Convergent, kilogram scale synthesis of an Akt kinase inhibitor

Article abstract of DOI:10.1021/op300031r

The development of a convergent, chromatography-free synthesis of an allosteric Akt kinase inhibitor is described. The route comprised 17 total steps and was used to produce kilogram quantities of the target molecule. A key early transformation, for which

Full text of DOI:10.1021/op300031r

Article
pubs.acs.org/OPRD
Convergent, Kilogram Scale Synthesis of an Akt Kinase Inhibitor
Pintipa Grongsaard,*,† Paul G. Bulger,† Debra J. Wallace,† Lushi Tan,† Qinghao Chen,† Sarah J. Dolman,†
Jason Nyrop,‡ R. Scott Hoerrner,† Mark Weisel,† Juan Arredondo,† Takahiro Itoh,† Chengfu Xie,§
Xianghui Wen,§ Dalian Zhao,† Daniel J. Muzzio,‡ Ephraim M. Bassan,‡ and C. Scott Shultz†
‡
†Departments of Process Chemistry and Chemical Process Development and Commercialization, Merck & Co., Inc., 126 East
Lincoln Avenue, Rahway, New Jersey 07065, United States
§WuXi AppTec Co., Ltd., No. 1 Building, #288 FuTe ZhongLu, WaiGaoQiao Free Trade Zone, Shanghai 200131, China
S
* Supporting Information
ABSTRACT: The development of a convergent, chromatography-free synthesis of an allosteric Akt kinase inhibitor is described.
The route comprised 17 total steps and was used to produce kilogram quantities of the target molecule. A key early
transformation, for which both batch and flow protocols were developed, was formylation of a dianion derived by deprotonation
and subsequent lithium-halogen exchange from a 2-bromo-3-aminopyridine precursor. Improved reaction yield and practicality
were achieved in the continuous processing mode. Further significant process developments included the safe execution of a high
temperature and pressure hydrazine displacement, separation of substituted cyclobutane diastereomers by means of
chemoselective ester hydrolysis, and a late-stage Suzuki fragment coupling under mild conditions.
safe and scalable synthetic route that enabled the production of
the required quantity of the target molecule.
INTRODUCTION
■
Cancer remains one of the leading causes of death in both
developed and developing countries, with the annual worldwide
death toll expected to exceed 11 million by 2030.1,2 The
emerging area of targeted cancer therapy entails finding target
proteins uniquely mutated in cancer cells and identifying their
upstream and downstream signaling pathways. The serine/
threonine kinase Akt pathway plays a pivotal role, as it is
involved in a number of apoptotic pathways and is abnormally
upregulated in various types of cancers, including colon, breast,
brain, lung, and prostate.3,4 Inhibition of Akt for the treatment
of cancer has been validated in both preclinical animal models
and early phase human clinical trials.2,5
A synthesis of compound 1 on small scale has previously
been reported, involving a late-stage Suzuki coupling between
pyridyl chloride 2 and aryl boronate 3 (Scheme 1).6 A similar
disconnection was favored for the larger preparation as well,
since it would allow for the convergent late-stage union of two
fragments of roughly equal size and complexity. The
preparation of N-phthalimido-protected boronate 3 required
10 steps, including multiple protecting group manipulations,
and necessitated the use of genotoxic hydrazine in the final
deprotection step to generate the API 1. An alternative
retrosynthesis of the corresponding N-Boc-protected boronate
6 was therefore proposed (Scheme 2), in which the primary
amine moiety would be obtained through Curtius rearrange-
ment, with a diastereoselective nucleophilic addition to ketone
7 serving to establish the relative stereochemistry around the
cyclobutane ring. Control of the diastereoselectivity of the
methyl addition was anticipated to be a potential challenge.
The aryl electrophile coupling partner (e.g., 2 or 4) would
arise from the sequential appendage of heterocyclic rings onto
the functionalized pyridine core structure 5, following the
general strategy previously disclosed.6 Key issues for scale up of
the previously reported route included the use of microwave
conditions for several steps run at high temperatures and
pressures, and multiple chromatographic purifications.
Compound 1 (Scheme 1) has been identified as a potent
allosteric Akt inhibitor.6 We were recently required to execute
the first kilogram scale delivery of this active pharmaceutical
ingredient (API). In this paper we report the development of a
Scheme 1. Previously Reported Suzuki Coupling Strategy
RESULTS AND DISCUSSION
■
Synthesis of Triflate 4. Pyridyl aldehyde 5 was previously
prepared on gram scale in 5 steps from picoline derivative 9.6 A
Special Issue: Continuous Processes 2012
Received: February 3, 2012
Published: March 23, 2012
© 2012 American Chemical Society
1069
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
the corresponding dianion (14) followed by quenching with
DMF (or alternative electrophilic reagents) were found to be
highly capricious even on gram scale.13 After extensive
experimentation, it was found that N-lithio monoanion 13
could be formed cleanly at low temperature by treating aniline
11 with 1 equiv of methyllithium (MeLi) in THF; addition of
n-hexyllithium (n-HexLi)14 then resulted in metal−halogen
exchange to give dianion 14, which could be quenched by DMF
to give the desired aldehyde 5 in good assay yield (75−80%).
One major impurity was generated (7−8%) that resulted from
competitive debromination of starting material 11; however,
this was efficiently rejected during the product crystallization.
Cryogenic temperatures were required (−45 °C or below) for
the dianion formation; the yield began to drop significantly if
the reaction was run at higher temperatures (e.g., 62% assay
yield at −20 °C).
Scheme 2. Retrosynthesis of Target Compound 1 for
Kilogram Scale Production
For the bulk production, this formylation step was first run in
2 × 5 kg batches in a 100 L vessel at −60 °C. During the
second batch, the reaction solution turned into a thick gel at the
dianion (14) stage, a phenomenon which had not been
observed in gram scale experiments. The batch temperature had
to be raised to −45 °C and the agitator manually manipulated
until the mixture was sufficiently mobile to allow stirring to
continue automatically. Fortuitously, these additional oper-
ations did not impact the reaction yield, as each 5 kg batch gave
a 76% assay yield of aldehyde 5. However, from a practical
perspective, the gelling problem encountered with the longer
hold times needed for temperature control in these kilogram
scale runs presented a concern for further scale-up in fixed
vessel equipment. This spurred a preliminary evaluation of the
feasibility of performing the formylation under flow conditions,
which have been applied to a number of other organolithium
processes,15,16 since the more efficient mixing and heat transfer
in a continuous operation could result in shorter hold times.
A solution of anion 13 in THF was preformed using MeLi
below −40 °C and then fed into the flow reactor, the schematic
for which is depicted in Figure 1 (see the Experimental Section
for a detailed description). The reactor was constructed of 0.25-
in. internal diameter stainless steel tubing and immersed in a
dry ice/acetone bath to maintain a low temperature. Reagent
streams were fed by peristaltic pumps with pressure gauges (PI)
through polytetrafluoroethylene (PTFE) tubing. Within the
flow reactor, the stream of anion 13 was mixed with n-BuLi
(commercial 2.5 M solution in hexanes, used as is with no
precooling) via a T-connector attached to the first static line
mixer.17 Upon exiting the mixer, the stream of the resulting
dianion 14 was immediately combined with the solution of
DMF/THF in a second T-connector/static mixer arrangement.
In-process temperature was monitored using three stainless
steel thermocouples (T1−3). The product stream was then fed
through PTFE tubing into an aqueous quench solution (acetic
acid/tert-butyl methyl ether/water) in a glass vessel for
neutralization and workup. Pump flow rates and mixer
residence times were optimized on laboratory scale to maximize
conversion of bromide 11 to the desired aldehyde 5. The
residence times for the dianion formation and DMF alkylation
steps could be reduced to 2 and 2.5 s, respectively, while still
achieving complete conversion of the starting material and a
reaction purity profile equal to or better than batch mode.
To demonstrate proof-of-principle for this process, a
preparative scale run was perfomed using the setup shown at
the bottom of Figure 1. The anion 13 derived from 1 kg of
bromide 11 was processed through the flow reactor in 1 h, at a
shorter synthesis was developed, as shown in Scheme 3, taking
advantage of the potential for selective functionalization at the
C-2 position of commercially available 3-amino-6-chloropyr-
idine (8). Treatment of pyridine 8 with NBS in DMF resulted
in highly regioselective monobromination7,8 to give compound
10, which was isolated in 81% yield by direct crystallization
from the reaction mixture upon the addition of water.9
Acetonitrile (MeCN) could also be used for this reaction;
however, a higher yield of isolated product 10 was obtained
using DMF.
Selective mono-Boc protection of aniline derivatives can be
difficult due to low nucleophilicity of the sp2-nitrogen.
Protection of aniline 10 using Boc2O/Et3N/DMAP in MeCN
was reported to proceed in 58% yield.10 In our hands, the use of
an established literature method (2 equiv of NaHMDS and
then 1 equiv of Boc2O in THF at room temperature)11 resulted
in improved yields but was plagued by incomplete conversion
of aniline 10 and the formation of up to 11% bis-protected
compound 12. After operational improvements to this step, a
reproducible protocol was developed that involved the addition
of NaHMDS to a THF solution of aniline 10 and Boc2O at −5
°C. This generated <1% bis-protected compound 12 on
multikilogram scale, and bromide 11 was isolated in 83% yield
and >99.5 wt % purity12 by aqueous (aq) workup followed by
crystallization from isopropanol (IPA)/water.
Conversion of bromide 11 to aldehyde 5 was initially very
problematic. Experiments using ≥2 equiv of n-BuLi to generate
Scheme 3. Preparation of Pyridyl Aldehyde 5
1070
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
Scheme 4. Conversion of Aldehyde 5 to Triflate 4
Figure 1. Flow setup: schematic (top) and kilogram-scale run
(bottom).
The displacement reaction to convert chloride 17 to
hydrazine 18 had used 7 equiv of hydrazine hydrate in pyridine
under microwave irradiation at 160 °C.6 Our own screening
efforts confirmed that very forcing conditions were indeed
required to effect this transformation. The most practical
procedure that could be developed, within the time frame of
meeting the target API delivery date, involved reacting chloride
17 in 4−5 volumes of 1,3-dimethyl-2-imidazolidinone (DMI)
with 18 equiv of hydrazine (35% aq solution) at 155 °C in an
autoclave.
flow rate of 114 mL/min. Efficient cooling of the system in the
rudimentary cold bath was maintained even at this high flow
rate (the temperatures recorded at steady-state were −70, −65,
and −55 °C at T1, T2, and T3, respectively). The solution
assay yield of aldehyde 5 at the end of the run (85%) was
higher than on the 5 kg scale in batch mode (76%), and the
level of debrominated side-product was lower (4% vs 7−8%).
Furthermore, even with a slightly higher concentration of
dianion 14 in flow mode (in 7 volumes of solvent compared to
10 in batch mode), no gelling or plugging of the reactor was
observed.
After workup, the organic layer from the 1 kg flow run was
combined with both 5 kg batches, and aldehyde 5 crystallized as
one lot by solvent-switch into IPA followed by addition of
water. A total of 6 kg of aldehyde 5 was isolated for a combined
isolated yield of 63%. A relatively high loss of product (13%) to
crystallization liquors was accepted for this delivery, since the
addition of too much water antisolvent resulted in oiling out of
the product.
The reported conditions for the conversion of aldehyde 5 to
naphthyridone 17 used phenylacetyl chloride and DBU in THF
at 100 °C.6 An alternative procedure was developed that
proceeded under milder conditions using methyl phenylacetate
and Cs2CO3 in MeCN. The reaction temperature was ramped
successively from room temperature (rt) to 45−50 °C to 70−
75 °C to funnel the proposed intermediates 15 and 16 along
the mechanistic pathway illustrated in Scheme 4.18 Cyclized
product 17 crystallized directly from the reaction mixture at the
end of the reaction, with excellent rejection of impurities.19
Water was then added to dissolve inorganic salts, and the
product was collected by filtration. The isolated yield (92 wt %,
60% corrected yield) was moderate; however, the conditions
were reproducible and the constraints of the project timeline
precluded further optimization of this step.
Such conditions involving the use of aqueous hydrazine at
high temperature clearly mandated rigorous safety evaluation
before being run on scale. A differential scanning calorimetry
(DSC) analysis of a sample of the reaction mixture prior to heat
up showed only a minor exotherm of 7.5 cal/g of reaction
mixture initiating at 221 °C. Further testing was performed
using a closed system rapid screening device (RSD). In one
experiment a reaction mixture sample was heated to 166 °C
and aged for 14 h, with no unexpected exothermic activity
being observed (Figure 2a). However, noncondensable gas was
generated, with the maximum gas generation rate (1.92 psi/
min) occurring at the start of the age, reaching a maximum
pressure of 382 psi after ∼4 h, and then decreasing to 282 psi
over the remaining ∼11.5 h. Assuming hydrazine was the
source of generated gas, 0.21 mol of gas was generated per mol
of hydrazine. In the second experiment, a sample was heated
from 22 to 255 °C (100 °C above the proposed reaction
temperature) at 3 C°/min (Figure 2b). Again, no unexpected
thermal activity was observed, but there was noncondensable
gas generation. The pressure of 652 psi was observed after cool-
down to 155 °C, equating to 0.48 mol of gas generated per
mole of hydrazine.20 Extrapolating these results to the proposed
production scale resulted in a predicted maximum gas
generation rate of 6.7 L/min and a maximum pressure for a
closed system (with no venting) of 1454 psig at 166 °C and of
2923 psig at 155 °C after first heating to 255 °C. These values
were well within the operating range of the equipment train to
1071
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
Water was then added to quench excess anhydride reagent and
reduce product liquor losses. The product cake that was first
isolated was subsequently reslurried in MeCN/water to
upgrade the purity, with triazole 19 being isolated in 91%
overall yield from hydrazine 18. This upgrade was necessary for
satisfactory performance of the material in the subsequent
triflation step.23
The conversion of triazole 19 to aryl chloride 2 could only be
effected, with any modest degree of success, using POCl3 under
microwave conditions at temperatures above 100 °C. We
proposed that the corresponding triflate 4 (Scheme 4) could be
generated under milder conditions and would serve as a viable
coupling partner in the Suzuki coupling. Treatment of triazole
19 with Tf2O (1.5 equiv) in the presence of a base indeed led
to the clean formation of the desired triflate 4 in good yield. A
small, focused screen of reaction conditions identified pyridine,
MeCN, and 40−50 °C as the optimum base, solvent, and
reaction temperature, respectively. Under these conditions,
>98% conversion of pyridone 19 was achieved within 2 h on
scale, with the product 4 then being crystallized by the addition
of water and isolated in 91% yield. A total of 3.4 kg of triflate 4
was produced in 7 steps, and 18% overall yield, from 3-amino-
6-chloropyridine 8. Notably, in five of these steps, the desired
product could be isolated directly from the reaction mixture,
with no aqueous workup or other purification procedures
needed.
Preparation of trans-Hydroxy Acid 27. The cyclobutane
ring was constructed in a one-pot operation involving
sequential C-alkylations of the dianion of 4-chlorophenylacetic
acid 20, generated using i-PrMgCl as base in THF, with
epichlorohydrin as the 1,3-bis-electrophile (Scheme 5). On 20
kg scale the cis-hydroxy acid product was isolated in 59% yield
and 96% purity following aqueous workup and crystallization
from toluene (PhMe).24 A small amount of dechlorination
occurred during the reaction. The resulting des-chloro impurity
22 (4%) cocrystallized with the desired product 21; however,
sufficient rejection of des-chloro impurities was observed
during later isolations. Esterification of acid 21 under standard
conditions (MeOH, cat. H2SO4) proceeded smoothly to give
methyl ester 23 (90% yield), which was used crude in the
subsequent bleach/TEMPO oxidation to afford cyclobutanone
7 in quantitative yield. Due to the fact that ketone 7 is a low-
melting solid, it was not isolated but instead carried forward as
a solution in PhMe into the methyl addition step.
Initial gram scale experiments evaluating the addition of
MeMgBr to ketone 7 were not promising. At 0 °C in THF the
diastereoselectivity was poor (24:25 ∼1.6:1), reactions would
stall at ∼80% conversion (due to competing enolate
formation), and chromatographic purification was required to
isolate samples of the desired trans-hydroxy ester 24 (since this
material is an oil). A breakthrough was found when the reaction
was performed under cryogenic conditions in a less polar
solvent mixture (10:1 PhMe/THF, a small amount of THF was
necessary to maintain solubility of the ketone starting material
7 at low temperatures); the addition proceeded to >95%
conversion and with an improved diastereoselectivity of ∼2.5:1.
A difference in the relative rates of hydrolysis of the ester
groups in compounds 24 and 25 was observed, with cis-isomer
25 undergoing saponification much more rapidly. This allowed
for separation of the resulting undesired cis-acid 26 from the
desired trans-ester 24 by simple aqueous extraction.
Figure 2. RSD testing of hydrazine displacement reaction: (a)
isothermal age at 166 °C; (b) temperature cycling.
be used,21 and they showed that the reaction was safe to run
under the conditions described.22
The conversion of chloride 17 to hydrazine 18 was run in 3
× 1.1 kg batches, each of which proceeded to >99% conversion.
The highest pressure generation was 1450 psi, at a maximum
rate of 7.6 psi/min, in excellent agreement with the
extrapolation from the RSD testing. At the end of each
reaction, the product slurry was diluted with water to minimize
product liquor losses and reduce the concentration of residual
NH2NH2. The batches were combined for filtration, affording
2.6 kg of hydrazine 18 in 98% purity and 84% yield. The small
amount (0.7%) of unreacted chloride starting material 17 was
not rejected during this isolation but could be purged during
downstream steps.
The optimum conditions for the subsequent triazole
formation employed 3 equiv of difluoroacetic anhydride and
5 equiv of Hunig’s base in MeCN at 5−20 °C. The use of less
̈
than 3 equiv of the anhydride gave inconsistent results with
long reaction times or stalled reactions. MeCN was preferred to
DMF as solvent due to a slightly cleaner reaction profile and
better impurity rejection. Although the conversion of hydrazine
18 to triazole 19 is a slurry-to-slurry process, it behaved
reproducibly on scale, with the reaction completing within 2 h.
On scale, the addition of MeMgBr to ketone 7 at −70 °C
proceeded to 98% conversion and with a diastereoselectivity of
1072
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
Scheme 5. Preparation of trans-Hydroxy Acid 27
Table 1. Screening of MeMgBr Addition to Cyclobutanone
Substrates
a
entry substrate
R
X
solvent
temp (°C)
dr
1
7
CO2Et
Cl
THF
THF
THF
PhMe
PhMe
THF
THF
THF
PhMe
THF
0
0
1.6:1
b
2
7
1:10
2.1:1
1.6:1
2.5:1
1:1
3
7
−78
0
4
7
5
7
−78
0
6
28
28
29
29
30
30
CO2H
NHBoc
CN
Cl
Br
Cl
c
7
0
1:1
8
0
1:1.5
1:2
9
0
10
11
0
>20:1
>20:1
PhMe
0
a
1
Ratio of trans/cis-adducts determined by H NMR analysis of crude
b
c
product mixtures. Cul (1 equiv) used as an additive. MeLi used
instead of MeMgBr.
Interestingly, a ratio of >20:1 in favor of the desired trans-
isomer was obtained when nitrile 30 was employed (entries 10
and 11). This promising screening hit could not be evaluated
further due to the timeline constraints of the kilogram-scale
delivery, but it clearly represents an opportunity that could be
investigated to favorably impact future API campaigns.
Completion of the Synthesis. The procedure developed
in the lab for the Curtius rearrangement of acid 27 involved the
initial generation of acyl azide intermediate 31, using
diphenylphosphoryl azide (DPPA) in THF, followed by the
addition of aqueous hydrochloric acid (HCl) at reflux to induce
rearrangement.26 In the pilot plant this step performed poorly
(Scheme 6), giving a low conversion to the desired amine 32
and a large amount of the urea byproduct 33. Urea 33 could be
separated and then hydrolyzed under very forcing conditions to
recover an additional quantity of amine 32, which was all
processed forward into the Boc protection step. The resulting
protected compound 34 was isolated by crystallization from
tert-butyl methyl ether (MTBE)/heptane. Noteworthy is the
fact that carbamate 34 is the first crystallized intermediate since
acid 21 (Scheme 5). At this stage carbamate 34 still contained
2.5% of the des-chloro impurity 35, which had been carried
over since the first step in the synthesis. Recrystallization of
carbamate 34 from acetone/water (95% recovery) reduced the
level of des-chloro impurity 35 to 0.4%, and improved the
purity of the product 34 to 99.3%.
A total of 2.87 kg of carbamate 34 was procured from an
input of 7.1 kg of acid 27 into the Curtius rearrangement step.
This modest amount of product was sufficient for the presently
described delivery. However, the low yield for this overall
transformation (30%) would not be sustainable in the longer
term. The root cause for the generation of urea 33 was
identified as being the prolonged (4 h) addition time of the aq
HCl that was required to control the rate of gaseous N2 and
CO2 evolution on scale. Subsequent lab-scale development
showed that inverse addition of the solution of acyl azide 31
into the aq HCl would provide a method to minimize the
formation of urea 33.
2.7:1. After selective hydrolysis of the product mixture using
LiOH in aq THF (5% of the desired trans-ester 24 was also
hydrolyzed under these conditions) and removal of the
resulting cis-acid 26, the resulting organic stream containing
trans-ester 24 was saponified under more forcing conditions (aq
NaOH, ethanol) to generate trans-acid 27.
The net effect of the steps from cis-acid 21 to trans-acid 27 is
only to convert a secondary alcohol group into a tertiary
alcohol plus one carbon unit (methyl). For the immediate
needs of the delivery, this sequence provided a timely, low-tech
solution that could be telescoped into a through process and
which proceeded in a serviceable 43% overall yield with no
chromatography required. It did not, however, solve the
fundamental problem of the low diastereoselectivity of the
methyl addition step. With a view to enabling subsequent larger
deliveries of API 1, this addition was subsequently screened
against a number of substrates on small scale using high-
throughput experimentation techniques.25 Selected data is
presented in Table 1.
As described above, moderate diastereoselectivity in the
direction of the desired trans-adduct was observed with
ketoester substrate 7 (entries 1 and 3−5) but could not be
improved further. Indeed, performing this addition in the
presence of a stoichiometric amount of CuI resulted in a switch
in selectivity to strongly favor the cis-isomer (entry 2). Addition
to ketoacid 28 using MeMgBr (entry 6) or MeLi (entry 7) was
nonselective. Marginal selectivity in favor of the undesired cis-
adduct was found with N-Boc substrate 29 (entries 8 and 9).
1073
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
Scheme 6. Conversion of Acid 27 to Akt Inhibitor 1
Table 2. Suzuki Coupling Screening
a
b
c
entry
ligand
solvent
assay yield (%)
1
2
3
4
5
6
7
8
X-Phos
DPPB
THF/H2O
THF/H2O
THF/H2O
THF/H2O
THF/H2O
2-MeTHF/H2O
DMAc/H2O
2-MeTHF
52
71
70
76
84
82
57
22
(R)-BINAP
DPPF
DIPPF
DIPPF
DIPPF
DIPPF
a
b
DIPPF = 1,1′-bis(di-i-propylphosphino)ferrocene. 10% v/v H2O in
c
mixed solvent systems. HPLC yield relative to internal standard.
% each, with only a very slight excess of the valuable boronate
coupling partner 6 (1.1 equiv) needing to be employed. On
kilogram scale the coupling proceeded to completion within 1 h
while warming from rt to 50 °C (Scheme 6). At the end of the
reaction, a biphasic mixture was formed, and the product 36
began to crystallize from the organic layer as the mixture was
cooled to ambient temperature. After addition of MTBE
antisolvent to reduce liquor losses, filtration and drying
afforded penultimate 36 in good yield (84%) and with excellent
rejection of Pd impurities (residual Pd in the product cake was
only 30 ppm).27 Treatment of penultimate 36 with 5 M aq HCl
(5 equiv) in ethanol (EtOH) at 50 °C for 24 h resulted in clean
deprotection of the Boc group; upon pH adjustment to 10−11
using aq NaOH, the desired polymorph of the API free base
crystallized directly from the reaction mixture. The target
compound 1 (2.6 kg) was isolated in 93% yield and met all
required purity specifications.
A palladium-catalyzed borylation of carbamate 34 afforded
boronate 6 in 91% isolated yield. During the aqueous workup,
removal of the pinacol byproduct by successive water washes
was necessary to ensure a reproducible crystallization of the
product 6 from heptane. A total of 3.4 kg of boronate 6 was
produced as a free-flowing white powder in 8 steps (7% overall
yield) from 4-chlorophenylacetic acid 20.
The reaction conditions for the Suzuki coupling between
triflate 4 and boronate 6 were screened in an effort to identify a
milder alternative to using an unoptimized method of 20 mol %
Pd(PPh3)4 at 140 °C in a microwave.6 Key results are
summarized in Table 2. Electron-rich bidentate phosphines
outperformed monodentate ligands (compare entry 1 to entries
2−5), allowing for good conversion at or slightly above room
temperature. Ferrocene derivatives (entries 4 and 5) proved
superior to acyclic (entry 2) or binaphthyl analogues (entry 3).
Ethereal solvents, e.g. THF (entry 5) or 2-methyltetrahy-
drofuran (2-MeTHF) (entry 6), gave better product yields than
more strongly dipolar aprotic solvents, such as dimethylaceta-
mide (DMAc) (entry 7). The presence of water cosolvent in
the reaction was found to be crucial, with the yield dropping
substantially under anhydrous conditions (entry 8). Ultimately,
2-MeTHF was chosen over THF due to the relative ease of
product isolation, since the coupled penultimate 36 could be
crystallized directly from the reaction mixture using the former
solvent.
CONCLUSION
■
A fit-for-purpose, kilogram scale synthesis of Akt kinase
inhibitor 1 has been developed. The convergent route consisted
of 17 total steps (10 steps longest linear sequence), required no
chromatographic purifications, and could be performed safely
on multikilogram scale. Key transformations included a high
temperature and pressure hydrazine displacement reaction, and
separation of a diasteromeric mixture of Grignard addition
products through chemoselective ester hydrolysis. Proof-of-
concept was established on kilogram scale for a key dianion
formylation reaction under flow conditions in which yield and
practicality improvements over batch mode were demonstrated,
providing a further example of the utility and potential of
continuous flow processing as an enabling technology in
process chemistry.
With further optimization, >99% conversion of triflate 4
could be achieved at lower catalyst and ligand loadings of 3 mol
1074
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
charged at rt to a separate 5 L round-bottomed flask equipped
with the feed line for FMI pump 3. The solution of anion 13
was pumped by FMI pump 1 at a rate of 114 mL/min. n-BuLi
was pumped by FMI pump 2 at a rate of 24 mL/min. The
anion 13 and n-BuLi streams were mixed in a Swagelok T
(0.125 in. internal diameter) followed by a temperature
indicator (T1), static mixer (0.25 in. internal diameter, 6 mL
volume, 2.0 s residence time), and second temperature
indicator (T2), to generate dianion 14. The DMF solution
was pumped by FMI Pump 3 at a rate of 37 mL/min. The
dianion 14 and DMF solutions were mixed in a Swagelok T
(0.125 in. internal diameter) followed by a static mixer (0.25 in.
internal diameter, 6 mL volume, 2.5 s residence time) and a
third temperature indicator (T3). During the kilogram run the
temperatures recorded at T1, T2, and T3 were −70, −65, and
−55 °C, respectively, and the pressure on the pumps was in the
range of 40−50 psi. The mixture then exited the flow reactor.
The first 30 s of product stream was discarded (while the
system reached steady-state) and then the remainder was fed
into a stirred quench solution (0.72 L acetic acid (AcOH), 1.5
L MTBE, and 6 L water) in a 30 L jacketed cylindrical vessel at
rt. After 1 h, all the solution of anion 13 had been processed.
The layers in the quench solution were separated, and the
organic layer was washed with saturated aqueous NaHCO3 (1
× 3 L) and then water (1 × 3 L). The organic layer (10.1 kg,
707 g assay 5, 85% assay yield) was collected and stored in a
plastic drum.
b. Batch Preparation. A 100 L round-bottomed flask was
charged with THF (36 L) and bromide 11 (4.8 kg, 15.6 mol,
1.0 equiv) at rt, and the resulting solution was cooled to −65
°C. MeLi (5.14 L of a 3.19 M solution in diethoxymethane,
16.4 mol, 1.05 equiv) was added over 1 h below −55 °C
(notemild effervescence, methane).28 The resulting dark
solution was cooled to −65 °C over 20 min and then n-HexLi
(7.46 L of a 2.3 M solution in hexane, 17.2 mol, 1.1 equiv) was
added over 50 min below −55 °C. The dark orange solution
was aged at −55 °C for a further 5 min and then a solution of
DMF (1.93 L, 25.0 mol, 1.6 equiv) in THF (2 L) was added
over 30 min below −45 °C. A separate 100 L cylindrical vessel
was charged with the quench solution of water (28 L), AcOH
(3.87 L, 60.4 mol, 3.9 equiv), and MTBE (8 L) at rt. The
reaction mixture was then transferred over 15 min into the
quench solution. The resulting biphasic mixture was aged at 15
°C for 15 min, and then the layers were allowed to settle. The
layers were separated, and the organic layer was washed with
saturated aqueous NaHCO3 (1 × 15 L) and then water (1 × 15
L). The organic layer (43.2 kg, 3.16 kg assay 5, 76% assay yield)
was collected and stored in plastic drums while a second
formylation batch was run using 4.7 kg of bromide 11.
EXPERIMENTAL SECTION
■
2-Bromo-6-chloropyridin-3-amine (10).10 2-Chloro-5-
aminopyridine (8; 8.0 kg, 62.2 mol) was dissolved in DMF (32
L) in a 100 L jacketed cylindrical vessel, and the resulting
solution was cooled to 15 °C. NBS (10.85 kg, 61.0 mol, 0.98
equiv) was added in portions over 2 h below 25 °C. The
mixture was stirred for a further 2 h, then water (48 L) was
added over 1 h below 25 °C, and the resulting slurry was aged
at rt for 16 h (liquors assayed for 13.3 mg/mL of product 10 at
this point, ∼10% of theoretical yield). The slurry was filtered,
and the resulting product was washed with 2:1 water/DMF (20
L) and then dried under vacuum with a N2 sweep at rt and then
40 °C for 3 days to afford the title compound (10.50 kg, 50.6
mol, 81% yield) as a brown solid.
tert-Butyl (2-Bromo-6-chloropyridin-3-yl)carbamate
(11). Bromide 10 (4.9 kg, 23.6 mol, 1.0 equiv), Boc2O (5.05
kg, 23.1 mol, 0.98 equiv), and THF (20 L) were charged to a
100 L jacketed cylindrical vessel, and the resulting solution was
cooled to −5 °C. NaHMDS (43.32 kg of a 40 wt % solution in
THF, 47.2 mol, 2.0 equiv) was added over 2.5 h below 0 °C.
The solution was warmed to 20 °C, at which point HPLC
analysis indicated complete conversion of bromide 10. The
reaction was quenched by addition of water (24 L) and
isopropyl acetate (IPAc) (24 L). The layers were separated, and
the organic layer was washed with 1 N aq HCl (1 × 24 L) and
then 10% aq NaCl (1 × 24 L). The organic layer was held in
plastic drums while a second reaction of the same batch size
was run. The combined organic layers from both batches were
concentrated under vacuum to a low volume and switched to
1
approximately 35 L of IPA (∼60 L total of IPA was used). H
NMR analysis was used to confirm that all THF and IPAc had
been removed. Water (20 L) was added slowly, and the
resulting slurry was aged at rt for 16 h (liquors assayed for 6.6
mg/mL of product 11 at this point, ∼6% of theoretical yield).
The slurry was filtered, and the resulting product was washed
with 2:1 water/IPA (20 L) and then dried under vacuum with a
N2 sweep at 40 °C for 24 h to afford the title compound (12.0
kg, 39.0 mol, 83%) as a pink solid. mp 95−96 °C; IR (neat)
3349, 2984, 1717, 1497, 1249, 1156, 1143, 1128, 767 cm−1; 1H
NMR (400 MHz, CDCl3) δ 8.43 (1H, d, J = 8.5 Hz), 7.23 (1H,
d, J = 8.5 Hz), 6.98 (1H, br s), 1.50 (9H, s); 13C NMR (101
MHz, CDCl3) δ 151.7, 142.3, 133.5, 129.6, 129.1, 123.7, 82.1,
28.1; HRMS calcd for C10H13BrClN2O2 [MH]+ 306.9849,
found 306.9854.
tert-Butyl (6-Chloro-2-formylpyridin-3-yl)carbamate
(5). a. Flow Preparation. The flow reactor was constructed
of stainless steel tubing, maintained at low temperature by
immersion in a dry ice/acetone bath. Feed lines to/from the
pumps, and from the reactor to the quench solution, were
Teflon-coated, and each pump was followed by a pressure
indicator and check valve. A 22 L round-bottomed flask was
charged with THF (4.5 L) and bromide 11 (1.0 kg, 3.25 mol,
1.0 equiv) at rt, and the resulting solution was cooled to −55
°C. MeLi (1.07 L of a 3.19 M solution in diethoxymethane,
3.41 mol, 1.05 equiv) was added over 30 min below −40 °C
(notemild effervescence, methane).28 The resulting solution
was stirred for a further 15 min, and then the flask was
equipped with the feed line for FMI pump 1. Concurrently, n-
BuLi (1.43 L of a 2.5 M solution in hexanes, 3.58 mmol, 1.1
equiv) was charged at rt to a 5 L round-bottomed flask
equipped with the feed line for FMI pump 2, and a solution of
DMF (0.38 L, 4.88 mol, 1.5 equiv) in THF (2.06 L) was
c. Product Isolation. The organic layers from both batches
were combined with the organic stream from the flow
preparation described in section a above, obtaining a total of
6.82 kg assay 5. The combined organic layers were then
concentrated under vacuum at 35−40 °C in a 100 L round-
bottomed flask to a volume of ∼30 L. The solution was then
flushed with IPA (30 L), maintaining the batch volume at ∼30
L, during which time the product began to crystallize. The
resulting slurry was then allowed to cool from 38 °C to rt over
2 h. Water (16 L) was then added over 1.5 h. By the end of the
water addition, the slurry had changed consistency, and a small
amount of the product had begun to oil out. Additional IPA (14
L) was added, and the batch heated to 65 °C over 1 h. The
resulting thin slurry was allowed to cool to rt over 3 h. The
1075
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
mixture was then filtered and the product was washed with 3:2
IPA/water (2 × 8 L) and then dried under vacuum with a N2
sweep at rt for 2.5 days to afford the title compound (5.99 kg,
95 wt % purity, 22.2 mol, 63% corrected yield) as a light beige
powder. mp 98−98.5 °C; IR (neat) 3294, 2985, 2843, 1735,
°C ranged from 1450 psi for the first batch to 900 psi for the
third batch). After the third batch the DMI vessel rinse slurry
(notethe same DMI rinse was used for all three batches) was
transferred into a 30 L extractor. The stirred slurry was diluted
with water (8 L) and agitated for 1 h. This slurry and the three
reaction mixture batches were then filtered together to isolate
the product. The cake was washed with water (1 × 8 L) and
EtOH (2 × 8 L), and it was then dried under vacuum with a N2
sweep for 16 h at rt and then 2 days at 35 °C to give the title
compound (2.585 kg, 98.4 wt % purity, 10.1 mol, 84%
corrected yield) as a muddy yellow powder. mp 255−258 °C
(dec); IR (neat) 3321, 3201, 2785, 1662, 1621, 1477, 1447,
1
1680, 1492, 1227, 1137, 898 cm−1; H NMR (400 MHz,
CDCl3) δ 10.16 (1H, br s), 9.95 (1H, d, J = 1.0 Hz), 8.85 (1H,
d, J = 9.0 Hz), 7.45 (1H, d, J = 9.0 Hz), 1.51 (9H, s); 13C NMR
(101 MHz, CDCl3) δ 195.6, 152.4, 143.2, 138.2, 135.7, 129.8,
129.6, 82.1, 28.1; HRMS calcd for C11H14ClN2O3 [MH]+
257.0693, found 257.0685.
6-Chloro-3-phenyl-1,5-naphthyridin-2(1H)-one (17). A
100 L jacketed cylindrical vessel was charged with MeCN (33
L) and aldehyde 5 (5.8 kg, 21.5 mol, 1.0 equiv) at rt. Methyl
phenylacetate (3.46 L, 24.0 mol, 1.1 equiv) was added to the
thin slurry, rinsing with MeCN (1 L). Cs2CO3 (24.5 kg, 75 mol,
3.5 equiv) was then added in portions over 10 min to the
orange solution, rinsing with MeCN (1 L). The resulting slurry
was aged at rt for 1 h (during which time the temperature had
risen slowly by 5 °C, from 22 to 27 °C). The slurry was then
heated to 50 °C over 1.5 h, aged at that temperature for a
further 1 h, then heated to 70−75 °C over 3 h, aged at that
temperature for a further 5 h, and then allowed to cool to rt
overnight (∼16 h). The slurry was then cooled to 10 °C, and
water (35 L) was added over 1 h below 25 °C. The resulting
biphasic slurry was aged at 20 °C for 1 h and then filtered. The
product cake was washed with 1:1 MeCN/water (2 × 8 L) and
then dried under vacuum with a N2 sweep at rt for 24 h and
then at 35 °C for 3 days to give the title compound (3.57 kg, 92
wt % purity, 60% corrected yield) as a cream-colored powder.
mp 262−264 °C (dec); IR (neat) 2781, 1652, 1591, 1443,
1
687, 620 cm−1; H NMR (400 MHz, DMSO-d6) δ 11.85 (1H,
br s), 7.79−7.75 (3H, m), 7.65 (1H, s), 7.48 (1H, d, J = 9.0
Hz), 7.45−7.35 (3H, m), 6.95 (1H, d, J = 9.0 Hz), 4.18 (2H, br
s); 13C NMR (101 MHz, DMSO-d6) δ 159.7, 157.8, 137.1,
136.3, 134.2, 133.9, 128.7, 128.2, 128.0, 127.9, 124.7, 112.0;
HRMS calcd for C14H13N4O [MH]+ 253.1089, found
253.1080.
1-(Difluoromethyl)-8-phenyl[1,2,4]triazolo[4,3-a][1,5]-
naphthyridin-7(6H)-one (19). A 50 L jacketed cylindrical
vessel was charged with MeCN (24.5 L), hydrazine 18 (2.45 kg,
9.5 mol, 1.0 equiv), and i-Pr2NEt (8.31 L, 47.6 mol, 5.0 equiv)
at rt. The yellow slurry was cooled to 5 °C over 20 min, and
difluoroacetic anhydride (3.19 L, 28.6 mol, 3.0 equiv) was
added over 45 min below 15 °C. The resulting yellow slurry
was warmed to rt and aged until <1% starting material 18 (as
measured by HPLC area count) remained (∼2 h). The thin
yellow slurry was cooled to 10 °C over 20 min, and water (400
mL) was added over 10 min below 17 °C. The mixture was
allowed to warm to rt, and additional water (5.6 L) was added
over 75 min. The yellow slurry was aged for a further 30 min
(liquors assayed for 1.3 mg/mL of product 19 at this point)
and then filtered. The product cake was washed with 7:2
MeCN/water (1 × 10 L) and then with EtOH (1 × 10 L), and
it was then dried under vacuum with a N2 sweep at rt for 2 days
1
1113, 918, 694, 669 cm−1; H NMR (400 MHz, DMSO-d6) δ
12.28 (1H, br s), 8.01 (1H, s), 7.80−7.77 (2H, m), 7.73 (1H, d,
J = 9.0 Hz), 7.59 (1H, d, J = 9.0 Hz), 7.47−7.39 (3H, m); 13C
NMR (101 MHz, DMSO-d6) δ 160.3, 143.7, 136.7, 136.6,
136.5, 135.2, 133.8, 128.9, 128.6, 128.1, 126.2, 125.2; HRMS
calcd for C14H10ClN2O [MH]+ 257.0482, found 257.0477.
6-Hydrazinyl-3-phenyl-1,5-naphthyridin-2(1H)-one
(18). a. Vessel Cleaning and Conditioning. A 5-gal
Hastelloy C autoclave,29 fitted with a rupture disk rated for
3100 psi, was cleaned and rinsed to neutral pH with deionized
(DI) water. The autoclave was charged with additional DI
water (12 L) and heated to 155 °C. Samples of the final DI
water rinse were submitted for metals analysis (Pd, Pt, Ru, Rh <
1 ppb; Ni 1 ppm; Cr 0.18 ppm; Mo 0.38 ppm; W 0.22 ppm).
The vessel was then conditioned for 1 h at 155 °C with 1% aq
NH2NH2 (10 L) prior to carrying out the reaction.
b. Hydrazine Displacement Reaction. Chloride 17 (1.11 kg,
92 wt %, 4.0 mol, 1.0 equiv) was slurried in DMI (4.5 L) and
charged by residual vacuum into the 5-gal autoclave. A 1 L
follow flush of DMI was used to rinse the lines. Hydrazine (6.5
L of a 35% aq solution, 72.0 mol, 18.0 equiv) was charged into
the autoclave by residual vacuum and the charge line rinsed
into the vessel with DI water (200 mL). Nitrogen pressure (80
psi) was placed on the vessel, and the batch temperature was
heated to 155 °C over 1.5 h. The batch was aged at 155 °C for
4 h and then cooled to 40 °C. The batch was discharged from
the vessel in two equal portions into two 5-gal polyjugs which
each contained 3 L DI water. The autoclave vessel was rinsed
with DMI (14 L), and the DMI was agitated overnight and then
discharged from the vessel into a third 5-gal polyjug.
1
to afford the title compound (3.2 kg, 88 wt % purity by H
NMR analysis) as a yellow solid. The solid was reslurried in 7:2
MeCN/water (12.25 L) in a 30 L jacketed cylindrical vessel at
rt for 4 h. The slurry was filtered, and the product cake was
washed with 7:2 MeCN/water (1 × 12 L) and then with EtOH
(1 × 12 L) and finally with MTBE (1 × 12 L). The product was
dried under vacuum with a N2 sweep at rt for 4 days to give the
title compound (2.72 kg, 100 wt % purity, 8.7 mol, 91% yield
from 18) as a yellow solid. mp >300 °C; IR (neat) 2777, 1644,
1
1577, 1128, 1018, 789, 699, 651 cm−1; H NMR (500 MHz,
DMSO-d6) δ 12.77 (1H, br s), 8.37 (1H, s), 8.10 (1H, d, J = 9.5
Hz), 8.01 (1H, t, J = 51.5 Hz), 7.80−7.78 (2H, m), 7.64 (1H, d,
J = 9.5 Hz), 7.52−7.48 (2H, m), 7.46−7.43 (1H, m); 13C NMR
(126 MHz, DMSO-d6) δ 159.3, 149.4, 140.7 (t, J = 27.5 Hz),
135.4, 132.6, 131.6, 128.7, 128.4 (2C), 127.6 (t, J = 6.0 Hz),
121.9, 118.1, 115.9, 110.0 (t, J = 234.5 Hz); 19F NMR (471
MHz, DMSO-d6) δ −115.1; HRMS calcd for C16H11F2N4O
[MH]+ 313.0901, found 313.0899.
1-(Difluoromethyl)-8-phenyl[1,2,4]triazolo[4,3-a][1,5]-
naphthyridin-7-yl Trifluoromethanesulfonate (4). A 50 L
jacketed cylindrical vessel was charged with MeCN (17.2 L),
triazole 19 (2.65 kg, 8.5 mol, 1.0 equiv), and pyridine (1.72 L,
21.2 mol, 2.5 equiv). The resulting yellow slurry was cooled to
5 °C over 15 min, and Tf2O (2.15 L, 12.7 mol, 1.5 equiv) was
added over 40 min below 15 °C. The slurry was warmed to 40
°C over 80 min, by which point an orange solution had formed
and <2% triazole starting material 19 remained (as measured by
This reaction was repeated twice more on the same scale
(notethe vessel pressure at the end of the reaction age at 155
1076
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
HPLC area count). The solution was cooled to 7 °C over 45
min, and water (300 mL) was added over 15 min below 25 °C.
The resulting reddish brown solution (15 °C) was transferred
under vacuum through a 1 μm in-line filter into a 50 L jacketed
cylindrical vessel at rt. Water (0.93 L) was then added in one
portion. The solution was aged for 5 min and then seeded and
aged for a further 30 min. Additional water (20.3 L) was added
over 2 h to the resulting slurry, which was then aged for a
further 30 min (liquors assayed for 3.7 mg/mL of product 4 at
this point). The slurry was filtered, and the product cake was
washed with 4:5 MeCN/water (2 × 8 L) and then dried under
vacuum with a N2 sweep at rt for 40 h to give the title
compound (3.41 kg, 100 wt % purity, 7.7 mol, 91% yield) as a
beige solid. mp 130.5−131.5 °C; IR (neat) 1521, 1421, 1209,
the resulting mixture was stirred for 30 min at 20−30 °C. Conc
H2SO4 (2.2 kg, 22 mol, 0.03 equiv) was added, and the mixture
was aged for 30 min before being heated to 50−60 °C and aged
at this temperature for 2.5 h (<2% acid starting material 21
remained at this point, as measured by HPLC area count). The
solution was cooled to 0−10 °C, then 8% w/v aq K2HPO4 (180
kg) was added below 30 °C, and the mixture was aged for a
further 1.5 h. The solution was then concentrated under
vacuum below 50 °C to a volume of ∼160 L. PhMe (99 kg)
was then added at rt, and the resulting biphasic mixture was
stirred for 1 h. The layers were separated, and the organic layer
was concentrated to dryness under vacuum at 45−60 °C to give
the title compound (23.65 kg, 66.2 wt % purity by HPLC
analysis, 65 mol, 90% corrected assay yield) as a crude oil which
was used directly in the next step. Residual solvent levels for
PhMe, water, and MeOH were 21.3%, 0.01%, and not detected,
respectively.
Data for an authentic sample of ester 23 purified by column
chromatography on silica gel: colorless oil; IR (neat) 3402,
2991, 2950, 1728, 1492, 1235, 1130, 1094, 720 cm−1; 1H NMR
(400 MHz, CDCl3) δ 7.32−7.27 (4H, m), 4.16−4.09 (1H, m),
3.61 (3H, s), 3.51 (1H, br s), 2.89−2.84 (2H, m), 2.73−2.68
(2H, m); 13C NMR (101 MHz, CDCl3) δ 175.6, 139.5, 132.7,
128.4, 128.3, 61.9, 52.5, 44.1, 42.4; HRMS calcd for
C12H17ClNO3 [MNH4]+ 258.0897, found 258.0894.
Methyl 1-(4-Chlorophenyl)-3-oxocyclobutanecarbox-
ylate (7). An aq KBr solution was prepared by dissolving KBr
(1.7 kg, 14.2 mol, 0.22 equiv) in water (15.6 kg). An aq
Na2S2O3 quench solution was prepared by dissolving Na2S2O3
(8.0 kg, 50.6 mol, 0.78 equiv) in water (51.5 kg). An aq NaOCl
solution was prepared by dissolving NaOCl (106.7 kg of a 10%
w/w solution in water, 144.0 mol, 2.2 equiv) and KHCO3 (81.9
kg, 818 mol, 12.6 equiv) in water (330 kg). A 1000 L glass-lined
reactor was charged with CH2Cl2 (211 kg) and the crude
solution of ester 23 from the preceding step (23.65 kg, 66.2 wt
% purity by HPLC analysis, 65 mol, 1.0 equiv) at rt. The
solution was cooled to 0−5 °C, and the aq KBr solution was
added below 5 °C. After a further 15 min, TEMPO (0.94 kg,
5.4 mol, 0.08 equiv) was added. After a further 30 min, the aq
NaOCl quench solution was added below 5 °C. After a further
30 min (HPLC analysis showed complete conversion of ester
starting material 23 by this point), the aq Na2S2O3 solution was
added below 5 °C. The mixture was then warmed to 23−28 °C
and aged for 40 min at this temperature. The layers were
separated, and the aqueous layer was extracted with CH2Cl2 (2
× 105 kg). The combined organic layers were washed with
water (1 × 105 kg) and concentrated to dryness under vacuum
below 20 °C. THF (47 kg) was then added, and the mixture
was concentrated to dryness under vacuum again to give the
title compound (18.05 kg, 87.7 wt % purity, 66.3 mol, ∼100%
assay yield) as a crude oil, which was diluted with PhMe (15.2
kg) for use directly in the next step.
1
1121, 1037, 849, 838, 831, 602 cm−1; H NMR (400 MHz,
CDCl3) δ 8.63 (1H, s), 7.66 (1H, d, J = 9.5 Hz), 7.53 (1H, d, J
= 9.5 Hz), 7.22−7.09 (5H, m), 7.05 (1H, t, J = 52.0 Hz); 13C
NMR (101 MHz, CDCl3) δ 150.4, 150.3, 142.7 (t, J = 28.5
Hz), 138.8, 132.2, 131.0 (t, J = 8.0 Hz), 131.0, 130.1, 129.9,
129.2 (2C), 127.2, 119.6, 118.2 (q, J = 321.0 Hz), 110.0 (t, J =
237.0 Hz); 19F NMR (377 MHz, CDCl3) δ −73.6, −115.7;
HRMS calcd for C17H10F5N4O3S [MH]+ 445.0394, found
445.0404.
cis-1-(4-Chlorophenyl)-3-hydroxycyclobutanecarbox-
ylic Acid (21). (4-Chlorophenyl)acetic acid (20; 21.0 kg, 123
mol, 1.0 equiv) and THF (110 kg) were charged into a 1000 L
glass-lined reactor, and the solution was concentrated under
vacuum below 45 °C to a final volume of ∼55 L, and a KF spec
of 200 ppm was met. The resulting solution was transferred
into a clean drum. i-PrMgCl (134.4 kg of a 2 M solution in
THF, 276 mol, 2.24 equiv) was charged into the reactor at rt,
and then the THF solution of starting material 20 was added
below 40 °C. The mixture was stirred at 30−40 °C for 70 min,
then epichlorohydrin (20.5 kg, 221 mol, 1.8 equiv) was added
at 35−40 °C, and the resulting mixture was aged at this
temperature for a further 3.5 h. The mixture was cooled to 20−
25 °C, and additional i-PrMgCl (122.1 kg of a 2 M solution in
THF, 250 mol, 2.03 equiv) was added below 32 °C. The
resulting mixture was stirred at 25−32 °C for 18 h and then was
cooled to 0−5 °C and quenched by the addition of 5 N aq HCl
(120 kg) below 40 °C. The resulting biphasic mixture was
cooled to 23−28 °C (pH of aqueous layer was 2) and aged for
30 min. The layers were then separated, and the organic layer
was washed with 25% w/v aq NaCl solution (1 × 79 kg) and
then concentrated under vacuum below 45 °C to a volume of
∼100 L. The resulting solution was solvent-switched into PhMe
under vacuum below 45 °C until a final volume of ∼100 L and
residual THF and water levels of 0.6% and 0.1%, respectively,
were achieved (target specifications: THF ≤1.0%, water by KF
analysis ≤0.3%). The resulting slurry was cooled to 10−15 °C,
aged for 1 h, and then filtered using a centrifuge. The product
cake was washed with PhMe (32 L) and dried under vacuum at
55−60 °C for 10.5 h to give the title compound (16.95 kg, 96.8
wt % purity, 72.4 mol, 59% corrected yield) as a white solid. mp
181−182 °C; IR (neat) 3316, 2948, 1697, 1261, 1240, 1129,
Data for an authentic sample of ketone 7 purified by column
chromatography on silica gel: white solid; mp 46−47 °C; IR
(neat) 2934, 1773, 1725, 1218, 1125, 1088, 821, 761 cm−1; 1H
NMR (400 MHz, CDCl3) δ 7.38−7.35 (2H, m), 7.32−7.29
(2H, m), 3.95−3.88 (2H, m), 3.71 (3H, s), 3.57−3.50 (2H, m);
13C NMR (101 MHz, CDCl3) δ 202.1, 174.1, 139.0, 133.6,
128.8, 128.6, 57.6, 53.0, 42.9; HRMS calcd for C12H15ClNO3
[MNH4]+ 256.0740, found 256.0738.
1
819, 629 cm−1; H NMR (400 MHz, DMSO-d6) δ 11.05 (1H,
br s), 6.03 (4H, s), 3.84 (1H, br s), 1.40−1.35 (2H, m), 1.18−
1.13 (2H, m); 13C NMR (101 MHz, DMSO-d6) δ 176.0, 141.1,
131.3, 128.9, 128.3, 60.2, 43.6, 42.5; HRMS calcd for
C11H15ClNO3 [MNH4]+ 244.0740, found 244.0734.
Methyl cis-1-(4-Chlorophenyl)-3-hydroxycyclobutane-
carboxylate (23). To a 500 L glass-lined vessel were charged
acid 21 (16.8 kg, 72 mol, 1.0 equiv) and MeOH (72 kg), and
trans-1-(4-Chlorophenyl)-3-hydroxy-3-methylcyclo-
butanecarboxylic Acid (27). a. MeMgBr Addition. A 1000
L glass-lined reactor was charged with PhMe (168 kg) and the
1077
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
crude solution of ketone 7 from the preceding step (18.05 kg,
87.7 wt % purity by HPLC analysis, 66.3 mol, 1.0 equiv) at rt.
The mixture was concentrated under vacuum below 50 °C,
azeotroping with additional PhMe (170 kg), until a final
volume of ∼70 L was reached (residual THF and water levels at
this point were 0.001% and 0.02%, respectively). THF (6.9 kg)
was added, and the solution was cooled to between −60 and
−70 °C. MeMgBr (25.85 kg of a 3.0 M solution in diethyl
ether, 76.3 mmol, 1.15 equiv) was added between −60 and −70
°C, and the mixture was stirred for a further 3 h at that
temperature, by which point HPLC analysis showed 98%
conversion of ketone starting material 7. The solution was
warmed to 20−30 °C and aged for 2 h, after which time the
degree of conversion had not increased. The solution was
cooled to 0−5 °C and quenched by the addition of MeOH (5.0
kg) below 10 °C. The quenched reaction mixture was
transferred to a 500 L glass-lined reactor, followed by a rinse
of the 1000 L reactor with PhMe (47 kg). To the combined
solutions was added 2 N aq HCl (50 kg) below 10 °C. The
resulting biphasic mixture was warmed to 20−30 °C and aged
for 30 min, and then the layers were separated. The organic
layer was washed with water (1 × 60 kg) and then concentrated
under vacuum below 50 °C to a volume of ∼40 L. The
resulting solution was solvent-switched into THF (a total of
323 kg THF was used) under vacuum below 50 °C and
concentrated to a final volume of ∼100 L (residual PhMe at
this point was 5.9%; target specification ≤10%). HPLC analysis
of this solution indicated a 2.7:1 ratio of trans-ester 24/cis-ester
25.
35% aq HCl (75 kg) was added, and the mixture was stirred for
2.5 h. IPAc (156 kg) was then added, and the layers were
separated. The organic layer was concentrated under vacuum
below 50 °C to a final volume of ∼70 L, to afford the title
compound 27 (66.9 kg, 10.9 wt % purity, 30.2 mol, 49%
corrected overall yield) as a crude solution in IPAc, which was
used directly in the next step.
Data for an authentic sample of trans-acid 27 purified by
column chromatography on silica gel: off-white solid; mp 136−
139 °C; IR (neat) 3404, 2945, 1693, 1244, 1132, 1092, 945,
1
819, 630 cm−1; H NMR (400 MHz, DMSO-d6) δ 12.56 (1H,
br s), 7.40−7.37 (2H, m), 7.28−7.24 (2H, m), 5.06 (1H, br s),
2.85 (2H, dd, J = 10.5, 2.5 Hz), 2.42 (2H, dd, J = 10.5, 2.5 Hz),
1.25 (3H, s); 13C NMR (101 MHz, DMSO-d6) δ 176.2, 143.7,
130.7, 127.9, 127.8, 66.5, 47.7, 42.5, 27.5; HRMS calcd for
C12H17ClNO3 [MNH4]+ 258.0897, found 258.0897.
tert-Butyl [trans-1-(4-Chlorophenyl)-3-hydroxy-3-
methylcyclobutyl]carbamate (34). a. Curtius Rearrange-
ment of Acid 27. To a 500 L glass-lined vessel were charged
the IPAc solution containing crude trans-acid 27 (66.9 kg, 10.9
wt % purity by HPLC analysis, 30.2 mol, 1.0 equiv) and THF
(131 kg) at rt. The resulting mixture was solvent-switched into
THF by azeotropic distillation under vacuum below 50 °C until
the residual IPAc and water levels were ≤5% and ≤0.1%,
respectively. The resulting solution (∼40 L total volume) was
diluted with additional THF (70 kg), and Et3N (7.4 kg, 73.1
mol, 2.4 equiv) was added. DPPA (7.9 kg, 28.7 mol, 0.95 equiv)
was added below 30 °C, and the reaction vessel was then rinsed
down with THF (7 kg). The mixture was aged at 25−35 °C, at
which point HPLC analysis showed 85% conversion of acid
starting material 27 to acyl azide intermediate 31. An additional
portion of DPPA (0.42 kg, 1.5 mol, 0.05 equiv) was added
below 30 °C, and the mixture was aged for a further 3 h. The
mixture was then heated to 55−65 °C, and 2 N aq HCl (72 kg)
was added at this temperature over 4 h. After a further 11.5 h at
55−65 °C, HPLC analysis showed the formation of desired
amine 32 and urea 33 in a ratio of 1.3:1. The mixture was
concentrated under vacuum below 50 °C to a target volume of
∼75 L to remove THF (residual level at this point 0.04%; target
specification ≤5%), then IPAc (70 kg) was added, and the
biphasic mixture was stirred at 40 °C for 1 h before being
cooled to 25−35 °C. The layers were separated. The organic
layer, containing urea 33, was washed with 2 N aq HCl (2 × 30
kg) and processed further as described in section b below. The
combined aqueous layers contained amine 32 (as the
corresponding HCl salt). These aqueous layers were processed
through the Boc protection step, using a procedure analogous
to that described in section c below, to provide 900 g of crude
carbamate 34, which was recrystallized as described in section d
below.
Data for an authentic sample of trans-ester 24 purified by
column chromatography on silica gel: colorless oil; IR (neat)
3397, 2951, 1727, 1250, 1125, 1090, 731 cm−1; 1H NMR (400
MHz, CDCl3) δ 7.17 (2H, d, J = 8.5 Hz), 7.07 (2H, d, J = 8.5
Hz), 3.78 (1H, br s), 3.49 (3H, s), 2.87 (2H, dd, J = 10.5, 2.5
Hz), 2.40 (2H, d, J = 10.5 Hz), 1.21 (3H, s); 13C NMR (101
MHz, CDCl3) δ 175.4, 142.1, 132.2, 128.1, 127.5, 67.7, 52.3,
47.4, 42.7, 27.2; HRMS calcd for C13H19ClNO3 [MNH4]+
272.1053, found 272.1048.
b. Hydrolysis and Removal of Undesired cis-Ester 25. To
this THF solution were added LiOH·H2O (0.548 kg, 13.1 mol,
0.2 equiv) and water (70 kg) at rt. The resulting mixture was
periodically analyzed by HPLC to monitor the conversion of
cis-ester 25 to cis-acid 26, as measured by the corresponding
increase in the ratio of trans-ester 24/remaining cis-ester 25.
After 2 h the ratio had increased from 2.7:1 to 3.8:1. A total of
0.905 kg (21.6 mol, 0.33 equiv) additional LiOH·H2O was then
added in 4 portions over 6 h. After a further 2 h the ratio of
trans-ester 24/remaining cis-ester 25 was 143:1, and MTBE
(168 kg) was added to the reaction to give a biphasic mixture.
The layers were separated, and the organic layer was washed
with 25% w/v aq NaCl (2 × 78 kg) and then concentrated
under vacuum below 50 °C to a target volume of ∼40 L.
c. Hydrolysis of Desired trans-Ester 24. To the resulting
solution was added EtOH (150 kg), water (70 kg), and NaOH
(13.2 kg, 330.0 mol, 5 equiv) at rt, and the mixture was stirred
for 3 h. The mixture was then concentrated under vacuum
below 50 °C to a target volume of ∼70 L, and solvent-switched
under vacuum below 50 °C from EtOH into ∼100 L of water
(achieved 0.02% residual EtOH, target specification <5%
EtOH). The resulting solution was diluted with water (32
kg) and MTBE (48 kg). The layers were separated, and the
aqueous layer was washed with MTBE (1 × 76 kg). The
resulting aqueous layer was diluted with water (47 kg), then
b. Hydrolysis of Urea 33. To the IPAc organic layer
containing urea 33 was charged 30% aq NaOH (26 kg, 195
mol) and water (40 kg) at rt. The mixture was stirred for 2.5 h
(to remove residual HCl and DPPA), and then the layers were
separated. The organic layer was concentrated under vacuum
below 50 °C to a target volume of 50−60 L, during which time
urea 33 crystallized. The resulting slurry was filtered and the
product cake dried to give 2.3 kg (5.1 mol uncorrected) of
crude urea 33 in 92 HPLC area % purity. To a 500 L glass-lined
vessel were charged the crude urea 33 (2.3 kg), water (25 kg),
diethylene glycol monoethyl ether (25 kg), and Ba(OH)2 (4.5
kg, 26.3 mol). The mixture was heated to 98−103 °C and aged
for 7 days. After cooling to 20−30 °C, water (25 kg) and
1078
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
MTBE (25 kg) were added, and stirring was continued for 30
min. The layers were separated, and the aqueous layer was
back-extracted with MTBE (2 × 25 kg). The combined organic
layers were then extracted with 1 M aq HCl (2 × 12 L). The
combined aqueous layers contained amine 32 as the
corresponding HCl salt (1.64 kg free-base amine by HPLC
assay, 7.75 mol) and were processed further as described in
section c below.
c. Boc Protection of Amine 32. The aqueous layer from
section b above, containing the amine 32 HCl salt (1.64 kg
free-base amine by HPLC assay, 7.75 mol, 1.0 equiv), was
adjusted to pH 7 using aq NaOH. Na2CO3 (0.5 kg, 4.7 mol, 0.6
equiv), THF (15 kg), and Boc2O (2.8 kg, 12.8 mol, 1.7 equiv)
were added, and then the mixture was heated to 60−65 °C and
aged at that temperature for 2 h. After cooling to 20−30 °C the
layers were separated and the aqueous layer was extracted with
MTBE (2 × 15 L). The combined organic layers were
concentrated under vacuum to a target volume of 10 L.
Heptane (20 L) was added, and the resulting solution was
concentrated under vacuum to a target volume of 10 L. To this
solution was added heptane (30 L), and the resulting slurry was
cooled to 15−20 °C and aged for 3 h. The slurry was filtered,
and the product cake was dried under vacuum at rt to give
crude carbamate 34 (2.3 kg, 7.4 mol uncorrected).
≤5%), during which time the product began to crystallize. The
slurry was aged at 20−30 °C for 30 min and then filtered. The
product cake was washed with n-heptane (1 × 8.6 kg) and then
dried under vacuum at rt for 24 h to give the title compound
(3.4 kg, 99.0 wt % purity, 8.4 mol, 91% corrected yield) as a
white solid. mp 112−115 °C; IR (neat) 3532, 3338, 2980,
1694, 1363, 1164, 1142, 1092, 655 cm−1; 1H NMR (400 MHz,
CDCl3) δ 7.77 (2H, d, J = 8.0 Hz), 7.40 (2H, d, J = 8.0 Hz),
5.22−5.04 (1H, br m), 2.67−2.54 (4H, br m), 1.90 (2H, s),
1.57 (3H, s), 1.38−1.26 (21H, br m); 13C NMR (101 MHz,
CDCl3) δ 154.2, 149.3, 134.8, 126.9, 124.8, 83.7, 79.5, 68.5,
51.1, 49.1, 29.4, 28.3, 24.8; HRMS calcd for C22H35BNO5
[MH]+ 403.2645, found 403.2651.
tert-Butyl (trans-1-{4-[1-(Difluoromethyl)-8-phenyl-
[1,2,4]triazolo[4,3-a][1,5]naphthyridin-7-yl]phenyl}-3-
hydroxy-3-methylcyclobutyl)carbamate (36). An aq
K3PO4 solution was prepared by dissolving K3PO4 (3.01 kg,
14.18 mol, 2.0 equiv) in water (2.52 L) at rt, followed by
degassing via subsurface N2 sparging for 20 min. A 50 L
jacketed cylindrical vessel was charged with 2-MeTHF (12.6 L),
triflate 4 (3.15 kg, 7.09 mol, 1.0 equiv), and boronate 6 (3.15
kg, 7.80 mol, 1.1 equiv), and the resulting thin brown slurry was
degassed for 20 min. To the vessel was then charged Pd(OAc)2
(48 g, 0. 21 mol, 0. 03 equiv) and 1, 1′-bis-
(diisopropylphosphino)ferrocene (DIPPF) (89 g, 0.21 mol,
0.03 equiv), and the mixture was degassed for a further 10 min.
The aq K3PO4 solution was then added, and the degassing
continued for a further 5 min. The mixture was aged for a
further 1 h, during which time the batch temperature rose
steadily from 25 °C to a maximum of 48 °C. After 1 h the
reaction was complete (<0.3% triflate 4 remained). The brown
solution was cooled to 20 °C over 35 min, during which time
the product began to crystallize. After stirring for 1 h at 20 °C,
the liquors were assayed for 31 mg/mL of product 36. Water
(3.78 L) was added in one portion, followed by the addition of
MTBE (12.6 L) over 1 h. The resulting yellow slurry was aged
for a further 30 min (liquors assayed for 18 mg/mL of product
36 at this point) and then filtered. The product cake was
washed with 1:1 2-MeTHF/MTBE (2 × 8 L) and then with 1:1
EtOH/water (2 × 8 L), and it was then dried under vacuum
with a N2 sweep at rt for 4 days to give the title compound
(3.66 kg, 93 wt % puritycontains residual EtOH and 2-
MeTHF, 5.96 mol, 84% corrected yield) as an off-white solid.
mp 147−148.5 °C; IR (neat) 3388, 2968, 1726, 1512, 1164,
d. Recrystallization of Crude Carbamate 34. The 0.9 kg of
crude carbamate from step a was combined with the 2.3 kg of
crude carbamate from step c and recrystallized from acetone/
water to give the title compound 34 (2.87 kg, 99.3 wt % purity,
9.2 mol, 30% overall yield from acid 27) as a white solid.
Data for recrystallized carbamate 34: mp 144−145 °C; IR
(neat) 3275, 2979, 1675, 1364, 1238, 1166, 1075, 1006, 626
1
cm−1; H NMR (400 MHz, CDCl3) δ 7.34−7.27 (4H, m),
5.51−5.05 (1H, br m), 2.66−2.54 (4H, br m), 2.05 (1H, s),
1.56 (3H, s), 1.39−1.28 (9H, br m); 13C NMR (101 MHz,
CDCl3) δ 154.2, 144.8, 132.2, 128.2, 127.0, 79.7, 68.4, 50.6,
49.3, 29.5, 28.3; HRMS calcd for C16H23ClNO3 [MH]+
312.1366, found 312.1362.
tert-Butyl {trans-3-Hydroxy-3-methyl-1-[4-(4,4,5,5-tet-
ramethyl-1,3,2-dioxaborolan-2-yl)phenyl]cyclobutyl}-
carbamate (6). A 100 L glass-lined vessel was charged with
carbamate 34 (2.87 kg, 99.3 wt % purity by HPLC analysis, 9.2
mol., 1.0 equiv), KOAc (2.7 kg, 27.5 mol, 3.0 equiv),
bis(pinacolato)diboron ((Bpin)2) (2.85 kg, 11.2 mol, 1.2
equiv), and MeCN (34.4 kg) at rt. After 20 min, X-Phos
(268 g, 0.56 mol, 0.06 equiv) and Pd(OAc)2 (68 g, 0.28 mol,
0.03 equiv) were added, and the resulting mixture was purged
with N2 (vacuum/backfill) 3 times. The mixture was heated to
78−83 °C and aged for 18 h, at which point HPLC analysis
indicated no detectable carbamate starting material 34 (target
specification: ≥99% conversion). The mixture was cooled to rt
and filtered through a pad (600 g) of Celite, rinsing with
MeCN (16.8 kg). The combined filtrates were concentrated
under vacuum below 50 °C and solvent-switched into a target
final volume of MTBE solution by azeotropic distillation
(achieved 1.5% residual MeCN; target specification ≤5%). The
resulting solution was washed with water (3 × 30 kg) to
remove pinacol and tert-butanol (t-BuOH) byproduct
(achieved 0.3% residual pinacol and 0.2% residual t-BuOH;
target specifications ≤5% for each). The organic layer was dried
with Na2SO4 (1.0 kg) for 1 h at rt and then filtered. The filtrate
was solvent-switched into n-heptane by azeotropic distillation
under reduced pressure below 50 °C to a final target volume of
∼30 L (achieved 0.02% residual MTBE; target specification
1
1063, 825, 784, 697 cm−1; H NMR (500 MHz, DMSO-d6) δ
8.52 (1H, s), 8.14 (1H, d, J = 10.0 Hz), 8.06 (1H, t, J = 51.5
Hz), 8.04 (1H, d, J = 10.0 Hz), 7.52 (1H, s), 7.40−7.29 (9H,
m), 4.97 (1H, s), 2.60 (2H, d, J = 12.5 Hz), 2.35 (2H, d, J =
12.5 Hz), 1.37−1.35 (9H, m), 1.15 (3H, s); 13C NMR (126
MHz, DMSO-d6) δ 156.0, 154.3, 150.1, 148.2, 142.8 (t, J = 28.0
Hz), 139.1, 138.4, 136.0 (2C), 132.3, 129.3, 129.2, 128.7, 128.2,
127.3 (t, J = 5.0 Hz), 125.8, 125.0, 118.4, 109.8 (t, J = 236.5
Hz), 77.6, 66.5, 50.0, 49.3, 29.2, 28.3; 19F NMR (471 MHz,
DMSO-d6) δ −116.9; HRMS calcd for C32H32F2N5O3 [MH]+
572.2473, found 572.2486.
trans-3-Amino-3-{4-[1-(difluoromethyl)-8-phenyl-
[1,2,4]triazolo[4,3-a][1,5]naphthyridin-7-yl]phenyl}-1-
methylcyclobutanol (1). A 100 L jacketed cylindrical vessel
was charged with EtOH (35.9 L) and carbamate 36 (3.59 kg,
93 wt % purity, 5.84 mol, 1.0 equiv). The resulting off-white
slurry was cooled to 17 °C over 15 min, and 5 N aq HCl (5.84
L, 29.2 mol, 5.0 equiv) was added in portions over 5 min below
25 °C. The reaction was then heated to 50 °C over 1 h and
1079
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
A.; Ripin, D. H. B.; Singer, R. A.; Tucker, J. L.; Wei, L. Org. Process Res.
Dev. 2003, 7, 873−878.
aged at this temperature for 24 h until <1.0% of starting
material 36 remained (measured by HPLC analysis). The off-
white slurry was then allowed to cool to rt over 12 h and then
diluted with water (27.14 L). The mixture was cooled further to
15 °C over 15 min, and 10 N NaOH solution was added over
45 min below 25 °C until a pH of 10−11 was reached (∼3.0 L
NaOH was added). The faint yellow slurry was aged for a
further 30 min at rt and then filtered. The product cake was
washed with 2:1 water/EtOH (2 × 8 L) and then with EtOH
(2 × 8 L), and it was then dried under vacuum with a N2 sweep
at rt for 40 h to give the title compound (2.57 kg, 99.3 wt %
purity, 5.42 mol, 93% corrected yield) as an off-white fluffy
solid. mp 260.5−261 °C (dec); IR (neat) 3336, 2961, 2921,
(9) Bromide 10 is commercially available, but a long lead time on
multikilogram scale precluded its use as the starting material in the
presently described campaign.
(10) Cheung, M.; Eidam, H. S.; Goodman, K. B.; Hilfiker, M. A. PCT
Int. Appl. WO/2011/119694, 2011.
(11) Kelly, T. A.; McNeil, D. W. Tetrahedron Lett. 1994, 35, 9003−
9006.
(12) Solution assay yields and purity of isolated products reported in
the Results and Discussion and Experimental Section were determined
by HPLC analysis, unless stated otherwise.
(13) Formylation of the des-chloro analogue of bromide 11 using n-
BuLi/N-formylpiperidine in THF has been reported to proceed in
90% yield, but these conditions could not be successfully scaled up
with bromide 11; see: Venuti, M. C.; Stephenson, R. A.; Alvarez, R.;
Bruno, J. J.; Strosberg, A. M. J. Med. Chem. 1988, 31, 2136−2145.
(14) n-BuLi could also be used, but n-HexLi was preferred as a more
easily handled reagent on scale for batch processing.
(15) For selected examples of organolithium chemistry in flow, see:
(a) Browne, D. L.; Baumann, M.; Harji, B. H.; Baxendale, I. R.; Ley, S.
V. Org. Lett. 2011, 13, 3312−3315. (b) Shu, W.; Pellegatti, L.; Oberli,
M. A.; Buchwald, S. L. Angew. Chem., Int. Ed. 2011, 50, 10665−10669.
(c) Nagaki, A.; Kenmoku, A.; Moriwaki, Y.; Hayashi, A.; Yoshida, J.
Angew. Chem., Int. Ed. 2010, 49, 7543−7547. (d) Gross, T. D.; Chou,
S.; Bonneville, D.; Gross, R. S.; Wang, P.; Campopiano, O.; Ouellette,
M. A.; Zook, S. E.; Reddy, J. P.; Moree, W. J.; Jovic, F.; Chopade, S.
Org. Process Res. Dev. 2008, 12, 929−939.
1
1619, 1516, 1133, 1113, 1039, 882, 821, 772, 793 cm−1; H
NMR (500 MHz, DMSO-d6) δ 8.51 (1H, s), 8.14 (1H, d, J =
10.0 Hz), 8.06 (1H, t, J = 51.5 Hz), 8.04 (1H, d, J = 10.0 Hz),
7.45−7.41 (3H, m), 7.40−7.33 (6H, m), 4.77 (1H, s), 2.37
(2H, dd, J = 10.5, 2.5 Hz), 2.16 (2H, dd, J = 10.5, 2.5 Hz), 1.93
(2H, s), 1.51 (3H, s); 13C NMR (126 MHz, DMSO-d6) δ
156.0, 152.5, 150.1, 142.8 (t, J = 28.0 Hz), 139.9, 138.5, 136.0,
135.8, 132.3, 129.4, 129.3, 128.8, 128.2, 127.3 (t, J = 6.0 Hz),
125.7, 124.6, 118.3, 109.8 (t, J = 235.5 Hz), 66.5, 51.8, 49.3,
29.8; 19F NMR (471 MHz, DMSO-d6) δ −116.9; HRMS calcd
for C27H24F2N5O [MH]+ 472.1949, found 472.1948.
ASSOCIATED CONTENT
■
(16) Lithium-halogen exchange of bromopyridines in flow micro-
reactors without the need for cryogenic conditions has recently been
reported; see: Nagaki, A.; Yamada, S.; Doi, M.; Tomida, Y.; Takabaya,
N.; Yoshida, J. Green Chem. 2011, 13, 1110−1113.
S
* Supporting Information
General experimental information, and copies of H, 13C, and
1
19F NMR spectra. This material is available free of charge via
the Internet at http://pubs.acs.org.
(17) Project timeline constraints meant that extension of the flow
process to include the generation of anion 13 could not be extensively
investigated. Initial attempts to generate dianion 14 directly from
bromide 11 in flow using 2 equiv of n-BuLi were low yielding.
(18) Structures of intermediates 15 and 16 tentatively assigned based
AUTHOR INFORMATION
Corresponding Author
*E-mail: pintipa.grongsaard@merck.com.
■
1
on HPLC-MS and H NMR analysis of reaction mixtures.
Notes
(19) There was one major impurity (8 HPLC area %), which
appeared to be a dimeric species by HPLC-MS analysis, in the liquors
together with numerous other unidentified lower-level impurities.
(20) The 652 psi of pressure included 102 psi from the vapor
pressure of water. With the assumption that the sole source of the
remaining 550 psi of gas (equal to about 0.077 mol based on the ideal
gas law) is hydrazine, then 0.48 mol of gas were generated per mole of
hydrazine.
(21) For the process described in this paper, a 5-gal Hastelloy C
autoclave rated for 5000 psig was used, fitted with a pressure-venting
rupture disk rated for 3100 psig. As described in the Experimental
Section, prior to running the main batches, the autoclave was
thoroughly cleaned and then conditioned with 1% aq hydrazine.
(22) For subsequent hydrazine displacement reactions, these
calculations must be repeated and, if necessary, additional safety
evaluation conducted, prior to making any changes to the equipment
being used and/or the proposed scale of the process.
(23) Subsequent lab experiments demonstrated that if the product
triazole 19 initially isolated from the reaction is washed more
thoroughly with organic solvent, then a higher weight % purity is
obtained and the reslurry is not required.
(24) Details of the development of this reaction, and the origin of the
high cis-diastereoselectivity, will be described in a forthcoming
manuscript from these laboratories.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
We thank Thomas J. Novak for assistance with HRMS analysis.
■
REFERENCES
■
(1) Jemal, A.; Bray, F.; Center, M. M.; Ferlay, J.; Ward, E.; Forman,
D. Ca−Cancer J. Clin. 2011, 61, 69−90.
(2) Mattmann, M. E.; Stoops, S. L.; Lindsley, C. W. Expert. Opin.
Ther. Pat. 2011, 21, 1309−1338.
(3) Lindsley, C. W. Curr. Top. Med. Chem. 2010, 10, 458−477.
(4) Pal, S. K.; Reckamp, K.; Yu, H.; Figlin, R. A. Expert. Opin. Invest.
Drugs 2010, 19, 1355−1366.
(5) Yun, J. Cancer Biol. Ther. 2010, 9, 504−506.
(6) Furuyama, H.; Goto, Y.; Kawanishi, N.; Layton, M. E.; Mita, T.;
Ogino, Y.; Onozaki, Y.; Rossi, M.; Sakamoto, T.; Sanderson, P. E.;
Wang, J. PCT Int. Appl. WO/2009/14887, 2009.
(7) As monitored by 1H NMR analysis of the reaction, no C-4
monobrominated product was observed. At the end of the reaction,
1% dibrominated product had been formed and 4% unreacted starting
material remained, but these impurities were efficiently rejected during
the crystallization.
(8) For examples of the regioselective bromination of 3-amino-6-
substituted pyridines, see: (a) Aldous, S. C.; Fennie, M. W.; Jiang, J. Z.;
John, S.; Mu, L.; Pedgrift, B.; Pribish, J. R.; Rauckman, B.; Sabol, J. S.;
Stoklosa, G. T.; Thurairatnam, S.; Van Deusen, C. L. PCT Int. Appl.
WO/2008/121670, 2008. (b) Beaudin, J.; Bourassa, D. E.; Bowles, P.;
Castaldi, M. J.; Clay, R.; Couturier, M. A.; Karrick, G.; Makowski, T.
W.; McDermott, R. E.; Meltz, C. N.; Meltz, M.; Phillips, J. E.; Ragan, J.
(25) For discussion of high-throughput experimentation and selected
examples, see: (a) Belyk, K. M.; Xiang, B.; Bulger, P. G.; Leonard, W.
R.; Balsells, J.; Yin, J.; Chen, C. Org. Process Res. Dev. 2010, 14, 692−
700. (b) Davies, I. W.; Welch, C. J. Science 2009, 325, 701−704.
(c) Dreher, S. D.; Dormer, P. G.; Sandrock, D.; Molander, G. A. J. Am.
Chem. Soc. 2008, 130, 9257−9259. (d) Rubin, A. E.; Tummala, S.;
1080
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081

Organic Process Research & Development
Article
Both, D. A.; Wang, C.; Delaney, E. J. Chem. Rev. 2006, 106, 2794−
2810.
(26) For the example of Curtius rearrangement plus carbamate
formation on scale, see: Yue, T.-Y.; McLeod, D. D.; Albertson, K. B.;
Beck, S. R.; Deerberg, J.; Fortunak, J. M.; Nugent, W. A.; Radesca, L.
A.; Tang, L.; Xiang, C. D. Org. Process Res. Dev. 2006, 10, 262−271.
(27) Target residual Pd specification in the final API 1 for this
delivery was 50 ppm.
(28) The methane generated was vented with the stream of N2,
which is sufficient for this scale of synthesis to ensure there is no
flammable atmosphere in the reactor or vent piping.
(29) Hastelloy C276 was used in the autoclave. Hastelloy C is
typically more robust than 316 Stainless Steel and is compatible with
hydrazine. For discussion on a high temperature experiment with 8.5%
aq hydrazine in 316 Stainless Steel, see: Brubaker, G. R.; Geoffrey, M.
M. Ind. Eng. Chem. Res. 1988, 27, 1149−1152.
1081
dx.doi.org/10.1021/op300031r | Org. Process Res. Dev. 2012, 16, 1069−1081