Development and large-scale preparation of an oral TACE inhibitor

Article abstract of DOI:10.1021/op800308t

An efficient, expedient synthesis of BMS-561392, 1, which enabled rapid delivery of drug substance for clinical development is described. The key features of the synthesis include an efficient synthesis of a phenolic α,α-disubstituted amino ester via carbon alkylation without protection of the phenol, an effective enzymatic resolution of this racemic amino ester, and a process for the preparation of a hydroxamic acid drug substance with undetectable levels of hydroxylamine.

Full text of DOI:10.1021/op800308t

Organic Process Research & Development 2009, 13, 510–518
Development and Large-Scale Preparation of an Oral TACE Inhibitor
Scott A. Savage,* Robert E. Waltermire, Silvio Campagna, Shailendra Bordawekar, and Joan Dalla Riva Toma
Bristol-Myers Squibb Company, Research and DeVelopment, One Squibb DriVe, New Brunswick, New Jersey 08903, U.S.A.
Scheme 1. Original synthesis of BMS-561392
Abstract:
An efficient, expedient synthesis of BMS-561392, 1, which enabled
rapid delivery of drug substance for clinical development is
described. The key features of the synthesis include an efficient
synthesis of a phenolic r,r-disubstituted amino ester via carbon
alkylation without protection of the phenol, an effective enzymatic
resolution of this racemic amino ester, and a process for the
preparation of a hydroxamic acid drug substance with undetect-
able levels of hydroxylamine.
Introduction
Hydroxamic acid, 1, is a potent, selective TNF-R convertase
enzyme (TACE) inhibitor with appropriate attributes for
development including excellent pharmacokinetics (PK) in
multiple species and efficacy in preclinical models of rheuma-
toid arthritis.1 From a development perspective, a small-
molecule, orally active, TACE inhibitor was considered to be
a potential high-impact medicine. Thus, from a process research
and development perspective we sought to prepare BMS-
561392, 1, as rapidly as possible to try and limit the impact of
drug substance availability on the rate of development.
The primary challenges of a rapid delivery were synthesis of
the quaternary stereogenic center2 and delivery of drug substance
with no more than trace levels of the mutagen hydroxylamine.
The initial goal of the project team was to rapidly
identify an enantioselective route to BMS-561392, 1, that
was practical, scalable, and reasonably cost-effective. A
preliminary delivery was rapidly needed to produce
supplies for toxicology studies, and ultimately larger
quantities were required for phase I/II studies. Our
Discovery colleagues provided a 12-step linear synthesis,
centered around forming the γ-lactam core from a phenyl
glycine derivative 5 and D-leucine methyl ester1 (Scheme
1). The resulting diastereomers were then chromatographi-
cally separated to give 6. This synthesis was appealing
from a process perspective due to the straightforward
approach along with the availability and extremely low
cost of the starting material D-4-hydroxyphenyl glycine,
2. We took advantage of this and devised a similar
synthesis that hinged upon an enantioselective preparation
of a compound resembling 4 and provided key intermedi-
ates as crystalline solids to facilitate purification.3
Results and Discussion
Our group focused on the set of retrosynthetic discon-
nects shown in Scheme 2. The goal was to identify an
(3) Waltermire, R. E.; Savage, S. A.; Campagna, S.; Magnus, N. A.;
Confalone, P. N.; Yates, M.; Meloni, D. J. (Bristol-Myers Squibb
Company, U.S.A.). Asymmetric synthesis of aminopyrrolidinones.
PCT Int. Appl. WO 2003/104220, A1 20031218 , 2003, and U.S.
Patent 6,770,763, B2 20040803.
* Author for correspondence. E-mail: scott.savage@bms.com.
(1) (a) Maduskuie, T. P., Jr.; Duan, J.; Mercer, S. E. (Dupont Pharma-
ceuticals Company, U.S.A.). Novel Lactam Metalloprotease Inhibitors.
PCT Int. Appl. WO 2002/004416 A2, 20020117, 2002. (b) Duan, J. J.-
W.; Chen, L.; Lu, Z.; Wasserman, Z. R.; Maduskuie, T. P.; Liu, R.-
Q.; Covington, M. B.; Vaddi, K. G.; Qian, M.; Voss, M. E.; Xue,
C.-B.; Hardman, K. D.; Ribadeneira, M. D.; Newton, R. C.; Magolda,
R. L.; Christ, D. D.; Decicco, C. P. Abstracts of Papers; 224th ACS
National Meeting, Boston, MA, U.S.A., August 18-22, 2002,
American Chemical Society: Washington, DC, 2002, MEDI-426.
(2) For reviews and syntheses of related tetra-substituted amino acids see:
(a) Cativiela, C.; Diaz-De-Villegas, M. D. Tetrahedron: Asymmetry
1998, 9, 3517–3599. (b) Ma, D.; Tian, H. Tetrahedron: Asymmetry
1996, 7, 1567–1570.
(4) (a) Vedejs, E.; Fields, S. C.; Schrimpf, M. R. J. Am. Chem. Soc. 1993,
115, 11612–11613. (b) Seebach, D.; Boes, M.; Naef, R.; Schweizer,
W. B. J. Am. Chem. Soc. 1983, 105, 5390–5398. (c) Belokon, Y. N.;
Kochetkov, K. A.; Churkina, T. D.; Ikonnikov, N. S.; Chesnokov,
A. A.; Larionov, O. V.; Parmar, V. S.; Kumar, R.; Kagan, H. B.
Tetrahedron: Asymmetry 1998, 9, 851–857. (d) Ooi, T.; Kameda, M.;
Maruoka, K. J. Am. Chem. Soc. 1999, 121, 6519–6520. (e) Horikawa,
M.; Busch-Petersen, J.; Corey, E. J. Tetrahedron Lett. 1999, 40, 3843–
3846.
(5) Jacques, J.; Collet, A.; Wilen, S. H. Enantiomers, Racemates, and
Resolutions; John Wiley: New York, 1981.
510
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development
10.1021/op800308t CCC: $40.75  2009 American Chemical Society
Published on Web 03/18/2009

Scheme 2. Retrosynthetic analysis
4-hydroxyphenyl glycine ethyl ester with PLE gave similar
results of 88% ee crude ester and 96% ee crystallized
product. The benzyl protected phenol was less effective
in the enzymatic hydrolysis, affording only 70% ee. The
benzyl group significantly lowered the substrate’s aqueous
solubility and consequently the reactivity in the enzymatic
hydrolysis. These results were appealing as we hoped to
avoid phenol protection and deprotection. In addition,
there were obvious cost benefits of alkylating the phenol
with 4-chloromethyl-2-methyl quinoline, 7, after the
resolution. The enzymatic resolution Via hydrolysis of 10
to 11 with PLE7 became the cornerstone of our new
synthesis of BMS-561392.
Enantioselective Synthesis of rr-Disubstituted Amino
Ester 12. Having identified an effective enzymatic resolution
of 10, we began development work towards its efficient
preparation. The Fischer esterification of D-4-hydroxyphenyl
glycine, 2, was straightforward using methanesulfonic acid
(MSA) and ethanol to afford the ethyl ester 8 as the corre-
sponding MSA salt.8 The free amine 8 was crystallized upon
neutralization of the reaction mixture with aqueous sodium
hydroxide and was isolated in 92% yield. Imine formation was
chosen to protect the amine during alkylation as our discovery
colleagues observed that allylation of the corresponding Boc-
protected phenyl glycinate 2 was not selective and resulted in
a mixture of C- and N-alkylation. The crystalline p-tolualde-
hyde-protected imine 9 was prepared in 97% yield by Schiff
base formation by way of azeotropic distillation of water using
toluene. This allowed for a single solvent for both the reaction
and the crystallization. The presence of the unprotected phenol
necessitated double deprotonation to form both the phenolate
and enolate ions with the expectation that the enolate anion
would be more reactive towards an allyl electrophile. Consider-
ing that competitive allylation of the phenol oxygen versus the
enolate carbon could be an issue, we selected a lithium base to
exploit the advantage of the strong Li-O bond9 to minimize
allyl ether formation. The initial condition of 2.1 equiv of lithium
bis(dimethylsilyl)amide (LiHMDS) and 1.05 equiv of allyl
bromide was successful with >99% conversion and almost none
of the allyl ether byproduct (Scheme 4). After in situ hydrolytic
asymmetric route to compounds similar to 4 while
avoiding benzyl protection and deprotection of the phenol.
Three approaches to form chiral R,R-disubstituted amino
esters were investigated in parallel: (1) asymmetric
alkylations of D-4-hydroxyphenyl glycine derivatives,4 (2)
traditional salt resolutions of R-substituted D-4-hydrox-
yphenyl glycine,5 and (3) enzymatic resolutions of R-sub-
stituted D-4-hydroxyphenyl glycine esters.6
The most promising lead came from a reported
enzymatic resolution with pig liver esterase (PLE),
whereby Van Betsbrugge et al. demonstrated the success-
ful enzymatic hydrolysis of R-allylphenyl glycine ethyl
ester6 (Scheme 3). By driving the hydrolysis of the racemic
Scheme 4. Enolate alkylation in the presence of a phenol
Scheme 3. Enzymatic resolution of r-allylphenyl glycine
derivatives
cleavage of the imine, the product 10 was crystallized as the
MSA salt in 80% yield. Lithium tert-butoxide (LiOt-Bu) was
shown to be an equally effective base. On the basis of a
significantly lower cost, lithium tert-butoxide was selected over
ester to 57% conversion, they obtained the R-ester with
95% ee. Our initial screen of the enzymatic hydrolysis of
(7) This work was conducted before BSE/TSE issues became a concern.
If this process is to be utilized for commercial manufacture, a further
screen of enzymes will be conducted to identify one lacking this
concern.
(6) PLE had previously been shown to be effective for the resolution of
analogues of 10. Van Betsbrugge, J.; Tourwe, D.; Kaptein, B.; Kierkels,
H.; Broxterman, R. Tetrahedron 1997, 53, 9233–9240. Moorlag, H.;
Kellogg, R. M. Tetrahedron: Asymmetry 1991, 2, 705–720.
(8) Wasserman, H. H.; Hlasta, D. J.; Tremper, A. W.; Wu, J. S. J. Org.
Chem. 1981, 46, 2999–3011.
(9) Arnett, E. M.; Small, L. E. J. Am. Chem. Soc. 1977, 99, 808–816.
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
511

LiHMDS for scale-up. The use of LiHMDS or LiOt-Bu allowed
the allyl bromide to be charged prior to the formation of the
bis-lithiate with no impact on the level of phenol alkylation.
This was important on scale, as the bis-lithiate was crystalline
and would form a thick slurry in tetrahydrofuran at the optimal
reaction concentration. Charging allyl bromide prior to LiOt-
Bu avoided pre-formation of the enolate which provided a
significant reduction in utilization of tetrahydrofuran. This
change resulted in an increase in volume efficiency and a
decrease in yield loss to aqueous waste streams. Experiments
using potassium tert-butoxide (KOt-Bu) were conducted to
further explore the effect of counterion on alkylation selectivity.
As expected, the carbon versus oxygen selectivity was signifi-
cantly reduced, as was the overall robustness of the allylation
due to tighter temperature constraints and the requirement to
pre-form the potassium enolate prior to the addition of allyl
bromide.
The development of the enzymatic resolution of 10 to 11
began by optimizing the conditions established by Van Bets-
brugge et al. Since PLE is quite expensive, loading studies were
conducted to identify the minimum quantity required for robust
conversion (a range of 50-380 kU/mol of 10 was studied with
a range of 130 to 150 kU/mol of 10 found to be optimal).
Potassium dihydrogenphosphate (0.2 mol %) was used as the
buffer (pH 8), and Triton-100X (2 wt %) was used as a
surfactant to help facilitate phase interaction in this heteroge-
neous reaction. The crystallization was straightforward as
enantiomerically pure 11 can also be conveniently isolated as
an MSA salt with an upgrade in ee, using essentially the
identical conditions as for the racemate 10. As shown in Table
1 (batches 1-7), these conditions resulted in 18 h reaction times,
produced foaming, leading to suspended substrate and ultimately
variable conversion and enantiomeric excess. This problem was
alleviated by switching to a nonfoaming surfactant (Antarox
BL-225). In addition, Tris(hydroxymethyl)aminomethane (TRIS)
buffer replaced potassium dihydrogenphosphate to improve the
solubility of 10. Enzyme loading and reaction temperature were
reevaluated using these new reagents, the results of which are
presented in Table 1 (batches 8-10). These changes resulted
in a significant increase in the rate of reaction (decrease in
reaction time in Table 1 batches 9-15). Looking to improve
the speed of the biomass filtration, the effects of various Celite
grades and loadings on filtration rates were studied. The results
of these experiments showed that the addition of 10 wt %, based
on reaction mass, of Celite 560 notably increased the flow rate
of filtration to ∼2 L/min with a high recovery of ∼97% of the
product. The final, and key change, was to heat the solution of
10 in water to 55 °C prior to pH adjustment and addition of
the enzyme. Performing the neutralization of the methane-
sulfonic acid salt 10 at elevated temperature in the presence of
surfactant was found to afford a stable emulsion. After cooling
the emulsion to 38 °C, the enzyme was added. The culmination
of these improvements was dramatic on production-scale
(batches 11-15). The reaction time was decreased to 2.5 h,
yields were increased to up to 45%, and the selectivity was
consistently 98% ee.
The final step to prepare our desired intermediate, 12, was
to protect the amine in preparation for the ozonolysis and
subsequent reductive amination with D-leucine. What was
expected to be one of the simpler steps in the synthesis proved
to be quite challenging due to the steric hindrance around the
amine and the nucleophilicity of the phenol. Typical conditions
for Boc protection using di-tert-butyl dicarbonate (Boc2O) and
triethylamine were extremely sluggish and nonselective for
reaction at the amine vs the phenol.10 When 4-(N,N-dimethy-
lamino)pyridine was used to catalyze the N-acylation, the
corresponding isocyanate was the predominant product. Potas-
sium carbonate in THF and water was our first reasonable
success using Boc2O, but it was not possible to drive the reaction
past 90% conversion without forming significant levels of the
tert-butyl phenol carbonate of 12. Drawing from our experience
with the the synthesis of 10, we postulated that lithium carbonate
might decrease the reactivity of the phenol. The use of lithium
carbonate in place of potassium carbonate was successful,
allowing the use of excess Boc2O at 40 °C to drive the reaction
to >95% conversion while keeping the level of the tert-butyl
phenol carbonate to <5%. Changing the reaction solvent from
THF to EtOAc enabled an efficient phase separation after the
reaction and an effective crystallization from EtOAc/heptane,
which afforded an 80% yield of 12.
Table 1. Enzymatic resolution of amino ester 10
input PLE temp
reaction yield ee
batch (kg) (kU/mol) (°C) surfactant buffer time (h) (%) (%)
1
2
3
4
5
6
7
45.0
45.0
45.0
45.0
45.0
45.0
30.0
12.0
12.0
12.0
130
130
130
130
130
130
130
130
150
150
150
150
150
150
150
28
28
28
28
28
28
28
28
40
40
38
38
38
38
38
Triton
Triton
Triton
Triton
Triton
Triton
Triton
Antarox TRIS
Antarox TRIS
Antarox TRIS
Antarox TRIS
Antarox TRIS
Antarox TRIS
Antarox TRIS
Antarox TRIS
KH2PO4
KH2PO4
KH2PO4
KH2PO4
KH2PO4
KH2PO4
KH2PO4
18
18
18
18
18
18
18
18
5
5
5
2.5
2.5
2.5
2.5
34
35
37
38
35
47
41
38
34
32
44
41
45
43
44
95
83
88
91
81
78
96
93
98
98
94
98
98
98
98
8
9a
10
11 309
12 311
13 331
14 336
15 328
a Surfactant was increased from 2 to 5 wt % on subsequent batches.
yields around 38%, and optical purities from 78 to 96% ee.
This in-house campaign was effective in that the 113 kg of 11
prepared in batches 1-7, afforded >99.5% ee drug substance
due to optical purity upgrades throughout the remainder of the
synthesis. Before transferring this technology to a contract
manufacturer, the opportunity was taken to further optimize the
enzymatic hydrolysis. The issues focused on were the following:
variable resolution results, long reaction times (18 h), and poor
filtration of enzyme residue. The first challenge was to improve
the heterogeneous nature of the reaction. Triton-100X surfactant
Having developed a robust process for the synthesis of the
chiral R,R-disubstituted amino ester, we scaled up in our pilot
plant (73 kg 12 was produced) and ultimately at a contract
manufacturer (405 kg was produced at g99% ee and g99%
purity). It is important to note that all intermediates in the
synthesis of 12 were isolated as crystalline solids (Scheme 5).
(10) Greene, T. W.; Wuts, P. G. M. Protecting Groups in Organic Synthesis;
John Wiley: New York, 1999.
512
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

Scheme 5. Synthesis of protected amino ester 12
intermediate to extract 15 into water using 33 wt % citric acid.
After phase separation, the acidic aqueous solution of 15 was
basified to enable extraction into EtOAc. This allowed complete
removal of all nonbasic byproduct. Care had to be taken to avoid
heat during the reductive amination and workup, as any amine
15 that cyclized to lactam 16 was not extractable and would
result in diminished yield. The reduction in efficiency during
production was more than compensated by the purification
provided by this extraction. As 16 is the only crystalline
intermediate in this sequence, its crystallization was a critical
aspect of this four-step telescoped process. The EtOAc solution
of 15 was heated to reflux to convert the amine to the
corresponding lactam, 16, and a final solvent exchange to
isopropanol (IPA) allowed for crystallization and isolation of
16 in 60% overall yield from 12 (Scheme 6).
Scheme 6. Synthesis of lactam core to prepare 16
Reductive Amination and Cyclization to Lactam 17. An
objective was to avoid the use of a benzyl-protected phenol in
the ozonolysis step, and we were pleasantly surprised to find
that installation of the quinoline group prior to the ozonolysis
posed no reactivity issues. Analysis found no detectable
oxidation of the quinoline nitrogen during ozonolysis. The
synthesis of the Boc-protected lactam 16 was a four-step
telescoped process, beginning with phenol alkylation in THF
using potassium tert-butoxide, 4-chloromethyl-2-methyl quino-
line, and 0.05 equiv of tetrabutylammonium iodide to afford
13. After neutralization with aqueous acetic acid, the compound
was extracted into EtOAc. An important part of this process
was the subsequent use of an aqueous citric acid wash to remove
residual 4-chloromethyl-2-methyl quinoline and its correspond-
ing impurities. This step is necessary since 13 is not a crystalline
compound. Originally, the ozonolysis was performed in meth-
ylene chloride at <-70 °C. In order to avoid using environ-
mentally unfriendly chlorinated solvents, we examined alter-
native solvents and ultimately selected EtOAc as the solvent.
We found the ozonolysis to be compatible with EtOAc and
that it could be run at a slightly warmer temperature. The
ozonolysis process entailed cooling a solution of 13 in EtOAc
to -65 °C and introducing ozone subsurface. After reaction
completion, nitrogen was introduced subsurface to purge excess
ozone from the organic solution. Triphenylphosphine was added
to reduce the intermediate ozonide to the desired aldehyde 14.
The EtOAc solution of 14 was cooled to -10 °C and treated
with DIPEA, NaBH(OAc)3, and D-leucine methyl ester to effect
a reductive amination.11 While the reaction was straightforward,
the reductive amination process required an extensive workup
due to the lack of crystallinity of aldehyde 14, and to enable
removal of the triphenylphosphine oxide byproduct and residual
triphenylphosphine. It was critical to remove the triphenylphos-
phine and triphenylphosphine oxide as they were found to
inhibit the crystallization of the lactam product 16. The process
we devised took advantage of the dibasic nature of the
As is typical for pharmaceutical processes, the purity of the
final intermediate was extremely important, and thus it was
critical to have a robust process and crystallization. In our case,
the deprotected lactam 17 fit this role perfectly as the simple
Boc cleavage could be carried out easily under a variety of
acidic conditions. This allowed for a thorough investigation of
salt forms and reaction solvents. The acid selection was critical
to provide a crystalline salt that gave the necessary purification
while also having suitable filtration and handling properties.
Again, methanesulfonic acid (MSA) was the chosen acid due
to the excellent purification and solid-state properties of the bis-
MSA salt 17. Isopropanol was initially selected as the solvent
as it provided a fast reaction, efficient impurity removal, and
low product solubility (<0.5 wt %). As the purity of 16
improved over time, a consequence was that the bis-MSA salt
17 began crystallizing out of solution much earlier during the
deprotection reaction. This resulted in a small amount (∼1.5%)
(11) Abdel-Magid, A. F.; Carson, K. G.; Harris, B. D.; Maryanoff, C. A.;
Shah, R. D. J. Org. Chem. 1996, 61, 3849–3862.
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
513

of the starting material 16 trapped within the crystals. The
solution to this problem was to run the reaction in a minimal
amount of methanol (3 kg MeOH/kg 16) at 55 °C to keep the
product in solution throughout the majority of the reaction.
Isopropanol was added after the reaction to complete the
crystallization. This modified procedure effectively converted
all of the 16 to17 without significantly increasing yield loss to
the filtrate. A side benefit was that this new procedure eliminated
the small amount of transesterification to the isopropyl ester
formed during deprotection in isopropanol (Scheme 7).
The biggest hurdle associated with the last step in our
sequence was the removal of excess hydroxylamine. The limit
determined in consultation with our drug safety evaluation group
was to have <5 ppm hydroxylamine in the drug substance. This
level was based both on review of the literature and most
importantly on a decision to have undetectable levels with our
current, best analytical methodology. At the time, our analytical
group had developed an ion chromatography method that
detected hydroxylamine in our compound with a detection limit
of 4 ppm. The original synthesis of BMS-561392 used 10 equiv
of hydroxylamine, and thus the removal of this potential
impurity was a major concern. Using the modified procedure
of sodium methoxide and hydroxylamine hydrochloride, we
were able to reduce the requirements to 4 equiv of NH2OH.
The initial API was the bis-HCl salt of 1, as shown in Figure
1. This target provided a significant challenge, as a portion of
Scheme 7. Boc cleavage to form final intermediate 17
Hydroxamic Acid Preparation and Purification. A major
yield hindrance in the conversion of the methyl ester 17 to the
hydroxamic acid 1 using hydroxylamine and potassium hy-
droxide was the competing hydrolysis to the corresponding
carboxylic acid 18 (Scheme 8).1 The solution to this problem
Figure 1. Solubility of BMS-561392 in isopropanol and water
mixtures.
Scheme 8. Hydroxamate formation to give target 1
the residual hydroxylamine crystallized as the HCl salt during
the bis-hydrochloride API formation. During our process, the
product was extracted into EtOAc by adjusting the pH to neutral
after the hydroxamate formation using HCl. After this aqueous
workup, there was typically >1000 ppm NH2OH in the organic
stream. A crystallization study showed that the optimal HCl
salt procedure could only tolerate ∼30 ppm of hydroxylamine
in solution prior to crystallization and still meet the limit. In
order to achieve this 30 ppm level, four water washes were
utilized after the initial aqueous workup to remove the excess
hydroxylamine. Our first and only delivery of BMS-561392
bis-hydrochloride provided 1.6 kg for use in preclinical toxicol-
ogy studies.
The bis-hydrochloride salt of 1 had significant issues with
crystal form which were related to the release of hydroxylamine.
The combination of its hygroscopicity (typically 5% water by
KF) and acidic nature resulted in hydrolysis during both drying
and storage to form detectable levels of hydroxylamine. Due
to this issue, we began investigating other crystalline forms of
1. Our solution came quickly, as we discovered the existence
of a crystalline neutral form. This neutral form was found to
have suitable properties for development, including excellent
solid-state stability and PK in multiple species. This new final
form not only solved all drying and storage issues, but afforded
new process opportunities to remove hydroxylamine during
was to utilize sodium methoxide to neutralize the hydroxylamine
hydrochloride salt.12 This change, along with minimization of
the water content of the reaction, was found to be effective at
reducing ester hydrolysis, as the methoxide would redundantly
react to afford the starting material 17. A small amount of
hydrolysis to 18 is unavoidable due to the small amount of water
present in the hydroxylamine hydrochloride.
(12) Thouin, E.; Lubell, W. D. Tetrahedron Lett. 2000, 41, 457–460.
514
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

crystallization. Unlike hydroxylamine hydrochloride, hydroxy-
lamine is soluble in alcohols as well as water. The crystalline
free base of 1 was first discovered by crystallization from
isopropanol, but it was slightly too soluble (∼2 wt %) to be
the process crystallization solvent. While looking for a suitable
single solvent, as well as antisolvent combinations, an interesting
trend was discovered. The solubility of 1 in IPA/water
combinations exhibited nonlinear behavior whereby the product
is not appreciably soluble in either single solvent, but was
surprisingly very soluble (16 wt % at 20 °C) in 80 vol %
isopropanol with 20% water (Figure 1).
Taking advantage of this solubility data, a new crystallization
procedure was identified. As stated earlier, the product 1 was
initially extracted into ethyl acetate. After solvent exchange from
EtOAc to IPA, water was added to the desired 4:1 ratio to afford
maximum solubility. This allowed 1 to be soluble in ∼3.5 L/kg
at 55-60 °C, improving the volume efficiency of our process.
Water was then added to a ratio of 2:3, and seed crystals were
added. After cooling to 45 °C, the remainder of the water was
added to adjust to a final composition of 80% water and 20%
isopropanol.
The discovery of the crystalline-free base of BMS-561392
and development of the isopropanol/water crystallization al-
lowed for significant improvements in removal of residual
hydroxylamine. The neutral form crystallization could tolerate
as much as 200 ppm of residual hydroxylamine in the rich,
organic stream, allowing the use of only two or three water
washes after the reaction. Of equal importance was that this
form does not liberate hydroxylamine upon drying or storage
as did the bis-HCl salt. Overall our team prepared over 150 kg
of BMS-561392 over a 2-year period, and all batches contained
no detectable level of hydroxylamine.
heated to 78 °C and aged for 2 h. The reactor was cooled to 55
°C, and the reaction was determined complete by HPLC. The
reactor was cooled to 40 °C and H2O (468 L) was added. The
reactor was cooled to 5 °C, and 17% aqueous sodium hydroxide
(prepared from 142 L of H2O, 71.3 kg of 50% sodium
hydroxide {0.89 kmol, 1.93 equiv}) was slowly added over 2 h
to provide a slurry of 8 at pH ) 7.0 to 7.5. The slurry was
stirred at 5 °C for 1 h. Ester 8 was isolated by centrifugation
and washed with H2O (3 × 106 kg). The solids were vacuum
dried at 50-55 °C to a constant weight to yield 82 kg (90%
yield) of a light-yellow crystalline solid. 1H NMR (400 MHz,
DMSO-d6) δ 7.18 (d, J ) 8.5 Hz, 2H), 6.73 (d, J ) 8.6 Hz,
2H), 4.40 (s,1H), 4.15-3.95 (m, 2H), 1.12 (t, J ) 7.1 Hz, 3H).
13C NMR (100 MHz, DMSO-d6) δ 174.7, 157.1, 131.7, 128.2,
115.4, 60.6, 58.0, 14.3. Analysis Calculated for C10H13NO3: C,
61.53; H, 6.71; N, 7.18. Found: C, 61.54; H, 6.57; N, 7.11.
(E)-Ethyl 2-(4-Hydroxyphenyl)-2-(4-methylbenzylidene-
amino)acetate (9). An inerted 300-gal reactor was charged with
toluene (685 kg) and cooled to 0 °C. Amine 8 (80 kg, 0.41
kmol) was added followed by p-tolualdehyde (51.4 kg, 0.43
kmol, 1.04 equiv) with a toluene (7 kg) flush. The reaction mass
was heated to reflux (∼90 °C) and distilled until approximately
180 kg of distillate was collected (∼115 °C). Fresh toluene,
equivalent in mass to the distillate, was added. The reaction
mass was cooled to 85 °C and aged for 1 h, then cooled to 20
°C over 1 h and aged for 2 h. Imine 9 was isolated by
centrifugation and washed with a toluene (200 kg)-heptane
(157 kg) mixture in three portions. The product 9 was vacuum
dried at 50 to 55 °C to a constant weight to yield 116 kg (95%
yield) of a white to light-yellow crystalline solid. 1H NMR (400
MHz, CDCl3) δ 8.25 (s, 1H), 7.67 (d, J ) 8.1 Hz, 2H), 7.26
(d, J ) 8.6 Hz, 2H), 7.16 (d, J ) 8.0 Hz, 2H), 6.76 (d, J ) 8.6
Hz, 2H), 5.13 (s, 1H), 4.25-4.11 (m, 2H), 2.33 (s, 3H), 1.20
(t, J ) 7.1 Hz, 3H). 13C NMR (100 MHz, CDCl3) δ 172.0,
164.0, 156.2, 141.8, 132.7, 129.3, 129.2, 129.1, 128.8, 115.9,
75.7, 61.6, 21.5, 14.0.
Ethyl 2-Amino-2-(4-hydroxyphenyl)pent-4-enoate Meth-
ane Sulfonate (10). An inerted 300-gal glass-lined reactor was
charged with tetrahydrofuran (215.5 kg, THF) and cooled to
-5 °C. Imine 9 (62.0 kg, 0.21 kmol) was added, followed by
allyl bromide (26.7 kg, 0.22 kmol, 1.06 equiv) with a THF (2.5
kg) chase. The reactor was cooled to -5 °C and charged with
2 M lithium tert-butoxide in THF (192.6 kg, 0.44 kmol, 2.1
equiv) over 1 h at e5 °C. The reaction mass was sampled for
conversion after 30 min and determined complete by HPLC
(criteria g30:1 ratio of 10:9). The reaction was quenched by
adding 2 M aqueous hydrochloric acid (326 kg, 0.625 kmol, 3
equiv), pH ) 1 after quench. Heptanes (107 kg) were added,
and the layers were separated, retaining the product-rich aqueous
phase. The pH of the aqueous phase was adjusted to 8-9 by
adding a 16% aqueous sodium hydroxide solution (prepared
by combining 32 kg of 50% aqueous sodium hydroxide {0.40
kmol, 1.9 equiv} and 62 kg of H2O). Sodium chloride (31 kg)
and ethyl acetate (369 kg) were added, and the layers were
separated, retaining the product-rich organic phase. The batch
was heated to reflux to distill off residual THF. The distillation
was deemed complete when the batch temperature reached 78
°C. Ethyl acetate was used to adjust to a final volume of ∼250
Conclusions
From the inexpensive starting material, D-4-hydroxyphenyl
glycine, the first enantioselective synthesis of the TACE
inhibitor BMS-561392 was identified. The synthesis consists
of a linear sequence of eleven chemical transformations with
eight isolations of crystalline intermediates. The features of the
synthesis include efficient synthesis of a phenolic R,R-disub-
stituted amino ester via carbon alkylation without protection
of the phenol, an effective enzymatic resolution of this racemic
amino ester, and a process for the preparation of a hydroxamic
acid drug substance with undetectable levels of hydroxylamine.
Discovery of the final form as neutral, stable, and crystalline
proved very important to the development of 1. The robustness
of the processes are supported by multiple process demonstra-
tions on manufacturing scale for early intermediates and pilot
scale for the final intermediate and API. This efficient, expedient
synthesis of BMS-561392 enabled its rapid entry into clinical
studies.
Experimental Section
(R)-Ethyl 2-Amino-2-(4-hydroxyphenyl)acetate (8). An
inerted 300-gal reactor was charged with absolute ethanol (123
kg) and cooled to 10 °C. D-4-Hydroxyphenyl glycine (78 kg,
0.47 kmol) was added, followed by methanesulfonic acid (86
kg, 0.89 kmol, 1.93 equiv) at e78 °C. The reaction mass was
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
515

L. The batch was cooled to 60 °C, and 2-propanol (9.6 kg)
was added. After cooling to 50 °C, methanesulfonic acid (20.1
kg, 0.21 kmol, 1 equiv) was added. The batch was then cooled
to 0 °C and aged for 1 h. Product 10 was isolated by
centrifugation and washed with ethyl acetate (2 × 56 kg).
Sulfonate 10 was vacuum dried at 50-55 °C to a constant
weight to afford 56 kg (80% yield) of a white to light yellow
C14H21NO6S: C, 50.74; H, 6.39; N, 4.23; S, 9.68. Found: C,
50.43; H, 6.06; N, 4.08; S, 9.74.
(R)-Ethyl 2-(tert-Butoxycarbonyl)-2-(4-hydroxyphenyl)-
pent-4-enoate (12). An inerted 300-gal glass-lined reactor was
charged sequentially with 11 (55 kg, 0.17 kmol), lithium
carbonate (26.8 kg, 0.36 kmol, 2.2 equiv), ethyl acetate (150
kg), di-tert-butyl dicarbonate (72.4 kg, 0.33 kmol, 1.95 equiv),
and water (269 L). The batch was heated to 40 ( 2 °C and
aged for 14 h. The reaction was sampled and determined
complete by HPLC. (criteria: >20:1, 12:11). After cooling the
reaction mass to 20 °C, acetic acid was slowly charged (32.9
kg, 0.55 kmol, 3.2 equiv). The pH was checked and adjusted
to <6.5 with acetic acid. The reaction was allowed to separate,
and the aqueous layer was removed. The organics were washed
with water (269 L). A vacuum distillation at 100 mmHg was
used to remove ethyl acetate from the organics by reducing
the volume to 400 L. The reaction was sampled for solvent
composition (criteria: >95:1 heptane:ethyl acetate, <1 wt % 12
in supernatant). The resulting slurry was cooled to 20 °C.
Product 12 was isolated by filtration on a nutsch and washed
with heptane (2 × 75 kg). The solids were vacuum dried at
50-55 °C to a constant weight to yield 43 kg (78% yield) of
a white to light-yellow crystalline solid. 1H NMR (400 MHz,
CDCl3) δ 7.21 (d, J ) 8.7 Hz, 2H), 6.57 (d, J ) 7.7 Hz, 2H),
6.19 (s, 1H), 5.75-5.58 (m, 1H), 5.18 (d, J ) 14.0 Hz, 1H),
5.14 (d, J ) 8.2, 1H), 4.25-4.00 (m, 2H), 3.48-3.30 (m, 1H),
3.25-3.10 (m, 1H), 1.44 (s br, 9H), 1.16 (t, J ) 14.2 Hz, 3H).
13C NMR (100 MHz, CDCl3) δ 172.5, 155.7, 154.0, 132.5,
130.6, 127.0, 119.4, 115.5, 79.9, 64.4, 62.2, 38.1, 28.4, 14.0.
Analysis calculated for C18H25NO5: C, 64.46; H, 7.51; N, 4.18.
Found: C, 60.03; H, 7.38; N, 4.14.
(R)-Ethyl 2-(tert-Butoxycarbonyl)-2-(4-((2-methylquino-
lin-4-yl)methoxy)phenyl)-4-oxobutanoate (14). A 400-L in-
erted glass-lined reactor was charged with THF (150 kg)
followed by potassium tert-butoxide (48.2 kg, 0.43 kmol,
1.2 equiv) and agitated for 30 min. A 3000-L inerted glass-
lined reactor was charged with 12 (120 kg, 0.358 kmol)
and THF (532 kg). While stirring, the solution of 12 was
cooled to 5 °C, and the prepared potassium tert-butoxide
solution was charged slowly, maintaining the batch
temperature below 10 °C. The reactor contents were mixed
for 15 min at 5 °C. Then 7 (68.5 kg charge corrected for
purity, 0.357 kmol, 1.0 equiv) was charged, followed by
tetrabutylammonium iodide (6.6 kg, 17.9 mol, 0.05 equiv).
After the addition, the reactor contents were heated to
40-45 °C for 2 h. The reaction was sampled, and HPLC
was used to determine completion (criteria: >30:1 13:12).
Once complete, the reactor contents were cooled to 20-25
°C. Water (725 L) was then charged, followed by glacial
acetic acid (18.2 kg, 0.286 kmol, 0.8 equiv) and ethyl
acetate (690 L). The reactor contents were mixed for 20
min and the layers allowed to separate. The aqueous phase
was removed. Aqueous citric acid solution (728 kg,
prepared from 630 L of water and 98 kg of citric acid
monohydrate {0.46 kmol, 1.3 equiv}) was charged to the
organic phase. The reactor contents were mixed for 10
min followed by layer separation. The aqueous layer was
discarded. The organic phase was washed with an aqueous
1
crystalline solid. H NMR (400 MHz, DMSO-d6) δ 9.90 (s,
1H), 8.83 (s, 3H), 7.29 (d, J ) 8.8 Hz, 2H), 6.85 (d, J ) 8.8
Hz, 2H), 5.83-5.71 (m, 1H), 5.30 (d, J ) 17.0 Hz, 1H), 5.24
(d, J ) 10.1 Hz, 1H), 4.30-4.15 (m, 2H), 2.97 (dd, J ) 6.8,
14.5 Hz, 2H), 2.35 (s, 3H), 1.18 (t, J ) 7.1 Hz, 3H). 13C NMR
(100 MHz, CD3OD) δ 171.4, 160.4, 131.1, 128.9, 127.1, 123.7,
117.4, 66.1, 64.7, 41.5, 40.0, 14.7.
(R)-Ethyl 2-Amino-2-(4-hydroxyphenyl)pent-4-enoate
Methane Sulfonate (11). Racemate 10 (331.4 kg, 1.0 kmol),
water (3164 L), tris(hydroxymethyl)aminomethane (20.7 kg,
0.17 kmol), and Antarox BL-225 (15 kg) were charged into a
3000-gal vessel. The pH of the solution was raised to 7.8-8.2
by adding 6 N NaOH solution (∼130 kg). The mixture was
heated to 55 °C to consume the methanesulfonic acid, causing
the starting material to oil out of solution. Utilizing a continuous
pH monitor, the pH of the mixture was adjusted to 7.8-8.2
using 6 N NaOH (∼30 kg). The batch was cooled to 38 °C,
and pig liver esterase (PLE, 3.5 kg) was added.13 The solution
was stirred while maintaining the desired pH range by adding
6 N NaOH as needed. After 2.5 h a sample was pulled to
determine the enantiomeric excess of the desired amino ester
(criteria g93% ee). After cooling to 20 °C, water (242 L) and
ethyl acetate (2740 L) were added while continuing to monitor
pH, adding 6 N NaOH (∼130 L) as necessary to keep the pH
in range. The pH was adjusted to 9.5 to 9.8 using 6 N NaOH
(∼90 kg), and Celite (390 kg) was then added to the mixture.
The batch was filtered and the cake washed sequentially with
ethyl acetate (635 L), water (120 L), and ethyl acetate (635 L).
The combined filtrate and washes were recharged to the vessel
and allowed to separate. The aqueous layer was removed and
back extracted with ethyl acetate (1800 L). The combined
organic layers were washed with a saturated aqueous sodium
chloride solution (500 L). An atmospheric distillation was
performed to reduce the solution to ∼1550 L. After cooling to
50 °C, isopropanol (45 kg) and methanesulfonic acid (MSA,
48.0 kg, 0.5 kmol) were added to form the desired salt. The
resulting slurry was cooled to 10 °C and aged for one hour.
The product was isolated by centrifugation and washed with
ethyl acetate (700 L). Sulfonate 11 was vacuum dried at 50 °C
to yield 148 kg (45% yield) of a white to light-yellow crystalline
solid. 1H NMR (400 MHz, DMSO-d6) δ 9.90 (s, 1H), 8.83 (s,
3H), 7.29 (J ) 8.8 Hz, 2H, d), 6.85 (d, J ) 8.8 Hz, 2H),
5.83-5.71 (m, 1H), 5.30 (d, J ) 17.0 Hz, 1H), 5.24 (d, J )
10.1 Hz, 1H), 4.30-4.15 (m, 2H), 2.97 (dd, J ) 6.8, 14.5 Hz,
2H), 2.35 (s, 3H), 1.18 (t, J ) 7.1 Hz, 3H). 13C NMR (100
MHz, CD3OD) δ 171.4, 160.4, 131.1, 128.9, 127.1, 123.7,
117.4, 66.1, 64.7, 41.5, 40.0, 14.7. Analysis calculated for
(13) This preparation was conducted in the late 1990s using PLE certified
to originate from food-grade animals. As the issue of TSE/BSE
emerged, a new synthesis that eliminated the need for PLE was devised
to support further clinical studies. This new synthesis will be the subject
of a future communication.
516
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

sodium chloride solution (720 kg of 10 wt %). To the
organic phase, water (690 L) was added, followed by 50
wt % aqueous NaOH solution to adjust the pH to between
6.5 and 7.5 (typically 50 L). After the pH adjustment, the
layers were separated, and the aqueous layer was removed.
The organic phase was washed again with aqueous sodium
chloride (720 kg of 10 wt %). The organic layer was
distilled down to ∼600 L, and ethyl acetate (∼1200 L)
was charged at a rate such that the volume remained at
∼600 L. The organics were sampled, and GC was used
to determine completion (criteria: < 0.5% THF) The
solution was transferred to an inerted 5000-L Hastelloy
reactor and diluted with ethyl acetate (1400 L). The
reaction was cooled to -65 to -70 °C. The oxygen flow
on the ozone generator was set to 100 Nm3/h, and oxygen
was introduced subsurface to the reactor. The ozone
generator was turned on, and a stream of ozone (18.4 kg,
0.383 kmol, 1.07 equiv) and oxygen was introduced
subsurface to the reactor over a period of ∼4 h (target 5
kg of O3/h). The ozone generator was switched off, and
the reactor was purged with oxygen, followed by nitrogen
to purge oxygen and unreacted ozone. The reaction mass
was sampled for completion (criteria: area ratio of 13 to
ozonide e 0.2). Once reaction completion was established,
a 0 °C, ethyl acetate solution of triphenylphosphine (113
kg, 0.430 kmol, 1.20 equiv in 500 L of ethyl acetate) was
added to the ozonide solution. The reaction was warmed
to 0 °C over 4 h and held for 12 h at 0-10 °C. The
reaction was sampled for completion (criteria: area ratio
of ozonide to 14 e 0.2). Once the reduction was complete,
the organic phase was washed with water (980 L). The
resulting aldehyde 14 in ethyl acetate was transferred to
a 5000-L glass-lined reactor for the synthesis of 15. The
solution was cooled to -10 °C. Diisopropylethylamine
(60.1 kg, 0.465 kmol, 1.3 equiv) and D-leucine methyl
ester hydrochloride (84.5 kg 0.465 kmol, 1.3 equiv) were
added. The reactor contents were aged at 0 to 5 °C for
1 h, and then sodium triacetoxyborohydride (110 kg 0.519
kmol, 1.45 equiv) was added. The reaction mass was aged
at 0 ( 5 °C for 2 h and sampled for completion (criteria:
conversion of 14 to 15 g 99.8 by area %). Once the
reaction was deemed complete, the excess sodium triac-
etoxyborohydride was quenched by adding water (2000
kg). The aqueous and organic phases were separated.
Toluene (150 kg) was added to the organic phase, and it
was then extracted with 28 wt % aqueous citric acid four
times (1000 L, 720 L, 720 L, 720 L). The organic layer
was discarded, and the aqueous citric acid extracts were
combined in a second 5000-L vessel. Ethyl acetate (788
L) and toluene (75 kg) were added as a wash, and the
aqueous layer of 15 was transferred back to the original
glass-lined reactor. EtOAc (550 L) was added to the
aqueous phase. The mixture was cooled to 5 ( 5 °C, and
then 30% aqueous sodium hydroxide (∼900 kg) was added
to adjust to a pH range of 4-5. The layers were separated
(the organic phase was retained), and the aqueous phase
was extracted with EtOAc (790 L). The second EtOAc
extraction was combined with the first, and the organics
were washed with water (500 L). EtOAc (1550 L) was
added, and the solution was heated to reflux. An initial
azeotropic distillation was performed to remove water and
raise the bp to ∼77 °C by concentrating to 700 L. The
solution was held at reflux for 8 h, and the reaction was
sampled for completion (criteria: g 99.8% conversion of
15 to 16). A constant volume distillation was then initiated
where isopropanol (∼2000 L) was added at approximately
the same rate of distillation to keep the total volume of
the solution at ∼700 L. The organic phase was sampled
and analyzed for EtOAc/IPA ratio (criteria: <0.2% EtOAc).
The solution was then cooled gradually to 5 °C, at a rate
of 10 °C/h. Seed crystals (0.5 kg) were added to induce
crystallization, and the resulting slurry was aged at 5 °C
for 2 h. The product was filtered by centrifugation and
washed with cold isopropanol (2 × 125 L). The product
16 was dried in a conical dryer at 40 °C to a constant
weight to yield 121.5 kg (59%) of a white to off-white,
1
crystalline solid. H NMR (400 MHz, CDCl3) δ 8.09 (d,
J ) 8.4 Hz, 1H), 7.91 (d, J ) 8.4 Hz, 1H), 7.72 (dd, J1
) 8.3, J2 ) 8.3 Hz, 1H), 7.53 (dd, J1 ) 8.3, J2 ) 8.3 Hz,
1H), 7.48 (s, 1H), 7.46 (d, J ) 9.2 Hz, 2H), 7.00 (d, J )
9.2 Hz, 2H), 5.64 (s, 1H), 5.49 (s, 2H), 4.93 (s,Br, 1H),
3.58 (s, 3H), 3.40 (m, 2H), 2.89 (t, 1H), 2.77 (s, 1H),
2.76 (s, 3H), 1.81-1.72 (m, 2H), 1.58 (m, 1H), 1.42 (s,
9H), 0.97 (m, 6H). 13C NMR (100 MHz, CD3OD) δ 173.7,
171.4, 158.9, 157.9, 154.7, 147.7, 142.0, 133.2, 129.4,
127.3, 126.1, 124.1, 122.6, 120.3, 114.9, 66.8, 63.2, 52.4,
52.1, 40.5, 36.9, 28.3, 25.4, 24.6, 23.3, 21.1. MS (DCI/
NH3): m/z: 576.3 (M + 1), 520.2, 459.1. Analysis
Calculated for C33H41N3O6: C, 68.85; H, 7.18; N, 7.30.
Found: C, 68.69; H, 7.06; N, 7.14.
(R)-Methyl 2-((R)-3-Amino-3-(4-((2-methylquinolin-4-yl)-
methoxy)phenyl)-2-oxopyrrolidin-1-yl)-4-methylpen-
tanoate Bis-MSA Salt (17). To an inerted 200-gal glass-
lined reactor was charged 16 (30 kg, 0.052 kmol) and
methanol (90 kg). Methanesulfonic acid (5.8 kg, 0.06
kmol, 1.15 equiv) was gradually charged at 25 °C. The
batch was heated slowly to 55 °C. Additional methane-
sulfonic acid (5.8 kg, 0.06 kmol, 1.15 equiv) was added
gradually over 30 min at 55 °C. The reactor contents were
aged at 55 °C for 2 h and sampled for reaction completion
(HPLC criteria: <0.5 area % 16). Isopropanol (214 kg)
was charged at 55 °C, and then the batch was cooled to
15 °C at a rate of 1 °C/min and aged at 15 °C for 1 h.
The product 17 was isolated on a centrifuge, and the cake
was washed with isopropanol (3 × 50 kg). The solids were
dried under vacuum at 50 °C to a constant weight to yield
1
34.9 kg (90%) of a white crystalline solid. H NMR (400
MHz, DMSO-d6) δ 9.07 (s, 4H), 8.51 (d, J ) 8.4 Hz,
1H), 8.32 (d, J ) 8.6 Hz, 1H), 8.14 (s, 1H), 8.14 (t, J )
8.6 Hz, 1H), 7.96 (t, J ) 8.6 Hz, 1H), 7.61 (d, J ) 8.9
Hz, 2H), 7.39 (d, J ) 9.0 Hz, 2H), 5.94 (s, 2H), 4.80 (dd,
J1 ) 4.0 Hz, J2 ) 11.3 Hz, 1H), 3.62 (s, 3H), 3.51 (t, J
) 9.5 Hz, 1H), 3.15-3.28 (m, 1H), 3.01 (s, 3H),
2.65-2.75 (m, 1H), 2.48-2.60 (m, 2H), 2.43 (s, 6H), 1.85
(t, J ) 12.2 Hz, 1H), 1.52-1.71 (m, 2H), 0.96 (d, J )
6.3 Hz, 3H), 0.93 (d, J ) 6.10 Hz, 3H). 13C NMR (100
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
517

MHz, DMSO-d6) δ 172.9, 172.5, 160.6, 159.7, 156.5,
139.1, 136.1, 131.2, 130.0, 128.9, 126.3, 126.0, 122.1,
117.2, 67.8, 64.3, 53.4, 50.1, 41.5, 40.1, 38.1, 33.8, 26.2,
24.0, 22.0, 21.5. Analysis Calculated for C30H41N3O10S2:
C, 53.96; H, 6.19; N, 6.29; S, 9.60. Found: C, 53.83; H,
5.97; N, 6.15; S, 9.76.
batch temperature was set to 55 °C, and purified water
(50 kg) was charged through a cartridge filter over 30 min.
The batch was heated to 60 °C, aged for 15 min, and then
cooled to 55 °C in 30 min. At 55 °C, a slurry of milled
seeds of 1 (1 L, ∼200 g of solids) in 1:4 isopropanol/
water (volume ratio) was charged. The batch was cooled
from 55 to 45 °C in 90 min. At 45 °C, purified water (67
kg) was charged through a cartridge filter gradually over
a period of 3.5 h. The batch was cooled from 45 to 20 °C
in 2 h, aged for 30 min, and filtered through a 36 in. nutsch
filter using a Dacron filter bag. The filter cake was washed
three times with a mixture of isopropanol/water (first wash:
38 kg of water, 8 kg of isopropanol; second and third
washes: 19 kg of water, 4 kg of isopropanol). The product
1 was dried in a tray dryer under vacuum at 50 °C to a
constant weight to yield 12.4 kg (87%) of a white,
crystalline solid. 1H NMR (400 MHz, DMSO-d6) δ 10.86
(s, 1H), 8.92 (s, 1H), 8.10 (dd, J1 ) 8.3 Hz, J2 ) 0.8 Hz,
1H), 7.97 (dd, J1 ) 8.3 Hz, J2 ) 0.8 Hz, 1H), 7.74 (ddd,
J1 ) 8.3 Hz, J2 ) 8.3 Hz, J3 ) 1.2 Hz, 1H), 7.58 (ddd, J1
) 8.3 Hz, J2 ) 8.3 Hz, Je ) 1.2 Hz, 1H), 7.54 (s, 1H),
7.36 (d, J ) 8.9 Hz, 2H), 7.07 (d, J ) 9.1 Hz, 2H), 5.57
(s, 1H), 4.55 (dd, J1 ) 9.1 Hz, J2 ) 5.8 Hz, 1H),
3.45-3.31 (m, 2H), 2.66 (s, 3H), 2.20 (bs, 2H), 2.18-2.01
(m, 2H), 1.71-1.62 (m, 1H), 1.57-1.43 (m, 2H), 0.93
(d, J ) 6.3 Hz, 3H), 0.90 (d, J ) 6.3 Hz, 3H). 13C NMR
(100 MHz, CDCl3) δ 176.6, 166.7, 158.5, 156.9, 147.3,
142.3, 137.4, 129.3, 128.8, 127.0, 125.8, 123.9, 123.6,
120.1, 114.3, 66.2, 62.1, 50.1, 39.8, 37.6, 36.5, 24.9, 24.1,
22.8, 21.9. Analysis Calculated for C27H32N4O4: C, 68.05;
H, 6.78; N, 11.76. Found: C, 68.00; H, 6.77; N, 11.77.
(R)-2-((R)-3-Amino-3-(4-((2-methylquinolin-4-yl)metho-
xy)phenyl)-2-oxopyrrolidin-1-yl)-N-hydroxy-4-methylpen-
tanamide (1). To an inerted 30-gal glass-lined reactor was
charged hydroxylamine hydrochloride (10 kg, 0.14 kmol,
4.7 equiv) and methanol (15.6 kg). The batch temperature
was set to 50 °C, and a 25 wt % solution of sodium
methoxide in methanol (64.6 kg, 16.15 kg, 0.3 kmol, 10
equiv) was charged followed by a methanol rinse of the
charging line. The reactor contents were heated to 55 °C
and aged for 15 min. The batch was then cooled to 25 °C
and filtered through a 36 in. nutsch filter using a
polypropylene filter bag. The filtrate was collected in a
100-gal glass-lined reactor and cooled to 10 °C. Final
intermediate 17 (20 kg, 0.03 kmol) was added, and the
batch was warmed to 25 °C for 1 h. The reactor contents
were sampled for reaction completion (HPLC criteria:
>99.5 area % 1). Once the reaction was deemed complete,
∼55 kg of 2 N HCl solution (prepared using 126 kg of
purified water and 30 kg of concentrated HCl) was added,
and the reaction mass was sampled for pH measurement
(acceptance criteria: pH ≈ 7.0). The batch was vacuum
distilled at ∼35 °C to remove ∼20 L of methanol. The
batch temperature was set to 25 °C, and the batch was
charged with purified water (75 kg) followed by ethyl
acetate (135 kg) and mixed for 15 min, and the layers
were separated. The organic phase was washed three times
with purified water (120 kg, 120 kg, 75 kg). The batch
was vacuum distilled at 35 °C to remove ∼70 kg of
solvent (ethyl acetate, methanol). Isopropanol (62 kg) was
charged, and the batch was vacuum distilled at 35 °C to
remove 55 kg of solvent (ethyl acetate/isopropanol azeo-
trope). Isopropanol (55 kg) and purified water (20 kg) were
charged, and the batch was vacuum distilled at 35 °C to
remove 25 kg of solvent (water/isopropanol azeotrope).
The batch was sampled for analysis (criteria: <5 area %
ethyl acetate, ∼4:1 isopropanol:water volume ratio, 10 L
of isopropanol/water per kg of 1. The batch temperature
was set to 40 °C, and the reactor contents were discharged
to a clean Teflon drum. The contents of the Teflon drum
were transferred to the cleaned reactor through a cartridge
filter, followed by an isopropanol chase (0.5 kg). The batch
was vacuum distilled at 35 °C to remove 35 kg of solvent
(isopropanol/water azeotrope). After the distillation, the
Acknowledgment
We thank Pat Confalone for many helpful scientific discus-
sions and especially his unwavering support and encouragement
throughout this project. We thank our Discovery colleagues,
especially Carl Decicco, Jim Duan, and Tom Maduskuie for
countless discussions and suggestions. We are grateful for the
support of our many collaborators including Reginald Cann,
Mark Guzman, John Castoro, Charles Ray, Robert Wethman,
Pascal Toma, Stephen Anderson, Sridhar Desikan, Ioannis
Valvis, and Michael Hrytsak. We thank Wendel Doubleday and
David Kronenthal for their careful review and editing of the
manuscript. We especially thank Sigma-Aldrich Fluka and
Uquifa for their scale-up efforts on the enzymatic resolution
and ozonolysis, respectively.
Received for review December 12, 2008.
OP800308T
518
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development