Asymmetric synthesis of letermovir using a novel phase-Transfer-catalyzed aza-michael reaction

Article abstract of DOI:10.1021/acs.oprd.6b00076

The development of a concise asymmetric synthesis of the antiviral development candidate letermovir is reported, proceeding in <60% yield over a total of seven steps from commercially available materials. Key to the effectiveness of this process is a nove

Full text of DOI:10.1021/acs.oprd.6b00076

This is an open access article published under an ACS AuthorChoice License, which permits
copying and redistribution of the article or any adaptations for non-commercial purposes.
Communication
pubs.acs.org/OPRD
Asymmetric Synthesis of Letermovir Using a Novel Phase-Transfer-
Catalyzed Aza-Michael Reaction
Guy R. Humphrey,* Stephen M. Dalby,* Teresa Andreani, Bangping Xiang, Michael R. Luzung,†
Zhiguo Jake Song, Michael Shevlin, Melodie Christensen, Kevin M. Belyk, and David M. Tschaen
Department of Process Chemistry, Merck and Co., Inc., P.O. Box 2000, Rahway, New Jersey 07065, United States
S
* Supporting Information
orphan drug status in 2011 and has since progressed to PhIII
ABSTRACT: The development of a concise asymmetric
synthesis of the antiviral development candidate letermovir
is reported, proceeding in >60% yield over a total of seven
steps from commercially available materials. Key to the
effectiveness of this process is a novel cinchonidine-based
PTC-catalyzed aza-Michael reaction to configure the single
stereocenter.
clinical trials. Herein we describe the first enantioselective
synthesis of letermovir to support ongoing clinical trials and
ultimately a manufacturing process.
From a structural perspective, letermovir presents a
fluorinated dihydroquinazoline core functionalized with a
piperazine and a trifluoromethylanisole together with a
stereogenic C4-acetate. Retrosynthetically, this structure is
most straightforwardly distilled down to the commercially
accessible building blocks 2, 3, and 4, which may be assembled
to generate the central dihydroquinazoline ring (Figure 1). This
had previously been established, whereby base-mediated
cyclization of urea 5 provided racemic dihydroquinazolinone
6, which was then condensed with piperazine 4 and subject to a
classical resolution to provide enantioenriched material.3a,c
While this proved sufficient for initial supplies of letermovir, in
view of its racemic mode and significant raw material costs, we
sought to define a more efficient and cost-effective route,
appropriate for more sustainable long-term supply.
espite advances in modern antiviral treatment, human
Dcytomegalovirus (HCMV) remains associated with
serious morbidity and mortality in immunocompromised
patients, particularly those with advanced HIV and those
undergoing organ transplant, up to 75% of whom experience
HCMV infection or reactivation without appropriate antiviral
treatment.1 Existing therapies almost universally target DNA
polymerase and thus suffer from severe toxicity and increasingly
prevalent drug resistance.2 Novel therapeutic agents are
therefore urgently sought-after, combining efficacy with a
unique mode of action to circumvent cross-resistance with
available HCMV agents.
Letermovir (1, Figure 1) shows great promise in this respect
for the next-generation treatment of HCMV infection.3
Endowed with potent anti-HCMV activity (cell culture EC50
= ca. 5 nM) with no observed dose-dependent toxicity, it
functions uniquely through inhibition of the DNA terminase,
via a mechanism distinct from other compounds targeting this
enzyme complex.3,4 Letermovir was granted both fast track and
Central to this effort would be the development and
implementation of an asymmetric method to define the C4-
acetate stereocenter. Initial work focused on catalysis of the
urea cyclization reaction, and some success was realized
utilizing cinchona alkaloid-based phase transfer catalysis
(PTC, Scheme 1).5−7 Treatment of urea 5 in a biphasic
Scheme 1. PTC Aza-Michael Cyclization of Urea 5
Received: March 8, 2016
Figure 1. Retrosynthetic analysis of letermovir (1).
© XXXX American Chemical Society
A
DOI: 10.1021/acs.oprd.6b00076
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
Scheme 2. Metal-Catalyzed Asymmetric Cyclizations12
mixture of toluene and aqueous K3PO4 with 20 mol % PTC 7
led to the smooth formation of quinazolinone 6 with moderate
levels of stereoinduction (56% ee). Further screening failed to
elevate the enantioselectivity, however, and although selective
crystallization of the residual racemate did enable the isolation
of highly enantioenriched material (>98% ee), epimerization of
the newly formed stereocenter was observed in the subsequent
Vilsmeier-type condensation with piperazine 4. While this
could be alleviated by the use of TFE as solvent,8 the need for a
more fundamentally revised approach was evident.
Among the range of potential approaches considered, those
targeting catalytic asymmetric cyclization of guanidine sub-
strates (which might thus circumvent epimerization during
piperazine condensation) were deemed most attractive. As
shown in Figure 2, the principal approaches explored involved
which likely also contributes to the difficulties suppressing
debromination in the reductive Heck reaction.13 Despite these
issues, productive asymmetric hydrogenation of 13 to provide 8
could be achieved, after optimization with 76% ee and 92 A%.
Evaluation of the asymmetric PTC aza-Michael cyclization of
guanidine 11 commenced with the conditions developed for
the analogous cyclization of urea 5 (20 mol % PTC, aq. K3PO4,
PhMe, 0 °C). The success of this approach would rely in part
upon the susceptibility of guanidine 11 to background
cyclization, which was particularly prominent in alcoholic
solvents.14 Fortunately, this was suppressed using toluene, and
no background cyclization was observed in the absence of the
PTC. Pleasingly, mono-quaternized PTCs were generally found
to effect a smooth cyclization reaction with encouraging levels
of enantioselectivity (10−40% ee). The corresponding bis-
quaternized catalysts offered immediate enhancements in both
reaction rate and enantioselectivity, however.15 Screening of
this catalyst manifold (Table 1) revealed acute sensitivity to
both the quinoline (R1, 14j, 14k) and quinuclidine (R2, 14h,
14l) substituents, leading to the identification of catalyst 14f,
which provided dihydroquinazoline 8 in almost quantitative
yield and 76% ee after 3 h at 0 °C at 5 mol % loading. Notably,
a range of other catalyst types proved ineffectual by
comparison,16 highlighting the utility of these new PTCs for
asymmetric synthesis.
Unfortunately, defining structure−activity relationships to
guide further PTC refinement to access 8 in >90% ee proved
unsuccessful and will likely rely upon greater mechanistic
understanding. Several general features were apparent however:
electron-deficient aromatic groups afforded much improved
catalyst performance; CF3 substituents tended to be beneficial
(14c); 2-substituted quinuclidine substituents led to diminished
activity and even reversal of enantioselectivity (14d), and
cinchonidine-based catalysts were superior to the correspond-
ing quinidine, quinine, and cinchonine-based catalysts (14i,
Figure 2. Principal asymmetric approaches considered.
(a) amino-carbonylation;9 (b) Heck cyclization10/hydrogena-
tion; (c) reductive Heck reaction;11 (d) PTC aza-Michael
reaction, aiming to exploit the initial success with urea 5.
The amino-carbonylation substrate, styrenyl guanidine 9
(Scheme 2a),12 was exposed to a range of Pd-mediated
oxidative carbonylative conditions based on the work of Sasai.9
Further screening of chiral ligands and additives led to the
identification of the conditions shown in Scheme 2a, which
provided dihydroquinazoline 8 in 76 A% (HPLC area %) and
70% ee. Relatively high Pd-loadings (10 mol %) were required
to attain full conversion, however, even at elevated temperature
(18h, 70 °C). While an attractive approach, attaining the
necessary improvements for process viability was expected to
be highly challenging.
Preparation of the tetra-substituted guanidine 10 proved
more complicated than for 9,12 which was ultimately reflected
in its reactivity profile. Pleasingly, screening of conditions for
the asymmetric reductive Heck cyclization of 10 did identify a
reagent set under which the desired dihydroquinazoline 8 could
be prepared in up to 86% ee (Scheme 2b). Competitive
reductive debromination to form 12 proved prohibitively
problematic, however, and was unable to be suppressed beyond
a ca. 2:1 product ratio of 8:12. In response, the stepwise Heck
reaction/hydrogenation strategy was explored, which might
offer more synthetic control. Accordingly, Heck cyclization
delivered the exocyclic acrylate 13 (Scheme 2c), which was
isolated as an equilibrating ca. 3:1 mixture of E:Z isomers.
Acrylate 13 also exhibited a propensity for alkene hydration,
presumably to alleviate allylic strain with the adjacent anisole,
B
DOI: 10.1021/acs.oprd.6b00076
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
a
Table 1. Optimization of the Phase-Transfer-Catalyst
Scheme 3. Preparation of Guanidine Salicylate 11a
b
PTC
X
R1
R2
ee (%)
14a
14b
14c
14d
14e
14f
H
2-Br, 5-OMe
″
″
3-NO2
24
24
H
3,5-F
H
3,5-CF3
58
H
″
2-CF3
−13
51
H
3,5-NO2
3,5-CF3
H
3-CF3
″
76
14g
14h
14i
H
4-CF3
″
″
″
″
67
H
3,5-CF3
3,5-CF3
2-F, 5-OCF3
3-F, 6-OCF3
3,5-CF3
3-F, 5-CF3
74
OMe
H
<30
71
14j
14k
141
14m
H
″
<10
50
H
3-CF3
OMe
3-F, 5-CF3
45
a
b
Subset of >200 PTC catalysts screened. Enantioselectivity was
determined by chiral HPLC analysis of the crude reaction mixtures.
rate, where sharp drop-offs were observed below a critical point
(Figure 3a);19 (ii) the concentration and equivalents of
aqueous base, where superstoichiometric amounts of K3PO4
14m). The corresponding benzyl or t-butyl ester substrates
offered no selectivity benefits.
(1.5 eq, 0.4 M aqueous) proved optimal (Figure 3b); (iii)
3−
PTC/base counterions, where deviation from Br− or PO4
,
Contrasting the remarkable operational simplicity and
effectiveness of the PTC aza-Michael reaction against the
preparative, stability, and reactivity issues associated with
alternative approaches led to its selection as the cornerstone
of the new synthesis of letermovir, where refined PTCs may be
directly implemented as they become available.17 Accordingly,
an effective means for preparation of guanidine 11 was
developed and optimized (Scheme 3). This commenced with
Heck reaction of commercially available bromoaniline 2 to
provide acrylate 15 in 99% assay yield (AY), utilizing just 0.2
mol % Pd precatalyst.18 Acylation under Schotten−Baumann
conditions afforded phenyl carbamate 16, which underwent
smooth condensation with aniline 3 to provide urea 5, whose
high crystallinity enabled it to be isolated directly from the
reaction stream. In practice, this sequence could be efficiently
carried out as a single solvent through-process, proceeding in
87% yield over the three steps. Dehydration of urea 5 with PCl5
in PhMe at 40 °C cleanly provided the carbodiimide 17. This
intermediate proved remarkably stable with respect to
hydrolysis, yet underwent rapid reaction with piperazine 4a as
part of a second through-process. The targeted guanidine 11
was then conveniently isolated from the crude reaction stream
as the corresponding salicylate salt (11a) in 90% yield from
urea 5. The combination of carbodiimide stability and
piperazine nucleophilicity underpinned the success of this
approach, in contrast to other permutations of urea/aniline
condensation, while the use of toluene as a nonpolar solvent
proved critical to preserving the acyclic integrity of guanidine
11 (<1% cyclization observed).
respectively, proved detrimental.20 Relatively constant enantio-
selectivity was observed over the course of the reaction,
suggesting a single active catalytic species. PTC loading could
be reduced to 3 mol % with minimal effect (6 h, 72% ee).21
Having defined suitable operating parameters, the critical aza-
Michael reaction was successfully implemented on multikilo-
gram scale to deliver, after aqueous workup, a toluene solution
of dihydroquinazoline 8 (Scheme 4). Upgrading the optical
purity from 76 to >99% ee could be conveniently achieved
through formation of the chiral DTTA (di-p-toluoyl-(S,S)-
tartaric acid, 20) salt 8a.3a,c In practice, this was accomplished
directly from the crude organic stream following aqueous
workup, providing 8a in 82% overall yield and 99.6% ee22 and
critically sufficient purity in preparation for generating
amorphous drug substance. Accordingly, free-basing of DTTA
salt 8a followed by ester saponification (NaOH, 70 °C)
provided letermovir (1), which was finally isolated through
precipitation as an amorphous white solid (94%).
In summary we have developed a concise and efficient
asymmetric synthesis of letermovir, a potent and selective new
treatment for infections of HCMV. A novel asymmetric aza-
Michael cyclization reaction was devised and implemented for
this purpose, exploiting bis-quaternized PTC catalysis. The
entire synthesis proceeds in >60% yield over seven steps,
including five steps incorporated into two single solvent
through-processes, without need for chromatography, and has
already enabled the preparation of over 1 t of letermovir for
further clinical development.
Process development of the aza-Michael cyclization revealed
a number of intriguing features which might suggest an atypical
PTC-type mechanism is operative. Both reaction rate and
enantioselectivity were found to be sensitive to (i) agitation
EXPERIMENTAL SECTION
All reactions were carried out under a nitrogen atmosphere. All
solvents and reagents were purchased from commercial sources
■
C
DOI: 10.1021/acs.oprd.6b00076
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
Figure 3. Aza-Michael reaction agitation/base sensitivity.
acrylate 15 (15.9 kg, 8.2 wt % 15, 6.66 mol) was added H2O
(6.50 L) and Na2HPO4 (1.42 kg, 10.0 mol) followed by phenyl
chloroformate (1.04 L, 8.33 mol) dropwise over 30 min. The
reaction mixture was stirred at RT for 12 h before being heated
to 60 °C and stirred for a further 2 h. The mixture was then
diluted with i-PrOAc (7.00 L) and held at 60 °C until all solids
were dissolved. The aqueous phase was then separated and the
organics washed with H2O (2.60 L) before being cooled to RT
to afford a crude i-PrOAc slurry of carbamate 16 (ca. 95% assay
yield) which was used directly without further purification. For
the purposes of characterization, a sample of pure carbamate 16
was isolated by crystallization from a sample of the crude
solution diluted with heptane. NMR spectra were complicated
by persistent rotamerspeaks are reported as observed. Mp
Scheme 4. Completion of the Synthesis of Letermovir (1)
1
154−158 °C; H NMR (500 MHz, DMSO-d6) δH 9.97/9.57
(1H, br s), 7.98−7.80 (1H, m), 7.75 (1H, m), 7.50−7.32 (4H,
m), 7.29−7.02 (3H, m), 6.75 (1H, m), 6.54/6.44/5.63 (minor
rotamers), 3.80−3.70 (3H, s); 13C NMR (500 MHz, DMSO-
d6) δC 166.5, 158.0 (d, JCF = 247.6 Hz), 157.2, 152.8, 150.8,
139.6 (d, JCF = 3.7 Hz), 138.6 (d, JCF = 3.2 Hz), 133.4, 129.6,
129.5, 128.6, 125.6, 122.8 (d, JCF = 3.1 Hz), 121.6, 120.7, 118.9,
117.7 (d, JCF = 20.6 Hz), 116.7, 116.2 (d, JCF = 18.8 Hz), 115.7
(d, JCF = 7.6 Hz), 115.3, 51.8, 51.3; HR-MS calcd for
1
and were used without further purification. H and 13C NMR
chemical shifts were reported relative to residual proton solvent
peaks. Melting points were determined on TA Instruments
Density Scanning Calorimetry Q20. All yields are corrected for
purity and determined by reverse-phase HPLC assay using
purified standards. High-resolution mass spectra (HR-MS)
were recorded using Waters Acquity UPLC and Waters Synapt
G1Mass Spectroscopy under positive ESI conditions.
Methyl-(E)-3-(2-amino-3-fluorophenyl)acrylate (15). A
stirred slurry of 2-bromo-6-fluoroaniline (2) (1.30 kg, 6.82
mol), methyl acrylate (1.24 L, 13.6 mol), c-Hex2NMe (1.75 L,
8.18 mol), and Solka-floc (ca. 250 g) in i-PrOAc (10.4 L) was
thoroughly degassed (N2 sparge) before addition of (t-Bu)3P−
Pd G2 (26.2 g, 5.12 mol). The solution was heated to 80 °C
and aged for 5 h before being cooled to RT and filtered (filter
cake washed with i-PrOAc, 7.80 L). The filtrate was washed
with citric acid (1 M aq., 2 × 2.00 L) to provide a crude i-
PrOAc solution of acrylate 15 (16.2 kg, 8.2 wt % 15, 99% assay
yield) which was used directly without further purification. For
the purposes of characterization, a sample of pure acrylate 15
was isolated by crystallization from a partially concentrated
C17H15FNO4 [M + H]+ 316.0980, found 316.0971 (Δ = 0.9
+
mmu).
Methyl-(E)-3-(3-fluoro-2-(3-(2-methoxy-5-
(trifluoromethyl)phenyl)ureido)phenyl)acrylate (5). To a
stirred crude i-PrOAc solution of carbamate 16 (<6.66 mol)
was added 2-methoxy-5-(trifluoromethyl)aniline 3 (1.40 kg,
7.33 mol). The slurry was diluted with i-PrOAc, heated to 40−
45 °C, and a constant volume distillation carried out to
azeotropically dry the mixture (KF < 400 ppm). DMAP (40.5
g, 331 μmol) was added, and the solution was heated to reflux
for 5 h. The ensuing slurry was then cooled to RT and aged for
2 h before being filtered. The cake was washed with i-PrOAc
(4.00 L) and air-dried to afford urea 5 (2.43 kg, 5.89 mol, 99 wt
%, 88% from acrylate 15) as an off white crystalline solid. Mp
212−216 °C; 1H NMR (500 MHz, DMSO-d6) δH 9.01 (1H, s),
8.78 (1H, s), 8.46 (1H, s), 7.76 (1H, d, J = 16.1 Hz), 7.72 (1H,
d, J = 6.4 Hz), 7.31−7.35 (1H, m), 7.21 (1H, d, J = 8.4 Hz),
6.69 (1H, d, J = 16.1 Hz), 3.99 (3H, s), 3.72 (3H, s); 13C NMR
(500 MHz, DMSO-d6) δC 166.5, 157.8 (d, JCF = 245.6 Hz),
152.9, 150.1, 139.1 (d, JCF = 2.5 Hz), 133.2, 129.2, 127.5 (d, JCF
= 8.1 Hz), 124.8 (d, JCF = 13.7 Hz), 124.5 (q, JCF = 271.2 Hz),
122.5, 121.1 (q, JCF = 31.5 Hz), 119.9, 119.0 (q, JCF = 4.1 Hz),
117.3 (d, JCF = 20.8 Hz), 114.1 (q, JCF = 3.6 Hz), 110.9, 56.3,
1
sample of the crude solution. Mp 48−50 °C; H NMR (600
MHz, CDCl3) δH 7.79 (1H, d, J = 15.9 Hz), 7.15 (1H, d, J = 8.0
Hz), 7.98 (1H, m), 6.67 (1H, td, J = 8.0, 5.3 Hz), 6.36 (1H, d, J
= 15.9 Hz), 4.07 (2H, br s), 3.79 (3H, s); 13C NMR (150 MHz,
CDCl3) δC 167.3, 151.9 (d, JCF = 238.6 Hz), 139.0 (d, JCF = 3.9
Hz), 134.2 (d, J = 12.7 Hz), 123.1 (d, JCF = 3.2 Hz), 121.8 (JCF
= 3.4 Hz), 118.8, 117.9 (JCF = 7.9 Hz), 116.1 (JCF = 19.1 Hz),
+
51.6; HR-MS calcd for C19H17F4N2O4 [M + H]+ 413.1119,
51.7; HR-MS calcd for C10H11FNO2 [M + H]+ 196.0768,
+
found 413.1112 (Δ = 0.7 mmu).
Methyl-(E)-3-(3-fluoro-2-((((2-methoxy-5-
(trifluoromethyl)phenyl)imino)methylene)amino)-
found 196.0773 (Δ = 0.5 mmu).
Methyl-(E)-3-(3-fluoro-2-((phenoxycarbonyl)amino)-
phenyl)acrylate (16). To a stirred crude i-PrOAc solution of
D
DOI: 10.1021/acs.oprd.6b00076
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
phenyl)acrylate (17). To a stirred suspension of urea 5 (2.00
kg, 4.85 mol) in toluene (28.0 L) at RT was added 2-picoline
(1.47 L, 14.6 mol) followed by PCl5 (1.24 kg, 5.82 mol)
portionwise over 30 min. The reaction mixture was heated to
40 °C for 5 h before being cooled to 0 °C and quenched by
dropwise addition of KOH (2 M aq., 14.6 L, 29.1 mol),
maintaining internal temperature <20 °C. The mixture was
then filtered (2.00 L toluene vessel rinse), allowed to warm to
RT, and stirred for a further 2 h. The aqueous phase was
separated, and the organic phase was washed with citric acid (1
M aq., 7.28 L) and H2O (7.28 L) to afford a crude toluene
solution of carbodiimide 17 (27.5 kg, 6.4 wt % 17, 92% assay
yield) which was used directly without further purification. For
the purposes of characterization, a sample of pure carbodiimide
17 was isolated by crystallization from a sample of the crude
stirred solution of cinchonidine (51.5 g, 175 mmol) in MeCN/
i-PrOH (7:1, 840 mL) was added 3,5-bis-trifluoromethylbenzyl
bromide (22, 32.2 mL, 175 mmol). The mixture was degassed
(N2 sparge) then heated to 50 °C and aged for 5 h or until
cinchonidine was consumed. 3-(Trifluoromethyl)benzyl bro-
mide (40.1 mL, 263 mmol) was then added, and the reaction
mixture was heated to 70 °C and aged for 12 h. The solution
was then cooled to RT and concentrated in vacuo to ca. 250
mL. The residue was diluted with EtOAc (3.00 L), seeded with
bis-Quat PTC 14f (1.46 g, 1.74 mmol), and the ensuing slurry
was stirred for a further 12 h at RT. The slurry was filtered and
the cake washed with ethyl acetate (500 mL) before being air-
dried to afford bis-Quat PTC 14f as a yellow solid (133.6 g, 159
mmol, 99 A%, 91%). An analogous procedure was used for the
preparation of bis-quaternized catalysts 14a−m. Mp 178−181
°C; 1H NMR (600 MHz, DMSO-d6) δH δ 9.92 (1H, d, J = 6.2
Hz), 8.86 (1H, dd, J = 8.7, 1.3 Hz), 8.65 (1H, s), 8.62 (1H, d, J
= 9.0 Hz), 8.53 (1H, d, J = 6.1 Hz), 8.33 (1H, s), 8.27 (1H,
ddd, J = 8.6, 7.0, 1.2 Hz), 8.12 (1H, t, J = 7.7 Hz), 7.97 (1H, s),
7.74 (1H, d, J = 7.7 Hz), 7.67 (1H, d, J = 7.9 Hz), 7.62 (1H, t, J
= 7.8 Hz), 7.25 (1H, d, J = 5.2 Hz), 6.85 (1H, d, J = 4.4 Hz),
6.57 (1H, d, J = 15.7 Hz), 6.51 (1H, d, J = 15.7 Hz), 5.70−5.62
(1H, m), 5.55 (1H, d, J = 12.3 Hz), 5.24 (1H, dt, J = 17.3, 1.3
Hz), 4.93 (1H, dt, J = 10.5, 1.3 Hz), 4.46 (1H, t, J = 13.5 Hz),
4.12 (1H, d, J = 12.3 Hz), 4.08 (1H, t, J = 9.5 Hz), 3.44−3.38
(2H, m), 3.31 (1H, td, J = 11.6, 4.8 Hz), 2.65 (1H, s), 2.14−
2.03 (1H, m), 2.03−1.99 (1H, m), 1.82 (1H, t, J = 13.5 Hz),
1.43 (1H, ddd, J = 12.8, 9.1, 4.8 Hz); 13C NMR (150 MHz,
DMSO-d6) δC 158.5, 150.3, 138.1, 137.5, 135.7, 135.3, 135.0,
132.0, 131.6, 131.4, 131.2 (q, JCF = 33.4 Hz), 130.5, 130.2 (q,
JCF = 31.9 Hz), 127.5, 126.6, 125.9, 125.0, 124.2 (q, JCF = 272.8
Hz), 124.2, 123.6 (q, JCF = 272.8 Hz), 122.1, 120.3, 116.8, 68.0,
65.1, 60.4, 59.7, 59.5, 51.1, 37.3, 26.5, 24.6, 21.8; HR-MS calc
for C36H33F9N2O2+ [M]2+ 340.1219, found 340.1208 (Δ = 1.1
mmu).
M e t h y l - ( S ) - 2 - ( 8 - fl u o r o - 3 - ( 2 - m e t h o x y - 5 -
(trifluoromethyl)phenyl)-2-(4-(3-methoxyphenyl)-
piperazin-1-yl)-3,4-dihydroquinazolin-4-yl)acetate
(2S,3S)-2,3-bis((4-Methylbenzoyl)oxy)succinate Ethyl Ac-
etate Solvate (8a). To a stirred suspension of guanidine
salicylate 11a (498 g, 687 mmol) in PhMe (5.30 L) was added
K3PO4 (1 M aq., 1.03 L, 1.03 mol). The biphasic slurry was
heated to 45 °C and stirred for 1 h or until all solids were
dissolved. The aqueous layer was then separated; fresh K3PO4
(0.43 M aq., 2.40 L, 1.03 mmol) was added and the mixture
cooled to 0 °C. Vigorous stirring was maintained while a
solution of bis-Quat PTC 14f (28.9 g, 34.4 mmol) in DMF
(53.0 mL) was added rapidly and continued for a further 5 h at
0 °C. The aqueous phase was then separated, glycolic acid (1 M
aq., 1.03 L, 1.03 mol) was added, and the mixture was heated to
50 °C and stirred for 2 h. Without cooling, the aqueous phase
was separated and the organics washed with H2O (687 mL).
Darco KB-G (ca. 50 g) was then added and the suspension
stirred for 2 h while being allowed to cool to RT before being
filtered through Celite. The solution was concentrated in vacuo
to ca. 3.50 L with concomitant azeotropic removal of water to
provide a crude toluene solution (KF < 200 ppm) of
dihydroquinazaline 8 (4.98 L, 13.5 wt % 8, 98% assay yield,
76% ee). The solution was heated to 45 °C and a solution of
(S,S)-di-p-toloyltartaric acid (20, DTTA, 53.1 g, 137 mmol) in
EtOAc (160 mL) was then added and the mixture seeded with
dihydroquinazaline DTTA salt 8a (1.00 g, 943 μmol). The
mixture was stirred for 2 h at 45 °C during which time a thin
1
solution diluted with heptane. Mp 80−84 °C; H NMR (500
MHz, DMSO-d6) δH 7.95 (1H, d, J = 16.1 Hz), 7.70 (1H, d, J =
8.0 Hz), 7.59 (1H, dd, J = 8.7, 1.9 Hz), 7.46 (1H, d, J = 2.1 Hz),
7.40 (1H, ddd, J = 10.4, 8.4, 1.2 Hz), 7.23−7.31 (2H, m), 6.76
(1H, d, J = 16.2 Hz), 3.91 (3H, s), 3.74 (3H, s); 13C NMR (500
MHz, DMSO-d6) δC 166.8, 157.9 (d, JCF = 247.5 Hz), 157.1,
138.7 (d, JCF = 3.5 Hz), 135.4, 130.4, 126.8 (d, JCF = 8.4 Hz),
126.3, 125.7 (d, JCF = 13.7 Hz), 124.7 (q, JCF = 3.9 Hz), 124.4
(q, JCF = 271.1 Hz), 123.9 (d, JCF = 3.1 Hz), 122.1 (q, JCF
=
32.7 Hz), 121.8 (q, JCF = 3.7 Hz), 121.0, 117.8 (d, JCF = 19.5
+
Hz), 112.7, 57.2, 52.1; HR-MS calcd for C16H15O3N2F4 [M +
H]+ 395.1008, found 395.1013 (Δ = 0.5 mmu).
Methyl-(2E)-3-(3-fluoro-2-((((2-methoxy-5-
(trifluoromethyl)phenyl)amino)(4-(3-methoxyphenyl)-
piperazin-1-yl)methylene)amino)phenyl)acrylate 2-Hy-
droxybenzoate (11a). To a stirred solution of carbodiimide
17 in PhMe (27.5 kg, 6.4 wt % 17, 4.46 mol) was added a
solution of piperazine bis-hydrochloride 4a (1.38 kg, 5.09
mmol) in H2O (13.0 L) followed by Et3N (1.49 L, 10.7 mol).
The reaction mixture was stirred at RT for 30 min before being
heated to 40 °C and held for 1 h. The aqueous phase was
separated and the organic phase washed with NaH2PO4 (2 ×
9.70 L) and H2O (7.30 L) to afford a solution of crude
guanidine free base 11. Salicylic acid (19, 744 g, 5.34 mol) was
then added, and the solution was stirred at RT for 1 h before
being cooled to 0 °C and stirred for a further 2 h. The ensuing
slurry was filtered and the cake washed with cold (0 °C) PhMe
(4.00 L) and dried under air to afford guanidine salicylate 11a
(3.16 kg, 4.36 mol, 98 wt %, 90% from 5) as a white crystalline
solid. mp 112−118 °C; 1H NMR (500 MHz, CD3CN) δH 12.9
(1H, br s), 7.75 (1H, dd, J = 7.8, 1.8 Hz), 7.72 (1H, d, J = 16.1
Hz), 7.40 (1H, td, J = 7.2, 1.7 Hz), 7.27 (1H, d, J = 7.8 Hz),
7.17 (1H, m), 7.16 (1H, t, J = 8.2 Hz), 7.02 (1H, br s), 6.95
(1H, t, J = 8.6 Hz), 6.88−6.81 (3H, m), 6.78 (1H, br s), 6.60
(1H, dd, J = 8.2, 2.0 Hz), 6.54 (1H, m), 6.48 (1H, d, J = 16.1
Hz), 6.43 (1H, dd, J = 8.0, 2.1 Hz), 3.73 (3H, s), 3.71 (3H, s),
3.69 (4H, br s), 3.68 (3H, s); 13C NMR (125 MHz, CD3CN)
δC 173.1, 166.4, 161.9, 160.7, 156.6, 156.5 (d, JCF = 247.7 Hz),
155.4, 152.0, 137.7, 134.4, 130.5, 129.9, 127.9 (d, JCF = 8.2 Hz),
127.9, 126.2, 125.0 (q, JCF = 3.7 Hz), 124.8 (d, JCF = 12.8 Hz),
123.5 (q, JCF = 273.1 Hz), 122.3, 122.0 (q, JCF = 34.8 Hz),
121.4, 121.0, 118.4, 117.2 (d, JCF = 11.8 Hz), 116.6, 115.5,
115.3, 112.1, 108.8, 105.4, 102.5, 56.1, 54.8, 51.4, 48.0, 47.9;
HR-MS calc for C30H31F4N4O4 [M + H]+ 587.2276, found
+
587.2272 (Δ = 0.4 mmu).
(1S,2S,4S,5R)-1-(3,5-Bis(trifluoromethyl)benzyl)-2-((R)-
hydroxy(1-(3-(trifluoromethyl)benzyl)quinolin-1-ium-4-
yl)methyl)-5-vinylquinuclidin-1-ium Bromide 14f. To a
E
DOI: 10.1021/acs.oprd.6b00076
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
slurry formed. Further DTTA (212 g, 550 mmol) in EtOAc
(640 mL) was then added via syringe pump over 12 h. The
slurry was allowed to cool to RT and stirred for 3 h before
being filtered. The cake was washed with EtOAc (3 × 500 mL)
and air-dried to afford dihydroquinazaline DTTA salt 8a (597
g, 563 mmol, 99 wt %, 82%, 99.6% ee) as a crystalline white
solid. Mp 126−133 °C; 1H NMR (600 MHz, CD3CN) δH 7.87
(4H, d, J = 8.2 Hz), 7.62 (1H, dd, J = 8.8, 2.2 Hz), 7.23 (4H, d,
J = 8.0 Hz), 7.16 (1H, m), 7.14−7.08 (3H, m), 6.95 (1H, dd, J
= 7.1, 1.7 Hz), 6.42 (1H, dd, J = 8.1, 2.3 Hz), 6.36 (1H, dd, J =
8.2, 2.3 Hz), 6.31 (1H, t, J = 2.4 Hz), 5.81 (2H, s), 5.05 (1H, t,
J = 7.0 Hz), 4.08 (2H, q, J = 7.1 Hz), 3.77 (3H, bs), 3.74 (3H,
s), 3.66 (3H, s), 3.52−3.41 (4H, m), 3.08 (1H, dd, J = 15.7, 7.3
Hz), 2.93−2.72 (4H, m), 2.70 (1H, dd, J = 15.7, 7.3 Hz), 2.35
(6H, s), 2.00 (3H, s), 1.23 (3H, t, J = 7.1 Hz); 13C NMR (150
MHz, CD3CN) δC 170.7, 170.5, 168.7, 165.2, 160.6, 156.6,
154.3, 153.2 (d, JCF = 249.1 Hz), 152.0, 144.4, 132.4, 129.8,
129.7, 129.5, 129.4 (d, JCF = 13.6 Hz), 129.2, 126.7, 126.0 (q,
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.oprd.6b00076.
■
S
1H and 13C NMR spectra, further characterization data,
ligand structures for Scheme 2, and expansions of Figure
3 (PDF)
AUTHOR INFORMATION
Corresponding Authors
*E-mail: stephen.dalby@merck.com.
*E-mail: guy_humphrey@merck.com.
■
Present Address
†M.R.L.: Process Research and Development, The Bristol-
Myers Squibb Pharmaceutical Research Institute, 1 Squibb
Drive, New Brunswick, New Jersey 08903.
Notes
JCF = 3.0 Hz), 125.7, 124.3, 124.0 (q, JCF = 271.0 Hz), 122.9 (q,
The authors declare no competing financial interest.
JCF = 33.2 Hz), 121.2, 117.3, 115.4 (d, JCF = 18.6 Hz), 113.4,
108.6, 104.9, 102.3, 71.7, 60.1, 60.0, 56.3, 54.7, 51.6, 47.7, 47.6,
ACKNOWLEDGMENTS
■
+
39.1, 20.7, 20.2, 13.5; HR-MS calcd for C30H31F4N4O4 [M +
We gratefully acknowledge A. Kalinin and E. Sirota (Merck &
Co., Inc.) for engineering contributions and G. Li and R. Xiang
(Merck & Co., Inc.) for analytical support.
H]+ 587.2286, found 587.2276 (Δ = 1.0 mmu); Rt 5.1 min
(undesired)/6.3 min (desired) (Chiral Pack IC-3, 150 × 4.6
mm, 3.0 μm), 70/30 MeCN/(0.05% aqueous TFA with 5 mM
Na2B4O7), 0.9 mL·min−1, λ = 210 nm.
REFERENCES
■
(S)-2-(8-Fluoro-3-(2-methoxy-5-(trifluoromethyl)-
phenyl)-2-(4-(3-methoxyphenyl)piperazin-1-yl)-3,4-di-
hydroquinazolin-4-yl)acetic Acid (Letermovir, 1). Dihy-
droquinazaline DTTA salt 8a (40.0 g, 37.7 mmol) was slurried
in MTBE (80 mL) and Na2HPO4 (0.5 M aq., 188 mL, 94.0
mol) added. The mixture was stirred for 2 h before the aqueous
layer was separated and NaOH (1 M aq., 151 mL, 151 mmol)
added. The stirred mixture was then heated to 60 °C, with
concomitant removal of MTBE by distillation. The resulting
aqueous reaction mixture was stirred at 60 °C for 5 h before
being cooled to RT. H2O (80 mL) and MTBE (200 mL) were
then added, and the organic phase was separated (discarded).
Further MTBE (200 mL) was added to the aqueous residue,
which was then acidified (pH 5−6) with HCl (3 M aq., ca. 40
mL). The organic phase was separated and concentrated to ca.
100 mL. Removal of MTBE through constant volume solvent
switch with acetone (600 mL) was carried out, providing a final
solution of ca. 60 mL. This solution was then added over 30
min to H2O (320 mL), and the ensuing slurry was stirred for a
further 1 h before being filtered. The cake was washed with
H2O (3 × 50 mL) and air-dried to provide letermovir (1, 20.2
g, 35.3 mmol, 100 wt %, 94%) as an amorphous white powder.
1H NMR (DMSO-d6, 600 MHz) δH 7.52 (dd, J = 8.7, 1.7 Hz,
1H), 7.40 (brs, 1H), 7.21 (m, 1H), 7.07 (t, J = 8.2 Hz, 1H),
7.04 (m, 1H), 6.87 (m, 2H), 6.44 (dd, J = 8.2, 1.9 Hz, 1H), 6.40
(t, J = 2.3 Hz, 1H), 6.36 (dd, J = 8.0, 2.0 Hz, 1H), 4.89 (t, J =
7.2 Hz, 1H), 3.80 (brs, 3H), 3.68 (s, 3H), 3.39−3.48 (m, 4H),
2.82−2.95 (m, 4H), 2.80 (dd, J = 14.8, 7.4 Hz, 1H), 2.46 (dd, J
= 14.9, 7.4 Hz, 1H); 13C NMR (DMSO-d6, 150 MHz) δC
171.8, 160.2, 156.5, 154.6 (d, JCF = 246.3 Hz), 153.2, 152.2,
134.2, 132.3 (d, JCF = 11.2 Hz), 129.6, 124.1 (q, JCF = 271.3
Hz), 123.8 (q, JCF = 3.7 Hz), 122.4, 122.1 (q, JCF = 7.1 Hz),
121.4 (q, JCF = 29.2 Hz), 120.8, 114.5 (d, JCF = 19.5 Hz), 113.3,
108.3, 104.6, 101.9, 59.0, 56.3, 54.8, 47.9, 45.6, 40.0; HR-MS
(1) (a) Kotton, C. N. Am. J. Transplant. 2013, 13, 24. (b) Boeckh,
M.; Geballe, A. P. J. Clin. Invest. 2011, 121, 1673.
(2) (a) Komatsu, T. E.; Pikis, A.; Naeger, L. K.; Harrington, P. R.
Antiviral Res. 2014, 101, 12. (b) Lurain, N. S.; Chou, S. Clin. Microbiol.
Rev. 2010, 23, 689. (c) Scott, G. M.; Weinberg, A.; Rawlinson, W. D.;
Chou, S. Antimicrob. Agents Chemother. 2007, 51, 89.
(3) (a) Goosen, K.; Kuhn, O.; Berwe, M.; Krueger, J.; Militzer, H.-C.
CN101863843, DE10227517, EP1893587, JP2008543797,
US2009221822, US8084604, US2012130072, US8372972,
WO2006133822. (b) Lischka, P.; Hewlett, G.; Wunberg, T.;
Baumeister, J.; Paulsen, D.; Goldner, T.; Ruebsamen-Schaeff, H.;
Zimmermann, H. Antimicrob. Agents Chemother. 2010, 54, 1290.
(c) Verghese, P. S.; Schleiss, M. R. Drugs Future 2013, 38, 291.
(4) (a) Chemaly, R. F.; Ullmann, A. J.; Stoelben, S.; Richard, M. P.;
Bornhauser, M.; Groth, C.; Einsele, H.; Silverman, M.; Mullane, K. M.;
̈
Brown, J.; Nowak, H.; Kolling, K.; Stobernack, H. P.; Lischka, P.;
̈
Zimmermann, H.; Rubsamen-Schaeff, H.; Champlin, R. E.; Ehninger,
̈
G. N. Engl. J. Med. 2014, 370, 1781. (b) Goldner, T.; Hewlett, G.;
Ettischer, N.; Ruebsamen-Schaeff, H.; Zimmermann, H.; Lischka, P. J.
Virol. 2011, 85, 10884. (c) De Clercq, E. Med. Res. Rev. 2013, 33, 1249.
(5) (a) Shirakawa, S.; Maruoka, K. Angew. Chem., Int. Ed. 2013, 52,
4312. (b) Herchl, R.; Waser, M. Tetrahedron 2014, 70, 1935.
(6) For examples of asymmetric PTC aza-Michael reactions see:
(a) Bandini, M.; Eichholzer, A.; Tragni, M.; Umani-Ronchi, A. Angew.
Chem., Int. Ed. 2008, 47, 3238. (b) Miyaji, R.; Asano, K.; Matsubara, S.
Org. Lett. 2013, 15, 3658. (c) Sallio, R.; Lebrun, S.; Schifano-Faux, N.;
Goossens, J.-F.; Agbossou-Niedercorn, F.; Deniau, E.; Michon, C.
Synlett 2013, 24, 1785.
(7) Preliminary work by AiCuris had demonstrated low levels of
enantioselectivity (20−30% ee) in the PTC cyclization of urea 5.
(8) A transient TFE adduct was observed by LCMS; however, its
presumably improved configurational stability is not fully understood.
(9) Shinohara, T.; Arai, M. A.; Wakita, K.; Arai, T.; Sasai, H.
Tetrahedron Lett. 2003, 44, 711.
(10) (a) Oestreich, M. The Mizoroki-Heck Reaction; Wiley: New
York, 2009. (b) Beletskaya, I. P.; Cheprakov, A. V. Chem. Rev. 2000,
100, 3009.
(11) Gottumukkala, A. L.; de Vries, J. G.; Minnaard, A. J. Chem. - Eur.
J. 2011, 17, 3091 and references therein..
calcd for C29H29F4N4O4 [M + H]+ 573.2119, found 573.2117
+
(Δ = 0.2 mmu).
(12) See the Supporting Information for details.
F
DOI: 10.1021/acs.oprd.6b00076
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
(13) DFT ground state energy calculations at the B3LYP 6-31G**
level for the corresponding quinazalinone indicated the (E)-acrylate to
be + 4.0 kcal mol−1 relative to its hydratesee the Supporting
Information for details. We note the competetive reductive
debromination process may also relate to the relative inaccessibility
of an intermediate arylpalladium complex conformer appropriate for
migratory insertion of the acrylate group.
(14) Cyclization of 11 in alcoholic solvents is presumed to be a
Bronsted acid-catalyzed processTFE: 100% cyclized in <15 min;
̈
MeOH: 100% in 1 h; i-PrOH: 78% in 22 h; DMSO: <1% after 48 h.
(15) Xiang, B.; Belyk, K. M.; Reamer, R. A.; Yasuda, N. Angew. Chem.,
Int. Ed. 2014, 53, 8375.
(16) Other catalysts evaluated included chiral phosphoric acids,
Maruoka PTCs, and chiral Lewis acids.
(17) Luzung, M.; Humphrey, G. R.; Xiang, B.; Belyk, K. M.; Dalby, S.
M.; Schwab, W.; Klenke, B.; Moody, T.; Brown, G. Process for the
preparation of substituted quinazoline compounds. WO 2015088931 A1,
2015.
(18) Bruno, N. C.; Tudge, M. T.; Buchwald, S. L. Chem. Sci. 2013, 4,
916.
(19) “Sufficient” agitation could be reliably modelled and scaled using
the just dispersed speed (NJD). See Armenante, P. M.; Huang, Y. T.
Ind. Eng. Chem. Res. 1992, 31, 1398 Notably, restoring sufficient
agitation after an initial period did not restore the corresponding
reaction rate and enantioselectivity.
−
(20) Alternative PTC counterions included: Cl− (70% ee), NO3
−
−
(61% ee), BF4 (51% ee), OTf− (27% ee), SbF6 (23% ee). Spiking
KBr led to increasingly diminished reaction rate and enantioselectivity
(0.25 eq KBr − 77% conv after 14 h, 60% ee) together suggesting that
counterion exchange plays a critical role in this process.
(21) 1 mol% PTC loading led to diminished rate and selectivity (18
h, 60% ee). Slow catalyst degradation via Hoffmann elimination was
observed during the reaction; this degradate was only weakly
catalytically active.
(22) The application of asymmetric catalysis thus translates into a
significant yield enhancement compared to resolution of racemic 11
(ref 3a). An 82% isolated yield of 8a (99.6% ee) from 11a via 8 (76%
ee, 88:12 er) represents an effective overall yield of 93% for this
transformation.
G
DOI: 10.1021/acs.oprd.6b00076
Org. Process Res. Dev. XXXX, XXX, XXX−XXX