4.5 (q, J ) 6.5 Hz, 2H), 7.25-7.34 (m, 10H),
8.06 (s, 2H).
concentrated under reduced pressure to give a clear, orange
oil (45.5 g). Ethanol (20 mL) was added to the oil, and the
slurry was stirred in an ice/water bath for 30 min. The
mixture was filtered and the solid rinsed with cold ethanol
and vacuum-dried at 50 °C to give 14.0 g (29% yield) of
1
diamine 13 as a white solid, mp ) 110-111 °C. Lit. mp )
1
119-122 °C. [R]25 ) +196 (c 1, MeOH). H NMR (500
D
MHz, CDCl3): δ 1.26 (d, J ) 6.5 Hz, 6H), 2.20 (broad
singlet, 2H), 3.37 (s, 2H), 3.4 (q, J ) 6.5 Hz, 2H), 6.92-
7.27 (m, 20H).
A 500-mL three-neck flask equipped with mechanical
stirrer, 500 mL addition funnel, and Claisen adapter with
septum, nitrogen inlet, and thermocouple was charged with
a solution of N,N′-bis-[(R)-1-phenethyl]ethane diimine (30.1 g,
0.114 mol) in anhydrous diethyl ether (150 mL) via cannula
under nitrogen. The clear, orange solution was cooled to
-78 °C with a dry ice/acetone bath. Phenyl magnesium
chloride (2 M in THF, 171 mL, 0.342 mol) was added over
3.5 h, maintaining the temperature below -73 °C. The
reaction stirred for 1 h at -78 °C. The ice bath was then
removed, and the cloudy mixture was allowed to warm to
0 °C over 45 min. Acetic acid (39 mL, 0.684 mol) in THF
(210 mL) was added slowly to the reaction mixture, keeping
the temperature below 0 °C by using a dry ice/acetone bath.
Deionized water (200 mL) was added to give a clear, biphasic
mixture that was allowed to warm to room temperature. The
mixture was transferred to a separatory funnel, and the layers
were separated. The aqueous layer was extracted with 2 ×
200 mL of MTBE. The combined organic layers were
washed with 1 N NaOH (200 mL), deionized water (200 mL),
and then saturated aqueous NaCl (200 mL). The combined
organic layers were then dried over Na2SO4, filtered, and
Acknowledgment
We are grateful to Professor Peter Wipf for the original
suggestion and encouragement to pursue the epoxide rear-
rangement chemistry and to Professors Marvin Miller,
William Roush, and Paul Wender for numerous helpful
discussions. We thank Mr. Robert Waggoner for sourcing
and procurement of raw materials and Ms. Lisa Zollars for
providing NMR spectral data.
Supporting Information Available
Schemes for the preparation of diamines 19-21; spectral
data and experimental procedures for the rearrangement of
epoxides 4c and i. This material is available free of charge
via the Internet at http://pubs.acs.org.
Received for review October 28, 2006.
OP060225F
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
559