Development of a robust process for the preparation of high-quality dicyclopropylamine hydrochloride

Article abstract of DOI:10.1021/op500031z

A short and efficient process for the preparation of high-quality dicyclopropylamine HCl salt is described. An oxygen-mediated Chan-Lam coupling of N-cyclopropyl 4-nitrobenzenesulfonamide with cyclopropylboronic acid was followed by an optimized p-nosyl deprotection with 1-decanethiol, providing the title compound in high chemical yield. This process addresses many of the challenges and liabilities inherent in previous synthetic approaches to this challenging molecule. The collection of key safety data enabled implementation of an oxygen-mediated process on-scale and ensured safe operation throughout development, optimization, and processing.

Full text of DOI:10.1021/op500031z

Article
pubs.acs.org/OPRD
Development of a Robust Process for the Preparation of High-
Quality Dicyclopropylamine Hydrochloride
Boguslaw Mudryk,* Bin Zheng,* Ke Chen, and Martin D. Eastgate
Chemical Development, Bristol-Myers Squibb Company, One Squibb Drive, New Brunswick, New Jersey 08903, United States
ABSTRACT: A short and efficient process for the preparation of high-quality dicyclopropylamine HCl salt is described. An
oxygen-mediated Chan−Lam coupling of N-cyclopropyl 4-nitrobenzenesulfonamide with cyclopropylboronic acid was followed
by an optimized p-nosyl deprotection with 1-decanethiol, providing the title compound in high chemical yield. This process
addresses many of the challenges and liabilities inherent in previous synthetic approaches to this challenging molecule. The
collection of key safety data enabled implementation of an oxygen-mediated process on-scale and ensured safe operation
throughout development, optimization, and processing.
(N-propyl- and N-ethyl-cyclopropylamines). Additional issues
with the use of sodium cyanoborohydride and allyl bromide are
well-known.
INTRODUCTION
■
The cyclopropylamino group, with its unique electronic and
steric properties, has become an important structural motif in
medicinal chemistry, as evidenced by its presence in numerous
drugs and drug candidates.1 Compared to cyclopropylamine,
dicyclopropylamine-containing compounds are relatively rare.2
This may be partially attributed to the difficulty of preparing
this system, which, despite its structural simplicity, presents a
number of synthetic challenges. For example, dicyclopropyl-
amine (DCPA) has unique electronic properties that impact
both its chemical compatibility and stability in addition to the
challenging bond disconnections required for installation of the
cyclopropyl rings.3 Our requirements for this compound came
with high-quality specifications and necessitated the preparation
of DCPA largely free of other secondary or primary amine
impurities, a significant complicating factor in developing an
approach to the bulk supply of this complex small molecule.
In 2011, we reported several synthetic approaches to DCPA.3
The most efficient process at that time was based on a reductive
amination/allyl deprotection strategy (Scheme 1). Although
Here, we describe the development of a new, robust, shorter,
and more efficient process for the preparation of dicyclopropyl-
amine hydrochloride. This approach is based on a Chan−Lam
coupling4 of N-cyclopropyl sulfonamide 2 with cyclopropylbor-
onic acid (CPBA) followed by a SNAr-type deprotection of the
p-nosyl (Ns) protecting group by a thiolate (Scheme 2).
Considering our previous experience with this system,3 we
realized that the selection of the proper protecting group would
be a critical success factor to ensure the delivery of high-quality
DCPA. We desired a protecting group that could be removed
under conditions in which the DCPA itself would remain
chemically stable. Our previous strategy (allylation) lacked the
appropriate deprotection practicality for large-scale synthesis,
and the allyl group was suboptimal in terms of its own stability.
Benzylation was not an option because of the known liabilities
associated with that approach.3 A critical area of concern was
the observed sensitivity of DCPA under strongly basic
conditions, resulting in degradation of DCPA to cyclopropyl-
amine.5 To obviate these issues, we envisioned that the Ns
group would provide the appropriate balance of chemical
compatibility and stability. Deprotection could be accomplished
through an aromatic nucleophilic substitution using a weakly
basic but highly nucleophilic thiolate anion, whereas the Ns
group would provide appropriate reactivity upstream.6 The
precursor to the Ns-protected DCPA could, in turn, be
prepared by a Chan−Lam coupling between the Ns-protected
cyclopropylamine and CPBA, leveraging conditions recently
described in the literature for other alkyl cyclopropyl amines.7
The Ns group would then be both the protecting and activating
group, enabling one cyclopropane ring to be derived from
readily available cyclopropylamine.8 It was apparent from
inception that this Chan−Lam approach would have several
technical and safety challenges in its development and
Scheme 1. Previous Synthetic Route to DCPA·HCl3
that route provided multikilogram quantities of the material, it
had several drawbacks. For example, one of the key reagents,
the cyclopropyl ketal, is expensive and difficult to source on a
large scale, the process involves palladium chemistry in two
steps, and the product is prone to contamination with several
difficult to purge impurities, such as other secondary amines
Received: January 27, 2014
Published: March 11, 2014
© 2014 American Chemical Society
520
dx.doi.org/10.1021/op500031z | Org. Process Res. Dev. 2014, 18, 520−527

Organic Process Research & Development
Article
Scheme 2. New Synthesis of DCPA·HCl
implementation; thus, a proactive approach to chemical safety
would be needed should proof of concept be achieved.
rather than performing the trials under a controlled
atmosphere. As the scale increased, the use of air appeared to
reduce reaction performance. An additional concern was the
formation of a gummy residue during the reactions in either
1,2-dichloroethane (DCE) or toluene, typical solvents reported
in the literature for this transformation.7a−c Thus, extensive
screening was initiated to identify conditions that would
address these challenges and result in a robust process that is
amendable for the preparation of sulfonamide 3 on scale.
We first screened various solvents to improve the efficiency
of oxygen mass transfer and to address the solubility issues,
using DCE and toluene as our initial reference for reactivity.
The results are summarized in Table 1.
RESULTS AND DISCUSSION
■
Nosylation of Cyclopropylamine. Preparation of sulfo-
namide 2 via nosylation of cyclopropylamine was straightfor-
ward and readily generated 2 as a highly crystalline solid in high
yield and quality. Our initial process to prepare 2 on a lab scale
employed i-Pr2NEt as the base and dichloromethane as the
solvent followed by an aqueous workup to remove the resulting
i-Pr2NEt·HCl salt. However, because of the limited solubility of
the sulfonamide in dichloromethane, addition of a secondary
solvent, such as 2-methyl THF, was required to solubilize the
sulfonamide during the acidic workup. After examining other
common solvents, the sulfonamide was found to have greater
solubility in ethyl acetate, allowing the use of a single solvent.
After a solvent exchange from ethyl acetate to n-heptane and
crystallization, sulfonamide 2 was produced in excellent yield
and quality (>99.5 AP, 95% yield).
a
b
Table 1. Conversions of Chan−Lam Reactions at 1 and 18
h in Various Solvents
entry
solvent
1 h (2/3)
18 h (2/3)
1
DCE
32/67
30/68
48/49
2/95
2/96
2
toluene
PhCF3
PhCl
3/93
A more efficient one-drop process was later developed and
adopted for larger scale preparation. The extractive workup and
the solvent exchange were eliminated by introducing toluene as
the reaction solvent with triethylamine as the base. In this
system, the product precipitated directly from the reaction
mixture. Addition of water to dissolve the triethylamine
hydrochloride byproduct produced a triphasic system. Filtration
of the triphasic system was rapid, and after washes with water
and toluene, the dried product was isolated with comparable
yield and quality. The use of triethylamine was preferred over i-
Pr2NEt to prevent contamination of the product with the
corresponding amine hydrochloride because triethylamine
hydrochloride easily purged to the aqueous phase during
isolation. This straightforward process was successfully
performed on up to a 110 kg scale without any issues,
providing product 2 in 90% yield and >99.5 AP purity.
Chan−Lam Coupling. The Chan−Lam coupling to
generate dicyclopropyl sulfonamide 3 plays a central role in
the preparation of DCPA and, as predicted, proved to be the
most challenging step to develop, optimize, and implement.
Following a recently reported procedure for the coupling of
simple alkyl toluenesulfonamides with boronic acids,7b our
initial results were encouraging, with >90% conversion to the
desired product being obtained after 3−5 h on the 50−100 mg
scale. However, when the reaction was carried out on 1 g scale,
the coupling became very sluggish and reached only 50% yield
after 24 h, presumably because of a less efficient mass transfer
of the oxidant. Oxygen is the stoichiometric oxidant in this
process; we therefore utilized air on a small scale (<100 mg)
3
12/83
0.2/94
47/35
0.6/99
0/96
4
5
DMF
82/6
6
MeCN
acetone
t-AmOH
THF
26/74
22/77
43/52
60/35
48/51
98/2
7
8
18/78
2/89
9
10
11
i-PrOAc
IPA
5/94
93/7
a
b
HPLC area percent of 2/3. Reactions were carried out on a 50 mg
scale with 1.0 mL of the solvent, 0.5 equiv of Cu(OAc)2, 0.5 equiv of
2,2′-bipyridyl, 2.0 equiv of CPBA, and 2.0 equiv of Na2CO3 at 55 °C
under an atmosphere of air.
As shown in Table 1, in addition to DCE and toluene,
couplings in chlorobenzene, MeCN, and acetone provided
excellent conversions. However, similar to our observations in
DCE and toluene, reactions in chlorobenzene produced
gummy reaction mixtures, and the conversions declined upon
larger-scale experimentation. In contrast, reactions in acetoni-
trile and acetone performed better, with more consistent
reaction profiles across various scales. Eventually, we selected
acetonitrile as the solvent for further development because it
provided the best balance of reaction rate and impurity profile.
After selecting the solvent, we turned our attention to
catalyst optimization. The common copper salts, such as Cu(II)
triflate or Cu(II) bromide, were not effective (<30%
conversion, Figure 1). Although the less soluble Cu(II) sulfate9
521
dx.doi.org/10.1021/op500031z | Org. Process Res. Dev. 2014, 18, 520−527

Organic Process Research & Development
Article
Consistent with previous reports,7a−d we found that a ligand/
metal ratio of 1:1 provided the best coupling rates. Although
stoichiometric amounts of copper have been generally
reported,11 0.5 equiv of the optimized catalyst system appeared
to be sufficient to reach completion, maintaining a similar
kinetic profile to that observed with 1 equiv of catalyst.
Reducing the catalyst loading to 0.35 equiv resulted in extended
reaction times (11 vs 8 h with 0.5 equiv on a 3 g input scale),
and the use of less catalyst led to incomplete conversion,
showing the limited number of catalyst turnovers available to
this system.
It is known that Chan−Lam couplings require an active
oxidant to drive the catalytic cycle.12 Although the most
common oxidant for this transformation is molecular oxygen,
which was utilized in our initial work, the use of other oxidants
in combination with air have been reported to reduce the
loading of the catalyst.11b To identify alternative oxidants, we
screened numerous reagents,13 including pyridine N-oxide,
TEMPO, and KMnO4; however, only MnO2 produced
promising results. Couplings with 2 equiv of MnO2 reached
>95% conversion in 15 h on the scale used in our screening (50
mg); however, the reaction stalled on the 0.5 g input scale,
reaching only ∼75% conversion after 24 h. Without a viable
alternate oxidant, the use of molecular oxygen seemed
inevitable. Thus, understanding the safety implications and
controls required for the use of molecular oxygen on scale
became critical. It should be noted, however, that despite
challenges associated with the use of oxygen on scale, it is the
greenest oxidant.
Our first goal was to understand the flammability of the
solvent system under the reaction conditions. No flammability
data on MeCN at elevated temperatures could be found in the
literature; thus, the limiting oxygen concentration (LOC) was
experimentally determined.14 It was found that a LOC of 11 to
12% oxygen in the headspace was required for acetonitrile at
either 45 or 55 °C to sustain a flame, which excluded the use of
air. However, this provided the possibility of operating below
this level using diluted oxygen in nitrogen, and we quickly
found that working at levels much lower than 11% were
sufficient to drive the reaction to completion. To ensure an
ample safety margin, we decided to investigate the use of
Figure 1. Conversions of the Chan−Lam reactions with copper salts.
Conversions are given as the HPLC area percent of 3 excluding the
peak of the ligand. Reactions were carried out on a 50 mg scale in 1.0
mL of acetonitrile with 0.5 equiv of the metal precursor, 0.5 equiv of
2,2′-bipyridyl, 2.0 equiv of CPBA, and 2.0 equiv of Na2CO3 at 55 °C
under an atmosphere of air.
and Cu(II) chloride gave relatively good yields, Cu(II) acetate
was found to be the most active copper source for this coupling.
In addition, the use of Cu(II) acetate of various purities (97−
99.5%) and particle sizes (powder to granular) showed
comparable reaction rates under the same reaction conditions,
contributing to the processes robustness on scale. Other metal-
catalyzed couplings, including the known Ni couplings of
boronic acids,10 were investigated without success.
We also performed an extensive screen of ligands. Four
common chemotypes known to coordinate copper were
analyzed: 2,2′-bipyridyls, 1,10-phenathrolines, 1,2-diamines,
and 1,3-diketones (Figure 2). Among them, both 2,2′-bipyridyls
and 1,10-phenathrolines showed better performance than either
1,2-diamines or 1,3-diketones. Because 2,2′-bipyridyl is readily
available in bulk quantities and is less expensive than 1,10-
phenanthroline, it was our choice for further development.
Figure 2. Conversions with various ligands (0.5 equiv) as the HPLC area percent of 3. Reactions were carried out in acetonitrile (20 v) with 0.5
equiv of Cu(OAc)2, 2.0 equiv of CPBA, and 2 equiv of Na2CO3 at 55 °C under an atmosphere of air for 15 h. 4-MeBipy, 4,4′-dimethyl-2,2′-
bipyridine; Phen, 1,10-phenathroline; 4MeO-Phen, 4,7-dimethoxy-1,10-phenathroline; 4Me2-Phen, 4,7-dimethyl-1,10-phenathroline; Phenone, 1,10-
phenathroline-5,6-dione; DMCyDA, cis-N1,N2-dimethylcyclohexane-1,2-diamine; tBu2-AA, 2,2,8,8-tetramethylnonane-4,6-dione; and Py2-AA, 1,5-
di(pyridin-2-yl)pentane-2,4-dione.
522
dx.doi.org/10.1021/op500031z | Org. Process Res. Dev. 2014, 18, 520−527

Organic Process Research & Development
Article
commercially available 5% oxygen in nitrogen in all future
experiments.
We also investigated the impact of oxygen mass transfer
(affected by oxygen concentration, gas flow rate, agitation rate,
etc.). Unsurprisingly, this aspect of the process was found to be
one of the most critical factors impacting the overall rate of
reaction; for example, a reaction fed with 1% oxygen reached
only 31% conversion after 8 h, whereas a coupling reaction
performed in the presence of 5% oxygen (under otherwise
identical conditions) went to completion. Similar observations
occurred during the initial scale-up runs, where an initial slow
coupling was significantly accelerated once more vigorous
agitation and a higher gas flow rate were applied.17
Interestingly, despite slow conversion, reactions starved of
oxygen (such as under of 1% O2 in N2) did continue to
produce product, indicating that the catalyst system remained
active over extended periods of time.
With a safe window of operation established,15 we
investigated the impact of base on the reaction profiles, as
shown in Figure 3. Sodium carbonate was the most effective
With the transformation itself nearly optimized, our focus
turned to development of the workup and isolation, especially
with respect to gaining control over the residual copper from
the catalyst, as we needed to ensure no Cu contamination in
downstream chemistry. In our optimized workup procedure,
dichloromethane and water were added to the reaction mixture,
and after the phase split, the resulting rich organic phase was
washed with 6 wt % aqueous ammonium hydroxide. This
extraction transferred the majority of the Cu salts into the
aqueous phase. A subsequent wash with aqueous 1 N HCl
(mainly to remove the ligand and excess ammonia) removed
additional Cu (Table 2). This workup has proved to be very
robust, both in the lab and on scale, and always resulted in
observed Cu levels in isolated 2 below 10 ppm (proven limit of
quantitation).
Figure 3. Reaction profiles of Chan−Lam reactions with various bases
(2 equiv). Reactions were carried out in acetonitrile (20 v) with 0.5
equiv of Cu(OAc)2, 0.5 equiv of 2,2′-bipyridyl, and 2.0 equiv of CPBA
at 55 °C under 5% oxygen in nitrogen.
base, whereas potassium carbonate or phosphate showed lower
reactivity. The reactions without base or with organic bases
(pyridine, TEA) were much slower or did not reach
completion.
The Chan−Lam coupling was found to be sensitive to
temperature, with very slow reaction rates being observed at
ambient temperature (less than 1% product observed after 15
h). At 45 °C, the reaction proceeded at an appreciable rate and
reached completion in 15−20 h. Couplings at 55 °C were more
rapid (ca. 5−8 h) but resulted in accelerated and competitive
decomposition of CPBA.16 This side reaction appeared more
evident when the impact of a charging protocol was assessed
both in terms of rate and conversion (Figure 4). Although an
upfront charge of CPBA resulted in stalling, portionwise
addition enabled a reduction in overall stoichiometry of CPBA
(from 2 to 1.5 equiv) while simultaneously improving the
reaction rate (Figure 4). The latter charging CPBA protocol
was therefore adopted for further development.
Table 2. Residual Cu Levels in Various Process Streams
stage of rich DCM
after initial
phase split
post aqueous
NH4OH wash
post 1 N HCl
wash
a
stream
Cu, ppm
1000
50
<10
a
Determined by concentrating a sample of the stream to dryness and
analyzing the resulting solid residue.
After the aqueous workup, product isolation involved a
standard sequence of solvent exchange (from dichloromethane
to isopropanol) followed by crystallization and filtration. This
protocol proved robust on scale and consistently provided
highly crystalline product.
With an optimized process in place, the Chan−Lam step was
successfully demonstrated on scales ranging between 60 and 80
kg, producing high yields (85−90%) and excellent quality
(>99.5 AP) of sulfonamide 3.
Deprotection to DCPA. We next investigated the Ns
group deprotection, which is usually achieved by simple thiol-
based nucleophiles, such as thiophenol.6 Following several
literature procedures, our initial efforts to remove the Ns group
used a variety of thiols and provided unsatisfactory results
(Table 4). This was largely due to poor selectivity in the
competition between the displacement of the sulfonyl (pathway
a) versus the nitro group (pathway b) by the thiol (Scheme 3),
leading to 6 (desired) or 7 (undesired). The results are
summarized in Table 3.
Overall, 1-decanethiol displayed the best impurity profile and
was also preferred because of its reduced odor compared to
other common thiols (with the exception of thiosalicylic
acid).18 Thus, further optimization of the reaction conditions
focused on 1-decanethiol.
Figure 4. Reaction profiles of Chan−Lam reactions using two CPBA
charge protocols (consumption of 2 vs time in min). Red, 1.5 equiv
CPBA charged upfront; blue, 1.5 equiv CPBA charged in 5 × 0.3 equiv
portions every 1.5 h.
523
dx.doi.org/10.1021/op500031z | Org. Process Res. Dev. 2014, 18, 520−527

Organic Process Research & Development
Article
Scheme 3. SNAr Deprotection of the Nosyl Group
Table 5. Deprotections with 1-Decanethiol and t-AmOK in
Various Solvents
a
b
c
solvent
reaction time (h)
LCAP (3:6:7)
solution yield (%)
d
toluene
toluene
THF
1
7
1
1
1
62:34:1.2
0.6:89:3.3
0.5:85:1.6
48:31:10
57:25:13
NA
84
79
d
acetone
MeCN
NA
d
NA
a
Reactions were carried out on a 0.1 g scale with 2 equiv of 1-
decanethiol and 1.5 equiv t-AmOK (1.7 M in toluene). The base was
added to the mixture of the substrate and thiol in the solvent (1.0 mL)
b
at ambient temperature. HPLC area percent of 3:6:7 excluding the
c
peak of toluene. Quantified by derivatizing the resulting DCPA in an
Table 3. Results of Initial Screening with Various Thiols and
Bases
a
aliquot sample with excess TsCl and K3PO4 and assaying the amount
of the corresponding sulfonamide against a reference standard (by
d
b
HPLC analysis). Not assayed because of the poor reaction profiles.
thiol
PhSH
base/solvent
LCAP (3:6:7)
3 equiv K2CO3/DMF
3 equiv K2CO3/DMF
3 equiv K2CO3/DMF
3 equiv K2CO3/DMF
3 equiv K2CO3/DMF
1.5 equiv t-BuOK/NMP
2.3 equiv t-BuOK/NMP
1.5 equiv t-BuOK/NMP
0:44:32
55:0:0
DCPA degrades in the presence of a strong base, converting to
cyclopropylamine5 (our main impurity of concern), it is
important to ensure the presence of an excess of the thiol
over t-AmOK at all times. Our optimized procedure involved
addition of 1.30 equiv of potassium tert-amylate into the
solution of the nosyl-protected DCPA 3 and 1.35 equiv of the
thiol in the mixed THF and toluene solvent system (1:1, 2.5
mL/g input each) at 5−10 °C. Deprotection was typically
complete in 1 to 2 h after the base was added and gave >100:1
selectivity of 6 to 7 and >90% yield of DCPA in solution.
We also found that the quality of t-AmOK was critical for the
desired reaction rate and impurity profile. The presence of
solids, typically KOH in the t-AmOK solution, resulted in
reaction stalling and competitive reduction of the nitro group in
3 to azoxy byproduct 8 (Scheme 4).19 A selective titration
method for t-AmOK and KOH was therefore used20 along with
a visual inspection of the base solution prior to use to ensure
acceptable base quality.
HSCH2CO2H
p-MeOPhSH
o-HSPhCO2H
1-C10H21SH
1-C10H21SH
o-HSPhCO2H
p-MeOPhSH
5:26:37
79:0:0
28:40:19
0:36:41
68:0:0
0:27:48
a
Reactions were carried out on a 50 mg input scale with 2.0 equiv of
b
thiols in 1.0 mL of solvent at ambient temperature for 15 h. HPLC
area percent of 3:6:7 excluding the peak of thiol reagents.
With a pKa value of ca. 10, 1-decanethiol can be
deprotonated with mild bases, such as triethylamine or
DIPEA. However, the use of organic bases was not desirable
because of the complication in separating them from DCPA
product. Our next screening with NaOMe in various solvents
did not show promising results (Table 4). Although improved
selectivity was observed, the deprotections were relatively slow,
and the reaction mixtures contained a range of impurities.
Scheme 4. Reduction of 3 to 8
Table 4. Results of the Deprotections with 1-Decanethiol
a
and NaOMe
b
solvent
LCAP (7:3:6)
THF
41:17:0.9
51:26:1.2
51:4.8:0
PhCF3
DCM
In the isolation of the final product, an acid−base extraction
was developed. In this scenario, it was envisaged that all of the
nonamine organics would purge to the organic phase, whereas
DCPA would transfer to the aqueous phase as its hydrochloride
salt. To enable this approach, the stability of DCPA in aqueous
acid was quickly determined at ambient temperature and shown
to be acceptable. This, however, was not the case once the
DCPA salt was liberated and transferred to the organic phase
(toluene) as its free base. Because of the sensitivity of DCPA to
basic conditions, we developed a salt-liberation protocol with
gradual pH adjustment that was accomplished by a two-stage
addition of solid dibasic potassium phosphate followed by
tribasic phosphate to bring the final pH gradually into the 9−11
range. To provide additional control, the temperature during all
quenches and operations during the workup was kept at or
below ambient, as decomposition was also shown to be
temperature-dependent. At ambient temperature, the stability
of the resulting DCPA toluene solution was limited; however,
reducing the temperature at which the DCPA solution in
toluene was held dramatically enhanced the stability.
toluene
MeOH
25:59:0.7
61:8.7:0.2
a
Reactions were carried out on a 50 mg input scale with 2.0 equiv of
thiols and 1.5 equiv of 25% NaOMe in MeOH at ambient temperature
in a solvent (10 mL/g input) for 8 h. HPLC area percent of 3:6:7.
b
A stronger and sterically more hindered base (potassium tert-
amylate, t-AmOK) in combination with 1-decanethiol provided
a more efficient deprotection with a vastly improved impurity
profile and excellent selectivity for the desired substitution
pathway (a). This improvement resulted in a high yield of
DCPA, reaching 84 and 79% in toluene and THF, respectively
(Table 5).
With deprotection conditions in place, further optimization
was required to improve recovery and isolation of high-quality
DCPA. A minimal volume of THF was desired to facilitate the
phase splits and to reduce the loss of DCPA to the waste
stream in the subsequent aqueous workup (vide infra). Because
524
dx.doi.org/10.1021/op500031z | Org. Process Res. Dev. 2014, 18, 520−527

Organic Process Research & Development
Article
Final isolation was accomplished by precipitation of the
hydrochloride salt from toluene. During our initial development
studies, this was accomplished by the addition of a 5 to 6 N
solution of HCl in isopropanol followed by distillation to
remove the alcohol and maximize the recovery of the resulting
DCPA·HCl salt. This approach, although providing good yields
(ca. 76−85%) and quality on a lab scale, was not robust enough
upon scale-up. Again, because of the thermal instability of
DCPA·HCl, an elevated level of the cyclopropylamine impurity
(3.1 AP) was observed after prolonged distillation. In addition,
commercial solutions of 5 to 6 N HCl in IPA typically contain
1−3 wt % water, which adversely impacts both the DCPA
recovery and stability.
The above liabilities were addressed by eliminating the
distillation and introducing an in situ generation of anhydrous
HCl from TMSCl and isopropanol. Thus, a small amount of
isopropanol (∼7 vol %) was added to the toluene solution prior
to addition of TMSCl at 10−15 °C. The product then
precipitated as a white crystalline solid, and after isolation,
excellent yields (80−89%) and quality of DCPA hydrochloride
were obtained (<0.15 AP of cyclopropylamine as the single
observable impurity). Although the level of cyclopropylamine
was generally extremely well controlled, we also developed an
optional reslurry in acetone, which provided additional purging
of this impurity.
phase A: 95% water/5% acetonitrile with 0.05% trifluoroacetic
acid. Mobile phase B: 5% water/95% acetonitrile with 0.05%
trifluoroacetic acid. Dicyclopropylamine and related impurities
were analyzed by the HPLC as their tosylated derivatives.
N-Cyclopropyl 4-Nitrobenzenesulfonamide 2.21 To a
solution of 4-nitrobenzenesulfonyl chloride (100 g) in toluene
(500 mL, 5 vol) was added cyclopropylamine (38 mL, 1.2
equiv) at 0−15 °C over 15 min. Triethylamine (76 mL, 1.2
equiv) was added at 5−15 °C, and the thick slurry was agitated
at 20−25 °C for 30−60 min. A sample was taken to confirm
>99% conversion by HPLC. Aqueous 1 N HCl (400 mL) was
then added at 20−25 °C, and the triphasic mixture was agitated
at 20−25 °C for 4 h. The solids were filtered, washed
sequentially with water (300 mL) and toluene (300 mL), and
dried under vacuum at 40−50 °C to give 98.3 g (90% yield) of
off-white crystalline 2 with 99.7 AP purity by HPLC.
N,N-Dicyclopropyl 4-Nitrobenzenesulfonamide 3.
Acetonitrile (700 mL, 7 vol), copper(II) acetate (37.6 g, 0.5
equiv), 2,2′-bipyridyl (32.3 g; 0.5 equiv), sodium carbonate
(87.6 g, 2.0 equiv), 2 (100 g), and cyclopropylboronic acid
(10.6 g, 0.3 equiv) were sequentially charged at rt under an
atmosphere of 5% O2 in N2. The resulting mixture was heated
at 55 °C and vigorously agitated for 3 h while bubbling
subsurface 5% O2 in N2 through the mixture. Four additional
portions of CPBA (10.6 g, 0.3 equiv each) were charged every 3
to 4 h, and the progress of coupling was monitored by HPLC.
Desired >98% conversion was typically reached after 16−24 h.
After cooling to rt, 6 wt % aqueous ammonium hydroxide
(1.7 L) and dichloromethane (1.5 L) were added, and the
biphasic mixture was agitated for a 30 min. The bottom organic
(brown) and the top aqueous (blue) phases were separated,
and the organic layer was washed with 1 N aqueous HCl (1.5
L) and once with water (1 × 800 mL). The organic layer was
then subjected to distillation at atmospheric pressure to bring
the final volume to ∼500 mL. Isopropanol (500 mL) was
added, and the product partially precipitated. The resulting
slurry was again concentrated to ∼500 mL, and another portion
of isopropanol (500 mL) was added. The slurry was reheated to
78 °C and then cooled to 20 °C over 1 h. After agitation for 1 h
at 20−25 °C, the solids were filtered, washed with isopropanol
(∼400 mL), and dried under vacuum at 40−50 °C to give 105
g (90% yield) of 3 as a light yellow crystalline solid with 100 AP
purity by HPLC.
This deprotection process has been successfully demon-
strated on varying scales up to a 30 kg output (Table 6).
a
Table 6. Scale-Up Data for the Ns Group Deprotection of 3
b
batch no.
5 (kg)
yield (%)
LCAP
1
2
3
4
5
6
22
21
22
21
27
30
79
81
77
78
75
79
99.9
99.9
c
99.9
99.9
99.9
c
100.0
a
b
Combined scale-up summary from two different vendors. Quantified
by derivatization of the DCPA by treatment with excess TsCl and
K3PO4 and assayed as the corresponding sulfonamide against the
c
reference standard by HPLC analysis. The acetone reslurry was
implemented.
1
Data for compound 3: mp 168−169 °C. H NMR (500
CONCLUSIONS
MHz, CDCl3): 0.76 (m, 4H), 0.95 (m, 4H), 1.98 (m, 2H), 8.11
(d, J = 9.0 Hz, 2H), 8.41 (d, J = 9.0 Hz, 2H). 13C NMR
(CDCl3): 6.5, 32.4, 124.0, 129.4, 142.3, 150.2. HRMS [M +
H]+ calcd for C12H15N2O4S, 283.0748; found, 283.0753.
N,N-Dicyclopropylamine Hydrochloride 5.3 THF (310
mL), toluene (310 mL), 3 (100 g), and 1-decanethiol (98 mL,
1.35 equiv) were charged under nitrogen, and the mixture was
cooled to 0−5 °C. Potassium tert-amylate solution in toluene
(25 wt %, 229 g, 1.30 equiv) was added dropwise at 5−10 °C.
The mixture was agitated at 5−10 °C for 90 min and sampled
to confirm >99% conversion by the HPLC derivatization
method. Aqueous 0.5 N HCl (940 mL) and toluene (190 mL)
were added to the reaction mixture at 10−15 °C, partitioning 5
into the aqueous phase. The biphasic mixture was stirred for 10
min, and the phases were split. The rich aqueous layer was
washed with toluene (500 mL) and cooled to ca. 10 °C.
Toluene (500 mL), solid K2HPO4 (156 g, pH 7.0 at this point),
and solid K3PO4 (156 g, pH 9.9) were sequentially added at
10−20 °C, partitioning DCPA free-base 4 into the organic
■
We have developed a highly efficient and robust process for
large-scale production of dicyclopropylamine hydrochloride in
excellent quality, which addressed liabilities of previous
synthetic approaches. In the key oxygen-mediated Chan−Lam
coupling, the critical LOC safety data was collected, which
drove development and optimization of this step in a
fundamental way.
EXPERIMENTAL SECTION
■
General. Sulfonylation and nosyl deprotection were
performed under nitrogen, and commercial 5% oxygen in
nitrogen was used for the Chan−Lam couplings. Reported
yields are for isolated materials or calculated solution yields and
are corrected for potency. All transformations were monitored
by HPLC at λ = 220 nm using a Zorbax Eclipse Plus C18 1.8
μm 4.6 × 50 mm column at a 1.2 mL/min flow and a 10 μL
injection volume using the following gradient method: 15% B
(0 min) to 100% B (7 min) and then 100% (15 min). Mobile
525
dx.doi.org/10.1021/op500031z | Org. Process Res. Dev. 2014, 18, 520−527

Organic Process Research & Development
Article
phase. After the phase split, isopropanol (36 mL) was charged
to the organic layer, and TMSCl (46 mL, 1 equiv) was added at
15−20 °C within 15 min. The resulting white slurry was
agitated for 30−60 min at 20−25 °C. The product was filtered,
washed with toluene (200 mL), and dried under vacuum at
20−25 °C to give 40.7 g (86% yield) of 5 as a white crystalline
solid with 99.9 AP HPLC purity.
Optional Reslurry of 5 in Acetone. The slurry of 5 (40 g)
in acetone (160 mL, 4 vol) was agitated for at 1 to 2 h at 20−25
°C. The solids were filtered, washed with acetone (80 mL), and
dried under vacuum at 20−25 °C to give 38.4 g (96% recovery)
of 5 with upgraded purity.
HPLC Derivatization Method for 5. Two percent (w/v)
p-toluenesulfonyl chloride (TsCl) in acetonitrile was added into
a 50 mL volumetric flask followed by ca. 10 mg of solid 5 or an
equivalent amount of 4 or 5 in solution and 10 mL of a 2% (w/
v) potassium phosphate (K3PO4) solution in water. The flask
was filled with 80 vol % acetonitrile/20 vol % water diluent, and
the solution was shaken well for a few seconds. After at least 30
min aging at ambient temperature, a sample was taken for UV
HPLC analysis.
A similar radical bond scission has been reported for DCPA,
see: Maeda, Y.; Ingold, K. U. J. Am. Chem. Soc. 1980, 102, 328−
331.
(6) (a) Fukuyama, T.; Jow, C.-K.; Cheung, M. Tetrahedron Lett.
1995, 36, 6373−6374. (b) Csiki, Z.; Fugedi, P. Tetrahedron Lett. 2010,
51, 391−395. Also, see: (c) Wuts, P. G. M.; Greene, T. W. Greene’s
Protective Groups in Organic Synthesis, 4th ed.; John Wiley and Sons:
Hoboken, NJ, 2007; p 860 and references therein.
(7) For the Chan−Lam coupling with CPBA, see: (a) Tsuritani, T.;
Strotman, N. A.; Yamamoto, Y.; Kawasaki, M.; Yasuda, N.; Mase, T.
Org. Lett. 2008, 10, 1653−1655. (b) Benard, S.; Neuville, L.; Zhu, J. J.
Org. Chem. 2008, 73, 6441−6444. (c) Benard, S.; Neuville, L.; Zhu, J.
Chem. Commun. 2010, 46, 3393−3395. (d) Racine, E.; Monnier, F.;
Vors, J.-P.; Tailefer, M. Chem. Commun. 2013, 49, 7412.
(8) Other N-protected dicyclopropylamines, such as Boc-DCPA and
CF3CO-DCPA, although viable for facile deprotection, were found not
to be efficient in the Chan−Lam couplings with CPBA compared to
Ns-DCPA. We also found that the Chan−Lam coupling of p-nosyl
cyclopropylamine with CPBA provided higher yields and cleaner
impurity profiles than that of the corresponding o-nosyl derivative.
(9) Cu(II) sulfate has been used in Chan−Lam couplings, see: Xu,
H.-J.; Zhao, Y.-Q.; Feng, T.; Feng, Y.-S. J. Org. Chem. 2012, 77, 2878−
2884.
(10) Raghuvanshi, D. S.; Gupta, A. K.; Singh, K. N. Org. Lett. 2012,
14, 4326−4329.
AUTHOR INFORMATION
■
(11) For examples of catalytic Chan−Lam reactions, see: (a) Coll-
man, J. P; Zhong, M. Org. Lett. 2000, 2, 1233−1236. (b) Lam, P. Y. S.;
Vincent, G.; Clark, C. G.; Deudon, S.; Jadhav, P. K. Tetrahedron Lett.
2001, 42, 3415−3418. (c) Collman, J. P.; Zhong, M.; Zeng, L.;
Costanzo, S. J. Org. Chem. 2001, 66, 1528−1531. (d) Antilla, J. C;
Buchwald, S. L. Org. Lett. 2001, 3, 2077−2079. (e) Quach, T. D.;
Batey, R. A. Org. Lett. 2003, 5, 4397−4400. (f) van Berkel, S. S.; Van
den Hoogenband, A.; Terpstra, J. W.; Tromp, M.; Van Leeuwen, P. W.
N. M.; Van Strijdonck, G. P. F. Tetrahedron Lett. 2004, 45, 7659−
7662. (g) Lan, J.-B.; Chen, L.; Yu, X.-Q.; You, J.-S.; Xie, R.-G. Chem.
Commun. 2004, 188−189.
Corresponding Authors
*E-mail: boguslaw.mudryk@bms.com (B.M.).
*E-mail: bin.zheng@bms.com (B.Z.).
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Drs. David Kronenthal, Rajendra Deshpande, and
David Conlon for supporting this work. We also thank Drs. Yi
Xiao, Jacob Janey, Dimitri Skliar, and Paul Lobben for technical
contributions, Dr. Simon Leung and Alan Fritz for safety
evaluation, and Drs. Catherine Gatzemeyer, Cesar Florez, Liya
Tang, Lydia Breckenridge, Frank Rinaldi, and Jonathan
Marshall for analytical support.
(12) (a) King, A. E.; Brunold, T. C.; Stahl, S. S. J. Am. Chem. Soc.
2009, 131, 5044−45. (b) King, A. E.; Huffman, L. M.; Casitas, A.;
Costas, M.; Ribas, X.; Stahl, S. S. J. Am. Chem. Soc. 2010, 132, 12068−
12073.
(13) The list of oxidants also included NMO, NaBO3, NBS, m-
CPBA, SO3 pyridine, NaBrO3, and oxone tetrabutylammonium salt.
(14) The safety evaluations were conducted by Chilworth
Technology, Inc.
(15) We obtained DSC for all intermediates and raw materials. The
entire three-step process was deemed to be safe on the basis of our
internal hazard safety evaluation.
(16) Use of more stable CPBA MIDA ester resulted in no reaction
under the standard process conditions, see: Knapp, D. M.; Gillis, E. P.;
Burke, M. D. J. Am. Chem. Soc. 2009, 131, 6961−63.
(17) We briefly evaluated this Chan−Lam coupling under pressure
(150 psig) and found a significantly higher conversion rate; however,
the procedure had not been optimized prior to the scale-up.
(18) For use of other odorless thiols in deprotection of nosylamines,
see: Matoba, M.; Kajimoto, T.; Node, M. Synth. Commun. 2008, 38,
1194−1200.
(19) Alkaline reduction of nitroarenes to the azoxy compounds have
long been known, see: Lachman, A. J. Am. Chem. Soc. 1902, 24, 1178−
1200. For recent examples, see: (a) Prato, M.; Quintily, U.; Scapol, L.;
Scorrano, G. Bull. Soc. Chim. Fr. 1987, 1, 99−102. (b) Liu, Y.; Liu, B.;
Guo, A.; Dong, Z.; Jin, S.; Lu, Y. Synth. Commun. 2012, 42, 2201−
2206.
(20) This method, developed by our analytical colleagues, comprises
two analyses: (a) total base titration (alkoxide, hydroxide, and
carbonate) using a standardized aqeuous HCl titration and (b)
selective titration for combined hydroxide and carbonate, which
utilizes neutralization in anhydrous methanolic solution of glacial
acetic acid followed by titration with Karl Fisher reagent of the water
resulting from hydroxide and carbonate acid−base reaction.
REFERENCES
■
(1) For the reviews, see: (a) Salaun, J. Top. Curr. Chem. 2000, 207,
1−67. (b) Wessjohann, L. A.; Brandt, W.; Thiemann, T. Chem. Rev.
2003, 103, 1625−1647. (c) Reichelt, A.; Martin, S. F. Acc. Chem. Res.
2006, 39, 433−442.
(2) For selected examples, see: (a) Holy, A.; Votruba, I.; Tloustova,
E.; Masojidkova, M. Collect. Czech. Chem. Commun. 2001, 66, 1545−
1592. (b) Plettenburg, O.; Hofmeister, A.; Goerlitzer, J.; Loehn, M.
Patent WO 2008077553, 2008. (c) Purandare, A. V.; Grebinski, J. W.;
Hart, A.; Inghrim, J.; Schroeder, G.; Wan, H. Patent WO 201128864,
2011. (d) Chen, W.; Loury, D. J.; Mody, T. D.; Wang, L. Patent WO
201310136, 2013.
(3) Chen, K.; Eastgate, M. D.; Zheng, B.; Li, J. Org. Process Res. Dev.
2011, 15, 886−892 and references therein.
(4) For recent reviews, see: (e) Qiao, J. X.; Lam., P. Y. S. Synthesis
2011, 829−856. (f) Rao, K. S.; Wu, T.-S. Tetrahedron 2012, 68, 7735−
7754 and references therein.
(5) Because of the increased sp2 character of the cyclopropyl groups
and increased acidity of N-H, we postulate that the following anionic
degradation pathway may be occurring:
526
dx.doi.org/10.1021/op500031z | Org. Process Res. Dev. 2014, 18, 520−527

Organic Process Research & Development
Article
(21) Engers, D. W.; Gentry, P. R.; Wiliams, R.; Bolinger, J. D.;
Weaver, D.; Menon, U. N.; Conn, P. J.; Lindsley, C. W.; Niswender, C.
M.; Hopkins, C. R. Bioorg. Med. Chem. Lett. 2010, 20, 5175−78.
527
dx.doi.org/10.1021/op500031z | Org. Process Res. Dev. 2014, 18, 520−527