Practical asymmetric synthesis of RO5114436, a CCR5 receptor antagonist

Article abstract of DOI:10.1021/op100020z

A practical asymmetric synthesis of a 3,7-diazabicyclo[3.3.0]octane derivative (1), a representative of a new class of potent CCR5 receptor antagonists, is described. The benzylamine stereogenic center of 1 was introduced by a ruthenium-catalyzed asymmetric reductive amination using (R)-MeOBIPHEP as ligand. Aldehyde 4, prepared by Parikh-Doering oxidation, was used without workup in the reductive amination reaction, which not only simplified the process but also overcame the instability of 4. The 3,7-diazabicyclo[3.3.0]octane core was obtained by a [3 + 2] cycloaddition.

Full text of DOI:10.1021/op100020z

Organic Process Research & Development 2010, 14, 592–599
Practical Asymmetric Synthesis of RO5114436, a CCR5 Receptor Antagonist
Xiaojun Huang,* Erin O’Brien, Felicia Thai, and Gary Cooper*
Chemical Synthesis, Roche Palo Alto LLC, 3431 HillView AVenue, Palo Alto, California 94304, U.S.A.
Scheme 1. Retrosynthetic analysis of 1
Abstract:
A practical asymmetric synthesis of a 3,7-diazabicyclo[3.3.0]octane
derivative (1), a representative of a new class of potent CCR5
receptor antagonists, is described. The benzylamine stereogenic
center of 1 was introduced by a ruthenium-catalyzed asymmetric
reductive amination using (R)-MeOBIPHEP as ligand. Aldehyde
4, prepared by Parikh-Doering oxidation, was used without
workup in the reductive amination reaction, which not only
simplified the process but also overcame the instability of 4. The
3,7-diazabicyclo[3.3.0]octane core was obtained by a [3 + 2]
cycloaddition.
1. Introduction
The chemokine receptor CCR5 is a clinically validated target
for Human Immunodeficiency Virus (HIV) disease and a
potentially interesting target for the inflammation therapy area.
The first small-molecule CCR5 antagonist on the market,
maraviroc (Selzentry), was approved by the FDA for treatment
of HIV-1 infection.1 Medicinal chemistry research at Roche led
to the discovery of a series of 3,7-diazabicyclo[3.3.0]octane
compounds,2 represented by RO5114436 (1), that are potent
CCR5 antagonists. Compound 1 also showed high potency in
functional assays for inflammation. The PK properties of 1 were
superior to those of maraviroc in preclinical species, including
rat, dog, and monkey. In order to satisfy our need for kilogram
amounts of 1 for toxicity studies, an enantioselective synthesis
was envisioned.
We disclose herein our studies on the practical synthesis of
1. We focused on the set of retrosynthetic disconnection shown
in Scheme 1. Peptide coupling between primary amine 2 and
acid 3 would give the final active pharmaceutical ingredient
(API). Amine 2 could be synthesized by reductive amination
of aldehyde 4 with amine 5.
(Scheme 3). Claisen condensation between commercially avail-
able ester 8 and EtOAc in the presence of potassium tert-
butoxide produced ꢀ-ketoester 9 in near-quantitative yield.
ꢀ-Ketoester 9 was a liquid, which was used as crude in the
next step after aqueous workup with 3 N HCl. At least 2.5 equiv
of potassium tert-butoxide was required for full conversion (3.0
equiv of potassium tert-butoxide was used for pilot-plant-scale
preparations). It was also very important to add EtOAc last to
achieve full conversion. The major byproduct in the crude
product mixture was ethyl acetoacetate, which can be minimized
by lowering the reaction temperature from ambient to 0 °C.
ꢀ-Ketoester 9 was also prepared in quantitative yield by the
condensation of m-fluoroacetophenone with diethylcarbonate
using sodium hydride in THF. However, the cost of m-
fluoroactophenone was much higher than that of ethyl 3-fluo-
robenzoate (8);3 thus, this route was not selected.
2. Results and Discussion
2.1. Synthesis of aldehyde 4. Medicinal chemists had
prepared aldehyde 4 from BOC-protected ꢀ-aminoester 7 by
DIBAL-H reduction (Scheme 2). Ester 7 was prepared from
the corresponding acid 6, which was commercially available
but very expensive.
The preparation of enantiomerically pure 12 was a key
transformation for the whole process. One approach for the
transformation could be the resolution of (()-12 via diastere-
omeric salt formation. Other possible approaches included an
asymmetric Michael addition reaction to an appropriately
substituted cinnamic acid ester,4 and the enantioselective
The enantioselective synthesis of ꢀ-aminoester 12 was
investigated in order to satisfy our needs for large amounts
* Authors for correspondence. E-mails: xiaojun.huang5@gmail.com; gary.
cooper@roche.com.
(1) (a) Haycock-Lewandowski, S. J.; Wilder, A.; Åhman, J. Org. Process
Res. DeV. 2008, 12, 1094–1103. (b) Åhman, J.; Birch, M.; Haycock-
Lewandowski, S. J.; Long, J.; Wilder, A. Org. Process Res. DeV. 2008,
12, 1104–1113, and references therein.
(3) The list price was $245/100 g for m-fluoroactophenone and $71/100
g for ethyl 3-fluorobenzoate (8) at Fluorochem.
(4) See for example, Davies, S. G.; Ichihara, O.; Lenior, I.; Walters, I. A.
J. Chem. Soc., Perkin Trans. 1 1994, 1411.
(2) Lee, E. K.; Melville, C. R.; Rotstein, D. M. Chem. Abstr. 2005, 144,
69821. PCT Int. Publication Number WO/2005/121145 A2, 2005.
592
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development
10.1021/op100020z  2010 American Chemical Society
Published on Web 03/10/2010

Scheme 2. Medicinal chemistry synthesis of 4
Scheme 3. Synthesis of ꢀ-aminoester 12
Table 1. Reductive amination results for converting 9 into 12a
entry
catalyst/(loading)
solvent/(vol)
NH4Oac (equiv)
T (°C)
time (h)
H2 (psig)
conv.b (%)
yieldc/(ee)d, (%/%)
1
2
10(2.5%)
11(2%)
11(2%)
11(2%)
11(2%)
11(2%)
11(2%)
11(2%)
11(2%)
11(2%)
11(2%)
11(2%)
11(2%)
11(2%)
11(1%)
11(0.5%)
11(2%)
11(2%)
11(2%)
TFE/(38)
TFE/(10)
MeOH/(10)
EtOH/(10)
IPA/(10)
TFE/(20)
TFE/(10)
TFE/(5)
TFE/(10)
TFE/(10)
TFE/(10)
TFE/(10)
TFE/(10)
TFE/(10)
TFE/(10)
TFE/(10)
TFE/(10)
TFE/(10)
TFE/(10)
5
5
80
80
67
80
85
80
80
80
80
80
80
80
80
80
75
75
70
60
75
16
16
24
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
30
40
40
100
100
78
52e/(99.6)
55e/(99.8)
3
5
45
2.1
4
5
45
77
4.4
5
5
45
76
3.4
6
5
45
100
100
95
46
7
5
45
46
8
5
45
38
9
2
45
100
100
92
100
100
100
100
90
45
10
11
12
13
14
15
16
17
18
19f
7.5
10
5
45
41
45
30
100
200
300
100
100
100
100
100
44
5
45
5
43
5
47e
26e/(99.7)
38
5
5
93
78
100
5
17
4
45e/(100)
a Conditions: Ru catalyst, NH4OAc, H2, (TFE), heat. b Conversion was determined on the basis of area % by HPLC (210 nm). c Yield was determined on the basis of area
% by HPLC (210 nm). d Ee for the HCl salt of 12 is shown in parentheses. e Isolated yield. f Pilot-plant run (1.6 kg).
hydrogenation of a corresponding enamine.5 The preliminary
efforts for the chiral resolution of (()-12 and the enantiose-
lective hydrogenation of the unprotected enamine were unsuc-
cessful. The direct conversion of ꢀ-ketoesters to chiral ꢀ-ami-
noesters by catalytic asymmetric hydrogenation has been
described in the literature.6 The catalytic asymmetric amination
and hydrogenation of ketoester 9 was investigated, and the results
are summarized in Table 1. The reaction was first carried out
following the literature procedure using the Ru-ClMeOBIPHEP
catalyst 10. Full conversion was achieved, and ꢀ-aminoester
12 was isolated in 52% yield and 99.6% ee after recrystal-
lization from hexane as an acetic acid salt (entry 1, crude
mixture has ∼98% ee). The absolute configuration of 12 was
confirmed by comparing BOC-protected 13 to the ethyl ester
derivative of commercially available 6. We then turned our
attention to a similar catalyst, Ru-MeOBIPHEP catalyst 11,
which is more readily available to us.7 The reductive
amination performed equally well with catalyst 11, giving
product 12 in 55% isolated yield and 99.8% ee as a HCl
salt after a single recrystallization from isopropyl acetate
(entry 2, crude mixture has ∼96% ee). We tested different
solvents, including MeOH, EtOH, isopropanol (IPA), and
2,2,2-trifluoroethanol (TFE) (entries 2-5). TFE was the only
solvent to give reasonable yields. The reasons for the
beneficial results of using TFE were not disclosed in the
literature.6 Ten volumes of TFE were required to give full
conversion and reasonable yield (entries 6-8). It was found
(5) (a) Hsiao, Y.; Rivera, N. R.; Rosner, T.; Krska, S. W.; Njolito, E.;
Wang, F.; Sun, Y.; Armstrong, J. D., III; Grabowski, E. J. J.; Tillyer,
R. D.; Spindler, F.; Malan, C. J. Am. Chem. Soc. 2004, 126, 9918–
9919. (b) Hansen, K. B.; Hsiao, Y.; Xu, F.; Rivera, N.; Clausen, A.;
Kubryk, M.; Krska, S.; Rosner, T.; Simmons, B.; Balsells, J.; Ikemoto,
N.; Sun, Y.; Spindler, F.; Malan, C.; Grabowski, E. J. J.; Armstrong,
J. D., III. J. Am. Chem. Soc. 2009, 131, 8798–8804.
(7) Schmid, R.; Foricher, J.; Cereghetti, M.; Scho¨nholzer, P. HelV. Chim.
Acta 1991, 74, 370–389. We have multikilograms of catalyst 11 in
Roche stock.
(6) Bunlaksananusorn, T.; Rampf, F. Synlett 2005, 2682–2684.
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
593

Scheme 4. Synthesis of aldehyde 4
that the equivalency of ammonium acetate was very impor-
tant for this transformation (entries 7, 9-11). Greater than
7.5 equivalents of ammonium acetate led to lower conver-
sion. It was unclear why the conversion was affected. Four
equivalents of ammonium acetate were used for the final
scale-up studies. Hydrogen pressure (40-300 psig) was not
critical to the outcome of the reaction (entries 7, 12-14). A
middle point of hydrogen pressure at 100 psig was applied
for the large-scale synthesis. The effect of catalyst loading
was also investigated (entries 12, 15-16). Greater than 1
mol % catalyst was required to achieve good conversion and
2 mol % of catalyst 11 was used in the scale-up run. The
reaction was very slow when the temperature was lower than
70 °C (entries 17-18). Since lower temperature gave less
byproduct, the temperature for the large-scale synthesis was
set at 75 °C. Different additives were tested, including acetic
acid, p-toluenesulfonic acid, methanesulfonic acid, pyridine,
triethylamine, di-tert-butyldicarbonate. All these additives led
to either lower conversion and/or decreased yields. The pilot-
plant run, using 2 mol % catalyst 11, 4 equiv of ammonium
acetate, 10 volumes of TFE, hydrogen at 75 °C and 100 psig,
gave chiral amine 12 in 45% isolated yield and 100% ee as
the hydrochloride salt (entry 19). The major byproduct was
the cinnamic acid ethyl ester by the elimination of ammonia
from product 12, which was confirmed by 1H NMR analysis.
A dimer byproduct was also observed by LC-MS. Even
though the yield was moderate (45% over two step), large
amounts of enantiomerically pure 12 could be prepared in
two steps from readily available starting materials, which
satisfied our needs at the moment.
BOC protection of the primary amine 12 followed by
DIBAL-H reduction of 13 was investigated for the synthesis
of aldehyde 4 following Medicinal Chemistry procedures. The
DIBAL-H reduction gave inconsistent results, with a mixture
of desired aldehyde 4, over-reduced alcohol 14, and some
starting ester 13. In one of the better cases, when 2.4 equiv of
DIBAL-H was used, aldehyde 4 was isolated in 82% yield,
along with 13% of alcohol 14 after flash chromatography. The
need to conduct this reaction at -78 °C also encouraged
exploration of alternate routes. A two-step process, reduction
of 13 to alcohol 14 followed by oxidation of 14 to aldehyde 4,
was employed (Scheme 4). Several conditions were tried for
the reduction of ester 13 to alcohol 14 including sodium
borohydride in EtOH under reflux, and lithium aluminum
hydride in cold (0 °C) THF. Sodium borohydride gave
inconsistent results, but lithium aluminum hydride gave 14 in
near quantitative yield. Red-Al (Vitride) was the reagent of
choice for scale-up because, as a convenient 65% solution in
toluene, it is a less hazardous reagent, and it gave the same
clean, full conversion of 14. The amount of 2-methoxyethanol,
which was generated on workup for Red-Al reduction, was not
significant since no adduct peak with amine 5 was observed
after oxidation and reductive amination. The fact that there were
five following steps and three isolations also made 2-methoxy-
ethanol not a residue of concern. Process-friendly nitroxyl
radical (TEMPO)-based oxidation was investigated first for the
conversion of alcohol 14 to aldehyde 4. It also gave inconsistent
results and was hard to drive to high conversion; overoxidation
to the corresponding acid was significant. We examined a
stabilized formulation of IBX (SIBX) to oxidize 14 in acetone.8
About 90% conversion was achieved, with no overoxidation
to the corresponding acid when 1.2 equiv of SIBX was used.
However, SIBX is expensive ($1750/kg, quoted for 4 kg from
SIMAFEX in September, 2007), and the safety issues about
using it in the pilot plant were still a concern. We finally turned
our attention to activated DMSO oxidation procedures. It was
quickly realized that Parikh-Doering oxidation (SO3 ·py,
DMSO, and Et3N) was the preferred method for this transfor-
mation.9 Aldehyde 4 was usually obtained in greater than 90%
yield based on HPLC. Methylsulfanylmethyl ether 15 (4-8%)
was the major byproduct for this process. The purification of
aldehyde 4 was difficult since it was not a crystalline compound
and was unstable in air. This problem was addressed by using
the oxidation reaction mixture directly to the next (reductive
amination) step without workup. The reaction performed equally
well as when using purified 4. The dimethylsulfide (DMS)
byproduct was carried to the next step and purged off with
nitrogen after the reductive amination, using bleach as a trap.
This procedure was carried out in the pilot plant, and the odor
from DMS was well-contained.
2.2. Synthesis of Amine 5. The synthesis of amine 5 started
with N-benzylglycine (16) and N-benzylmaleimide (17) (Scheme
5). The bicyclic core was created by a [3 + 2] cycloaddition of
16, 17, and formaldehyde.10 Despite what might be expected
by the mechanisms proposed in the literature, which involve
evolution of water, it was found that aqueous formaldehyde
(8) Ozanne, A.; Pouyse’gu, L.; Depernet, D.; Franc¸ois, B.; Quideau, S.
Org. Lett. 2003, 5, 2903.
(9) Parikh, J. P.; Doering, W. E. J. Am. Chem. Soc. 1967, 89, 5505–
5507.
(10) Pennell, A. M. K.; Aggen, J. B.; Wright, J. J. K.; Sen, S.; Chen, W.;
Dairaghi, D. J.; Zhang, P. Chem. Abstr. 2005, 143, 306291. PCT Int.
Publication Number WO/2005/084667 A1, 2005.
594
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development

Scheme 5. Synthesis of amine 5
Table 2. Effect of acid additives on hydrogenation of 22a
(formalin) was a better source of formaldehyde for this particular
[3 + 2] cycloaddition. The use of paraformaldehyde was not
preferred because of the sublimation of paraformaldehyde to
the condenser. The reaction was carried out at the boiling point
of the toluene/water mixture, with formalin being added slowly
to control reaction/distillation rate. Chemoselective debenzy-
lation of N-benzyl tertiary amines in 18, using ammonium
formate in MeOH catalyzed by Pd/C, gave 19. Electronic factors
must have played a pivotal role in determining the selectivity
of the debenzylation reaction. Red-Al reduction under forcing
conditions (105 °C) converted 19 to the monobenzyl protected
diamine 20. Crude 20 was coupled with acid chloride 21 under
Schotten-Baumann conditions to give amide 22,11 which was
an oil and used in the next step without purification. EDCI-
mediated coupling of 20 with 4,6-dimethylpyrimidine-5-car-
boxylic acid also successfully produced 22.12
Deprotection of 22 was complicated by undesired side
reactions and the fact that the product was an oil that was
difficult to isolate. The N-methyl analogue of 5 was the major
side product, along with unidentified impurities on the baseline.
Transfer hydrogenation of 22 with ammonium formate required
forcing conditions to reach completion and generated the
N-methyl analogue of 5 as a significant impurity that carried
through the synthesis. An extensive screen of classical hydro-
genation conditions including solvent and catalyst showed that
10% Pd/C with acetic acid in MeOH worked best. Under these
conditions, both temperature and hydrogen pressure could be
raised to achieve a faster reaction rate, without the generation
of the N-methyl impurity. Unfortunately, the acetic acid salt of
5 failed to crystallize despite extensive solvent screening. A
search was conducted to find an acid additive that would, in
addition to accelerating the debenzylation of 22, provide a
crystalline product to enable purification of 5 and be suitable
acid
conversionb (%)
product form
acetic
oxalic
succinic
100
45
oil
oil
oil
oil
oil
oil
oil
oil
oil
oil
oil
solidc
oil
oil
oil
oil
100
100
100
100
58
phenylsuccinic
benzylsuccinic
glutaric
L-tartaric
citric
75
salicyclic
100
100
60
malonic
phenylmalonic
benzylmalonic
tartronic
100
69
100
0
96
2,5-dihydroxybenzoic
1-naphthoic
tricarballylic
a Reaction was performed with 1.0 equiv of acid in MeOH (10 vol) at 40 °C
under H2 (50 psig) with 10% Pd/C (10 wt %) for 24 h. b Conversions were
determined by HPLC area % (210 nm). c Crystallized from 2-propanol.
for scale-up. Screening of acids revealed that, while some
additives expedited the hydrogenation, only benzylmalonic acid
also afforded a crystalline product (Table 2). Hydrogenolysis
needed to be conducted in a solvent that the product was soluble
in so that it could be filtered away from the catalyst. MeOH
was chosen as the solvent because the resulting salt was soluble
in MeOH at room temperature but not in IPA. IPA proved to
be a good recrystallization solvent.
2.3. Coupling of 4 and 5. Reductive amination of aldehyde
4 with amine 5 was carried out using sodium triacetoxyboro-
hydride (STAB) in dichloromethane (Scheme 6).13 Aldehyde
4 was used as the crude reaction mixture from the Parikh-
Doering oxidation without workup. Amine 5 was used directly
as the benzylmalonic acid (BMA) salt, avoiding a low-yielding
(due to the high water solubility of 5) free amine formation
step. The BOC-protected amine 23 was not crystalline; there-
fore, after workup it was used directly in the next step without
purification. Deprotection of 23 was accomplished using aque-
ous HCl/toluene at 50 °C, giving amine 2 as a dry foam. An
(11) Kuo, S.-C.; Tsai, D. J.-S.; Liao, H. Chem. Abstr. 2006, 145, 124603.
PCT Int. Publication Number WO/2006/074264 A2, 2006.
(12) (a) McCombie, S.; Tagat, J. R.; Vice, S.; Lin, S.-I.; Steensma, R.;
Palani, A.; Neustadt, B.; Baroudy, B.; Strizki, J.; Endres, M.; Cox,
K.; Dan, N.; Chou, C.-C. Bioorg. Med. Chem. Lett. 2003, 13, 567–
571. (b) Palani, A.; Shapiro, S.; Clader, J. W.; Greenlee, W. J.; Vice,
S.; McCombie, S.; Cox, K.; Strizki, J.; Baroudy, B. M. Bioorg. Med.
Chem. Lett. 2003, 13, 709–712. (c) Draper, R. W. Chem. Abstr. 2005,
142, 176442. PCT Int. Publication Number WO/2005/007608 A2,
2005.
(13) Abdel-Magid, A. F.; Carson, K. G.; Harris, B. D.; Maryanoff, C. A.;
Shah, R. D. J. Org. Chem. 1996, 61, 3849–3862.
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
595

Scheme 6. Coupling of the fragments
extensive effort to find a solid derivative of 2 to effect
purification and facilitate isolation resulted in the finding that a
1:1 mixture of 2 and L-tartaric acid gave a nicely behaved solid
in good yield. Although the purity of 2 was modestly improved,
X-ray powder diffraction analysis showed that this salt was
completely amorphous. A crystalline derivative was not found.
Thus, amine 2 was treated with L-tartaric acid in a mixture of
EtOH and EtOAc to afford the tartaric acid salt of 2 in 76%
over all yield for five steps.
Completion of the synthesis required coupling of acid 3 with
amine 2. Acid 3 was converted to acid chloride 24 by reaction
with oxalyl chloride (cat. DMF) and isolation as a stable toluene
concentrate. The tartaric acid salt of amine 2 was converted to
its free base using Me-THF/3 N NaOH. Acylation of free amine
2 in Me-THF with the toluene solution of acid chloride 24 in
the presence of triethylamine gave the free base of API 1.
Despite a considerable effort, of the two salts found, only the
HCl salt of 1 was found to be useful. Treatment of an acetone/
water (95/5) solution of 1 with conc. HCl gave a nicely
crystalline solid (1·HCl, 1:1 molar ratio, mp 149-150 °C) in
86% yield that proved to be a stable monohydrate.
by SFC as the trifluoroacetamide using the following system:
Mettler-Toledo/Berger Analytical SFC with (R,R)-Whelk-O1
column, 4.6 mm × 250 mm, 5 um particles. Mobile phase
consisted of 95% carbon dioxide and 5% ethanol containing
0.4% (v/v) isopropylamine at a flow rate of 2.5 mL/min and
125 bar backpressure. Column temperature was 40 °C, detection
was UV at 262 nm. Run time was 6 min.
3-(3-Fluorophenyl)-3-oxopropionic Acid Ethyl Ester (9).
To a stirred and cooled (-10 °C) solution of potassium tert-
butoxide in THF (1.0 M, 33.77 kg, 37.44 mol) in an 80-L glass-
lined reactor was added ethyl 3-fluorobenzoate 8 (2.11 kg, 12.55
mol). EtOAc (3.01 kg, 34.16 mol) was added to the cooled
(-10 °C) solution over 90 min at -10 to -5 °C. After 1 h, the
mixture was treated with 3 N HCl (15.0 kg, 42.9 mol) and
concentrated under reduced pressure (30-125 Torr) at 25-30
°C to remove the solvents with low boiling points. The mixture
was extracted with MTBE (16.5 kg), and the organic layer was
washed with water (15.3 kg), a mixture of saturated aqueous
NaHCO3 (5.0 kg)/saturated aqueous Na2CO3 (10.0 kg), and with
brine (10.0 kg). Solvents were removed under reduced pressure
(30-125 Torr) to give 9 (2.90 kg, 90% pure, 2:1 ketone/enol
form based on 1H NMR, ∼99% yield) as an oil,14 which was
used directly to the next chemical step. An analytically pure
sample of 9 was obtained by column chromatography and
3. Conclusion
In summary, a convergent 13-step route (8 longest linear
steps) to 1 was developed and used to produce highly chemical
and chiral pure material suitable for toxicity studies. All the
chromatographic purifications were removed. The chiral amine
center was produced by a highly enantioselective reductive
amination reaction using chiral ruthenium catalyst 11 (95-96%
ee for crude mixture, and >99.5% ee after isolation). The
Parikh-Doering oxidation was used to efficiently produce
aldehyde 4, which was used for the next reductive amination
reaction directly without workup. This process was critical since
it not only simplified the overall process but also overcame the
instability of aldehyde 4. The overall yield for this process was
about 29% starting from ethyl 3-fluorobenzoate (8).
1
characterized. H NMR (300 MHz, CDCl3) 1.27 (t, J ) 7.16
Hz, 2 H), 1.35 (t, J ) 7.16 Hz, 1 H), 3.98 (s, 1.4 H), 4.17-4.35
(m, 2 H), 5.66 (s, 0.3 H), 7.11-7.22 (m, 0.3 H), 7.27-7.79
(m, 3.7 H), 12.56 (s, 0.3 H). MS m/z 211.1 [M + H]+. HPLC tR
) 7.12, 8.96 min.
(S)-3-Amino-3-(3-fluorophenyl)propionic Acid Ethyl Es-
ter Hydrochloride Salt (12 ·HCl). Ketoester 9 (3.38 kg, ca.
90%, pure, ∼14.5 mol) and TFE (48 kg) were added succes-
sively to a nitrogen-inerted 40-L glass-lined reactor with
NH4OAc (2.46 kg, 57.9 mol) and Ru[(R)-MeOBIPHEP](OAc)2
(11, 68 g, 0.072 mol),. The mixture was heated at 75 °C under
hydrogen at 100 psig. After 30 h, the reaction achieved full
conversion. The solvents was removed under reduced pressure
(30-125 Torr). Isopropyl acetate (37 kg) was added, and the
mixture was washed with saturated aqueous Na2CO3 (25 kg,
pH ) 9) and 1:1 brine/water (22 kg). HCl in IPA (4.82 N, 2.8
kg, 14.8 mol) was added to the organic phase, and the solvent
4. Experimental Section
General. HPLC analysis was performed using the following
system: Waters 2690 HPLC with Zorbax SB-C8 column 4.6
mm × 75 mm, 3.5 µm particles; mobile phase consisting of
solvent A, 0.1% trifluoroacetic acid in water, solvent B, 0.1%
trifluoroacetic acid in MeCN. Gradient from 10-90% mobile
phase B in 10 min; λ ) 210 nm. Chiral purity was monitored
(14) Several stripping to dryness operations remain that can be optimized
for mulitkilogram scale by using solvent exchange operations. This
was not completed due to the termination of the work.
596
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development

was removed under reduced pressure (30-125 Torr) at 45 °C
until the volume was reduced to about 10 L. The mixture was
slowly cooled to -10 °C and filtered, and the solids were
washed with isopropyl acetate and dried in an oven under
reduced pressure (30-50 Torr) to give 12 ·HCl (1.60 kg, 45%
yield, 100% ee) as a white solid. Mp 156.0-159.0 °C. IR νmax
(KBr) 3432, 2894, 2700, 1738, 1596, 1524, 1266, 1028 cm-1.
1H NMR (300 MHz, CDCl3) 1.16 (t, J ) 7.16 Hz, 3 H), 3.01
(dd, J ) 16.58, 7.91 Hz, 1 H), 3.15-3.33 (m, 1 H), 4.05 (q, J
) 7.16 Hz, 2 H), 4.78 (br s, 1 H), 6.95-7.13 (m, 1 H),
7.27-7.40 (m, 3 H), 8.84 (br s, 3 H). MS m/z 212.1 [M +
overnight. The reaction achieved full conversion to give
aldehyde 4 (88% purity by HPLC area), which was used directly
in the next step. An analytically pure sample of 4 was obtained
by column chromatography over silica gel and characterized.
Mp 63.7-66.2 °C. 1H NMR (300 MHz, CDCl3) 1.42 (s, 9 H),
2.79-3.15 (m, 2 H), 5.19 (br s, 2 H), 6.91-7.16 (m, 3 H),
7.31 (dt, 1 H), 9.74 (dd, J ) 2.26, 1.51 Hz, 1 H). MS m/z 212.1
[M - t-Bu + 2H]+.
2,5-Dibenzyltetrahydropyrrolo[3,4-c]pyrrole-1,3-dione (18).
Toluene (42 kg) was added to a mixture of N-benzylglycine
(16, 2.16 kg, 13.11 mol) and N-benzylmaleimide (17, 1.76 kg,
9.38 mol) in a 120-L glass-lined reactor, and the resulting
solution was heated to just under reflux (∼105 °C). Formal-
dehyde/water (1.06 kg, 37% w/w, 13.1 mol) was added at a
rate of 300 mL/h while collecting toluene/water distillate. After
the completion of the reaction, the reaction mixture was cooled,
and the solvent was distilled under reduced pressure (30-125
Torr) at 50 °C. MeOH (28 L) was added and then distilled off,
the mixture was cooled to ambient temperature, and more
MeOH (22.4 L) was added. Water (15 L) was slowly added to
the solution over 2 h, and the slurry was stirred at ambient
temperature overnight and then filtered. The solid was washed
with MeOH/water (3 × 4 L, 60:40 v/v) and dried at 50 °C
under reduced pressure (30-50 Torr) to give 18 (2.65 kg, 88%
yield, 99% purity by HPLC area) as a light-gray solid. 1H NMR
(300 MHz, CDCl3) 2.13-2.58 (m, 2 H), 3.07-3.25 (m, 2 H),
3.30 (d, J ) 10.2 Hz, 2 H), 3.55 (s, 2 H), 4.69 (s, 2 H),
6.90-7.67 (m, 10 H). MS m/z 321.1 [M + H]+.
H]+. [R]20 ) +4.9 (CH3OH). HPLC tR ) 4.0 min.
D
(S)-3-tert-Butoxycarbonylamino-3-(3-fluorophenyl)propi-
onic Acid Ethyl Ester (13). To a 40-L glass-lined reactor was
added successively amine 12 (460 g, 1.86 mol), Boc2O (405 g,
1.86 mol), Me-THF (4.6 L), and saturated aqueous NaHCO3
(2.66 kg, 2.79 mol) at ambient temperature. After stirring
overnight, the two phases were separated, and the organic layer
was washed with water (4.6 L) and brine (4.6 L). The organic
layer was distilled under atmosphere pressure to give 13 (97%
purity by HPLC area), which was used to the next step without
further purification. An analytically pure sample of 13 was
obtained by recrystallization from heptane and characterized.
Mp 56.0-57.0 °C. 1H NMR (300 MHz, CDCl3) 1.17 (t, J )
7.16 Hz, 3 H), 1.42 (s, 9 H), 2.63-3.03 (m, 2 H), 4.07 (q, J )
7.16 Hz, 2 H), 5.07 (br. s., 1 H), 5.58 (br. s., 1 H), 6.88-7.11
(m, 3 H), 7.28-7.34 (m, 1 H). MS m/z 256.1 [M - t-Bu +
2H]+. Anal. Calc’d for C16H22FNO4: C, 61.72; H, 7.12; N, 4.50.
Found: C, 61.97; H, 7.03; N, 4.70.
2-Benzyloctahydropyrrolo[3,4-c]pyrrole (20). A mixture
of 18 (2.62 kg, 8.18 mol), ammonium formate (0.57 kg, 0.90
mol), 10% Pd/C (0.26 kg, 62.4% water wet) and MeOH (21
kg) was heated at 60 °C. After 5 h, the reaction mixture was
cooled to ambient temperature, and the catalyst was filtered off
through a pad of Solka Floc. The solvent was removed under
reduced pressure (30-125 Torr) at 40 °C, and the residue was
treated with toluene (23 kg). The mixture was washed with
water, and the organic layer was concentrated to about half
volume under reduced pressure (30-125 Torr). The solution
(containing crude 19) was added slowly to a Red-Al solution
(4.13 kg, 65% in toluene, 13.3 mol) in toluene (2.6 L) at 50
°C. The reaction mixture was stirred under reflux, achieving
full conversion in 5 h. The mixture was cooled to 0 °C and
slowly treated with 15% aqueous NaOH solution. The organic
layer was concentrated under reduced pressure (30-125 Torr)
at 50 °C to give 20 (1.47 kg, ∼89% yield, 81% purity by HPLC
area) as an oil. 1H NMR (300 MHz, CDCl3) 2.09-2.46 (m, 2
H), 3.52-2.92 (m, 8 H), 3.52 (s, 2 H), 7.13-7.38 (m, 5 H).
MS m/z 203.1 [M + H]+1.
[(S)-1-(3-Fluorophenyl)-3-hydroxypropyl]carbamic Acid
tert-Butyl Ester (14). Red-Al (65 wt % in toluene, 1.19 kg,
3.83 mol) over 15 min was added to a stirred and cooled (0-15
°C) solution of 13 (ca. 1.86 mol) in Me-THF (6.9 L) in a 40-L
glass-lined reactor. The mixture was warmed to 24 °C after
addition. After 5 h, the reaction mixture was cooled to 10 °C,
quenched with 10% aqueous NaOH (4.6 L), and extracted with
toluene (4.6 L). The organic layer was washed with 10%
aqueous NaOH (2.3 L) and two times with brine (2 × 4.6 L).
The organic layer was concentrated under reduced pressure
(30-125 Torr) to give crude 14 (97% purity by HPLC area)
as a colorless viscous oil, which was used in the next step
directly. An analytically pure sample of 14 was obtained by
1
column chromatography over silica gel and characterized. H
NMR (300 MHz, CDCl3) 1.44 (s, 9 H), 1.69-1.90 (m, 1 H),
1.93-2.30 (m, 1 H), 2.89 (br s, 1 H), 3.56-3.79 (m, 2 H),
4.79-4.98 (m, 1 H), 5.12 (dd, J ) 8.57, 4.05 Hz, 1 H),
6.89-7.14 (m, 3 H), 7.28-7.38 (m, 1 H). MS m/z 214.1 [M -
t-Bu + 2H]+.
[(S)-1-(3-Fluorophenyl)-3-oxo-propyl]carbamic Acid tert-
Butyl Ester (4). Triethylamine (0.75 kg, 7.4 mol) was added
to a stirred and cooled (0 °C) solution of crude 14 (∼1.86 mol)
in DMSO (1.0 L, 13.9 mol) and DCM (2.5 L) in a 40-L glass-
lined reactor. In another 5-L flask, DMSO (1.0 L, 13.9 mol)
and pyridine (0.45 L, 5.6 mol) were added to sulfur trioxide
pyridine complex (45% SO3, 0.89 kg, 5.6 mol) at ambient
temperature. After agitating for 20 min, the resulting suspension
was added to the previously formed, cooled solution of alcohol
14 over 15 min. The mixture was warmed to 20 °C and stirred
(5-Benzylhexahydropyrrolo[3,4-c]pyrrol-2-yl)-(4,6-dim-
ethylpyrimidin-5-yl)methanone (22). To a suspension of
commercially available 4,6-dimethylpyrimidine-5-carboxylic
acid11 (1.015 kg, 6.67 mol) and DMF (5.6 mL) in acetonitrile
(7.1 L) was added oxalyl chloride (0.89 kg, 7.0 mol) at such a
rate as to maintain the reaction temperature between -5 to 5
°C. The mixture was then aged at 0 °C for 2 h to give a solution
of acid chloride 21.
In another reactor, to a heterogeneous mixture of K3PO4
(1.56 kg, 7.34 mol) and K2HPO4 (2.09 kg, 12.0 mol) in water
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
597

(3.4 L) and acetonitrile (6 L) at 0 °C was added a solution of
20 (1.484 kg, 81% purity by HPLC area) in acetonitrile (1.7
L).5 After 2 h at 0 °C, the solution of acid chloride 21 prepared
above was added at such a rate as to maintain the temperature
below 10 °C. After stirring at 10 °C overnight, water was added
(10 L). The organic phase was distilled under reduced pressure
(30-125 Torr), and EtOAc (11 L) was added. The mixture was
washed with 50% K2HPO4 aqueous solution (2 × 3 L), water
(6 L), and brine (6 L). The organic layer was decolorized with
charcoal (500 g) overnight. The mixture was then filtered
through Solka Floc, and the cake was washed with EtOAc (1
L). The filtrate was concentrated under reduced pressure
(30-125 Torr) to give 22 (1.92 kg, ∼86% yield, 84% purity
by HPLC area) as a honey-colored oil. An analytical pure
sample of 22 was obtained by column chromatography over
and brine (11 kg) to give a solution of adduct 23 (∼898 g,
89% purity by HPLC area), which was used in the next step
directly. 1H NMR (300 MHz, CDCl3, ∼1:1 rotamers) 1.42 (br
s, 9 H), 1.80-1.95 (m, 1 H), 2.23-3.14 (m, 18 H), 3.39 (ddd,
J ) 11.30, 8.10, 6.03 Hz, 1 H), 3.52-3.78 (m, 1 H), 3.84-4.18
(m, 0.5 H), 5.82 (d, J ) 7.35 Hz, 0.5 H), 6.84-7.10 (m, 3 H),
7.20-7.41 (m, 1 H), 8.96 (s, 1 H). MS m/z 498.2 [M + H]+.
{5-[(S)-3-Amino-3-(3-fluorophenyl)-propyl]hexahydropy-
rrolo[3,4-c]pyrrol-2-yl}-(4,6-dimethylpyrimidin-5-yl)metha-
none Tartaric Acid Salt (2 ·Tartaric Acid). A solution of 23
in toluene (∼898 g, 1.81 mol, 89% pure by HPLC area) was
filtered to a 40-L glass-lined reactor through Solka Floc (300
g) to remove trace insoluble impurities. The solvent level was
reduced to 2.7 L by vacuum distillation, followed by the addition
of 3 N HCl (1.8 kg) at 25 °C. The reaction mixture was then
heated to 35 °C and stirred. Upon completion (HPLC), the
mixture was cooled, and the layers were separated. The aqueous
layer was stirred with toluene (1 kg) for 5 min and separated.
The aqueous layer was basified with 50% aqueous NaOH
solution (800 g) and extracted with Me-THF (2 × 1.1 kg). The
Me-THF layer was filtered through Solka Floc (300 g) to
remove insoluble impurities. Me-THF was replaced with EtOAc
(12.5 kg) by atmospheric pressure distillation until less than
1% Me-THF remained by GC analysis. L-Tartaric acid (272 g,
1.81 mol), as a solution of EtOH (2.86 kg) was slowly added
with rapid stirring. Once addition was complete, the solution
was cooled to 25 °C and filtered. The cake was immediately
rinsed with MTBE (2.7 kg) and dried in a vacuum oven at 50
°C, 30-50 Torr to constant weight, to give 2 as a tartaric acid
salt (810 g, 76% yield over five steps, 96% purity by HPLC
area). Mp 180-186 °C (amorphous solid). 1H NMR (300 MHz,
D2O) 2.37 (s, 3 H), 2.38 (s, 3 H), 2.39-2.58 (m, 3 H), 2.88
(td, J ) 11.59, 4.33 Hz, 1 H), 3.12-3.81 (m, 8 H), 3.88-4.04
(m, 1 H), 4.34-4.48 (m, 5 H), 7.10-7.32 (m, 3 H), 7.37-7.59
1
silica gel and characterized. H NMR (300 MHz, CDCl3)
2.36-2.68 (m, 4 H), 2.43 (s, 3 H), 2.48 (s, 3 H), 2.75-3.03
(m, 3 H), 3.35 (dd, J ) 11.11, 8.10 Hz, 1 H), 3.50-3.65 (m,
2 H), 3.70-3.91 (m, 2 H), 7.13-7.43 (m, 5 H), 8.94 (s, 1 H).
MS m/z 337.2 [M + H]+.
4,6-Dimethylpyrimidin-5-yl)(hexahydropyrrolo[3,4-c]pyr-
rol-2(1H)-yl)methanone Benzylmalonic Acid Salt (5 ·BMA).
A nitrogen-inerted 1-L high-pressure reactor was charged with
22 (100 g, 0.298 mol), benzylmalonic acid (69 g, 0.357 mol),
10% Pd/C (10 g), and MeOH (0.5 L). The mixture was purged
with nitrogen, heated to 60 °C, and pressurized to 100 psig with
hydrogen. Upon reaction completion by HPLC the reaction
mixture was cooled to ambient temperature, filtered through
Solka Floc (40 g) and rinsed with MeOH (0.25 L). MeOH was
replaced with IPA (0.25 L) by atmospheric-pressure distillation
and the mixture aged at 25 °C for 18 h. The crystals were filtered
to provide 5 as the benzylmalonic acid salt (97 g, 73% yield,
100% purity by HPLC area). An analytical pure sample of free
amine 5 was obtained by column chromatography over silica
gel and characterized. 1H NMR (300 MHz, CDCl3) 2.07 (s, 1
H), 2.45 (s, 3 H), 2.48 (s, 3 H), 2.66 (dd, J ) 10.93, 4.14 Hz,
1 H), 2.76-3.01 (m, 4 H), 3.12 (dd, J ) 10.74, 6.97 Hz, 1 H),
3.21 (dd, J ) 10.74, 6.97 Hz, 1 H), 3.39 (dd, J ) 11.30, 7.91
Hz, 1 H), 3.60-3.74 (m, 1 H), 3.83-3.99 (m, 1 H), 8.95 (s, 1
H). 1H NMR (BMA salt, 300 MHz, DMSO-d6) 2.35 (s, 3 H),
2.38 (s, 3 H), 2.81-3.89 (m, 13 H, 3 H from BMA), 6.89-7.37
(m, 5 H, all from BMA), 8.94 (s, 1 H). MS m/z 247.1 [M +
H]+.
[(S)-3-[5-(4,6-Dimethylpyrimidine-5-carbonyl)hexahydro-
pyrrolo[3,4-c]pyrrol-2-yl]-1-(3-fluorophenyl)propyl]carbam-
ic Acid tert-Butyl Ester (23). Sodium triacetoxyborohydride
(0.59 kg, 2.8 mol) was added to a stirred and cooled (0 °C)
slurry of 5·BMA salt (0.82 kg, 1.86 mol) in dichloromethane
(5.6 L) in a 40-L glass-lined reactor, followed by addition of
the crude 4 (∼1.86 mol) solution from the previous step over
30 min (<10 °C). After 2 h at ambient temperature, the mixture
was warmed to 60 °C and stirred for another 2 h. The reaction
mixture was cooled to ambient temperature and quenched with
saturated aqueous NaHCO3 (15 kg) over 20 min. The mixture
was purged with a nitrogen stream overnight, using a bleach
trap, to remove most of the dimethylsulfide. The mixture was
extracted twice with toluene (6 and 2 kg). The combined organic
phase was washed with saturated aqueous NaHCO3 (10 kg)
1
(m, 1 H), 8.86 (s, 1 H). H NMR (free amine 2, 300 MHz,
CDCl3) 1.70-2.07 (m, 2 H), 2.23-2.66 (m, 13 H), 2.70-3.07
(m, 3 H), 3.37 (dd, J ) 11.30, 8.29 Hz, 1 H), 3.63-4.08 (m,
4 H), 6.82-7.13 (m, 3 H), 7.21-7.38 (m, 1 H), 8.95 (s, 1 H).
MS m/z 398.2 [M + H]+.
(R)-Tetrahydrofuran-3-carboxylic acid [(S)-3-[5-(4,6-
dimethylpyrimidine-5-carbonyl)hexahydropyrrolo[3,4-c]pyr-
rol-2-yl]-1-(3-fluorophenyl)propyl]amide Hydrochloride Salt
(1 ·HCl).15 Oxalyl chloride (46.0 g, 36.2 mmol) was added over
1 h to a solution of (R)-tetrahydrofuran-3-carboxylic acid 316
(40.1 g, 34.5 mmol) in toluene (310 mL) containing DMF (0.5
mL) with stirring, while maintaining the temperature at 10 °C
with an ice bath. After 40 min the ice bath was removed and
the solution allowed to warm to ambient temperature and stir
overnight. The light-yellow solution was partially evaporated
at 40 °C, gradually reducing the pressure to 30-35 Torr. The
precipitated Vilsmeier reagent was removed by filtration, leaving
(15) The scale-up study was not conducted for this final process due to the
termination of the program.
(16) (R)-Tetrahydrofuran-3-carboxylic acid 3 of 98% ee was purchased from
Chirotech Technology Ltd. For preparation, see: Johnson, N. B.;
Lennon, I. C.; Moran, P. H.; Ramsden, J. A. Acc. Chem. Res. 2007,
40, 1291–1299.
598
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development

92.1 g of a gold-colored solution of acid chloride 24 (0.504 g
per g of solution) that was stored under nitrogen in a freezer
until used.
N HCl (2.42 mL, 29.0 mmol). The clear solution was seeded
with authentic product and stirred for 3.5 h, after which the
resulting slurry was filtered and washed with ice-cold acetone
(12 mL in two portions). Drying in a vacuum oven at 68 °C,
30-50 Torr gave 1·HCl as a dry, white powder (10.57 g, 86.1%
theory, 97.8% purity by HPLC area). Analytical data for 1·HCl
2·tartaric acid salt (14.54 g, 26.54 mmol, 95.8% pure by
HPLC area) was stirred with Me-THF (140 mL) and 3 N NaOH
(21.5 mL) until the solids dissolved. The organic phase was
washed with 3 N NaOH (14 mL) and clarified by filtration
through a medium porosity fritted glass filter. Solvent evapora-
tion at 40 °C gave a gold-colored oil. To the oil dissolved in
Me-THF (95 mL) was added triethylamine (8.06 g, 79.62
mmol), and the resulting solution was cooled in an ice bath.
Acid chloride 24 (9.93 g of the solution prepared above, 5.00 g
of acid chloride, 37.18 mmol) was added over 30 min. The ice
bath was removed and the mixture allowed to stir overnight.
The reaction mixture was quenched by the addition of 3 N
NaOH (100 mL) and stirring for 1 h. The aqueous phase was
extracted with Me-THF (20 mL), and the combined organic
phase was evaporated at 40 °C to give the free base of 1 (11.44
g, 23.08 mmol) as a slightly tacky dry foam.
1
salt: mp 149-150 °C. H NMR (300 MHz, D2O) 1.73-1.91
(m, 1 H), 1.98-2.26 (m, 3 H), 2.34 (s, 3 H), 2.37 (s, 3 H),
2.82-3.37 (m, 8 H), 3.40-4.10 (m, 9 H), 4.81 (t, J ) 7.54
Hz, 1 H), 6.93-7.13 (m, 3 H), 7.33 (td, J ) 7.82, 5.84 Hz, 1
H), 8.83 (s, 1 H). MS m/z 496.2 [M + H]+.
Acknowledgment
We thank David Rotstein, Drs. Fernando Padilla and Chris
Melville from the Medicinal Chemistry department for helpful
suggestions, and Frida Dobrouskin and Tim Lane for analytical
support.
Received for review January 26, 2010.
OP100020Z
A solution of the free base of 1 (11.44 g, 23.08 mmol) in
acetone (57.4 mL) and water (1.45 mL) was acidified with 12
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
599