Synthesis of Fused Oxepane HIV Integrase Inhibitor MK-1376

Article abstract of DOI:10.1055/s-0040-1707994

Controlling the absolute and relative stereochemistry of a seven-membered oxepane in the formation of HIV integrase inhibitor MK-1376 was accomplished through a strategy involving the use of asymmetric allylation and stereoconvergent, substrate-directed installation of an amine fragment. Surprising reactivity was demonstrated during the asymmetric allylation in which the allyl-pyrimidone product was formed reversibly. The stereoconvergent amine addition was accomplished through an elimination/addition sequence involving a quinone methide reactive intermediate, and nucleophilic trapping of the reactive quinone methide intermediate with methylamine. This novel approach delivered MK-1376, offering 100-fold greater productivity and 50-fold less waste than the initial synthetic chemistry route.

Full text of DOI:10.1055/s-0040-1707994

SYNTHESIS
0
0
3
9
-
7
8
8
1
1
4
3
7
-
2
1
0
X
© Georg Thieme Verlag Stuttgart · New York
2020, 52, A–K
special topic
A
en
P. E. Maligres et al.
Special Topic
Synthesis
Synthesis of Fused Oxepane HIV Integrase Inhibitor MK-1376
Peter E. Maligres*a
Zhiguo Jake Song*
Neil A. Strotman
Jinquin Yin
Tao Pei1
Hallena R. Strotman2
0
0
0
0-
0
0
0
2-2
2
3
7-9
0
0
2
O
COOMe
1. π-allyl cyclization
OH
2. hydrogenation
HN
HO
CN
OH
OAc O
O
enantioselective
catalyst control
N
Et
COOMe
O
N
OMe
"quinone methide"
OH
N
Me
O
Et
O
O
O
Tetsuji Itoh
benzylic
O
OH
bromination
elimination
stereoselective
addition
NH
N
O
NMe
N
Edward C. Sherer
Guy R. Humphrey
O
O
O
O
N
N
substrate
control
O
NMe2
OMe
OMe
X = H, X = Br
X
Department of Process Research and Development,
Merck & Co., Inc., Kenilworth, NJ 07033, USA
peter_maligres@merck.com
MeNH2
MK-1376
11% overall
F
100-fold greater productivity & 50-fold less waste over original route
Published as part of the Special Topic Synthesis in Industry
Received: 22.01.2020
Accepted after revision: 20.02.2020
Published online: 16.03.2020
O
O
Me
OH
OH
O
MeN
N
H
N
DOI: 10.1055/s-0040-1707994; Art ID: ss-2020-c0045-st
N
O
O
N
N
N
O
Me Me
O
HN
HN
Abstract Controlling the absolute and relative stereochemistry of a
seven-membered oxepane in the formation of HIV integrase inhibitor
MK-1376 was accomplished through a strategy involving the use of
asymmetric allylation and stereoconvergent, substrate-directed instal-
lation of an amine fragment. Surprising reactivity was demonstrated
during the asymmetric allylation in which the allyl-pyrimidone product
was formed reversibly. The stereoconvergent amine addition was ac-
complished through an elimination/addition sequence involving a qui-
none methide reactive intermediate, and nucleophilic trapping of the
reactive quinone methide intermediate with methylamine. This novel
approach delivered MK-1376, offering 100-fold greater productivity
and 50-fold less waste than the initial synthetic chemistry route.
Me
N
N
Me
Me
O
Raltegravir
MK-0396
F
F
Me
O
O
N
OH
N
O
O
HN
Me
N
N
Me
Me
O
Key words HIV integrase inhibitor, oxepane, pyrimidine, -allyl cy-
clization, quinone methide
MK-1376
F
Figure 1 HIV Integrase inhibitors
Global estimates from the Joint United Nations Program
on HIV/AIDS indicate that, as of 2018, over 76 million peo-
ple have become infected with HIV, and 37 million have
died from AIDS related illnesses since the 1981 start of the
epidemic.3 In 2017 alone, nearly 37 million people were liv-
ing with AIDS, 1.8 million new cases had been diagnosed,
and nearly 1 million people had died as a result. These dis-
couraging statistics have spurred rapid expansion in
HIV/AIDS research, with the discovery of multiple drugs
targeting reverse transcriptase and protease to achieve suc-
cessful combinational therapies. Despite the advances af-
forded by these approaches, drug resistance inevitably be-
comes a problem. More recently, HIV integrase has been
targeted, with MSD’s integrase inhibitor Raltegravir (Figure
1) representing an especially potent option.4 More elabo-
rate HIV integrase inhibitors bearing an oxepane fused
pyrimidine have subsequently been identified, including
MK-0396, bearing a stereogenic center5 and MK-1376,
bearing two stereogenic centers.6
The original route for MK-1376 (Scheme 1) utilized
commercially available, but expensive, 1,2-epoxybutane for
the efficient preparation of an advanced intermediate that
was an epimeric mixture at the nitrogen-bearing stereo-
center. An intense effort was required for production of
gram quantities of final product through an inefficient syn-
thesis, mostly due to low cyclization yields, preparative
chromatography to separate the diastereomers, and the
complication of oxepane epimerization.
This existing chemistry was not amenable to scale, as
preparation of a single kilogram of MK-1376 would gener-
ate more than 60 metric tons of waste. As a result, a new
route was required to enable the preparation of clinical
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K
Georg Thieme Verlag KG, Rüdigerstraße 14, 70469 Stuttgart, Germany

B
P. E. Maligres et al.
Special Topic
Synthesis
Me
Me
O
1) NH2OH, MeOH
2) DMAD, o-xylene
150 °C
NC
NMeBoc
O
N
1) NaH, DMF
2) O3
O
N
OH
X
OH
OH
N
OH
X
+
O
NH
3) NaCN
MeNH2
Boc2O
40%
3) PFBA
30%
O
H
Et
O
O
O
Et
Me
(R)
NHBoc
O
N
O
OH
Et
A
1) MsCl, Et3N,
MeCN 70%
D
BocMeN
OH
NH
*
6
N
O
F
O
F
N
NH
2) K2CO3, DMAc,
140 °C,
microwave
7%
3) preparative
HPLC
10
O
H
N
Me
N
O
N
Me
O
OH
Me
O
C
B
X
O
BocMeN
1) HCl, EtOAc
X
A hydroamination
then asymmetric
hydrogenation
B C-10 amine
then π-allyl cyclization
C C-10 O-leaving group
then π-allyl cyclization
D π-allyl cyclization
O
2) Me2NCOCOOH,
EDCI, HOAt,
NMM, DBU
70%
F
O
N
N
N
Et
OH
H
N
OH
X
NH
NH
O
O
O
O
O
N
O
O
OH
X
OMs
F
NHBoc
then C-10 oxidatiion
Me
O
N
PFBA =
Scheme 2 Retrosynthetic strategy
Me2N
O
F
N
N
Et
H
N
NH2
O
AuCl3
AgSbF6
ligand
Me
O
N
OH
O
OH
R
HN
<0.1% MK-1376 overall
O
Scheme 1 Initial route for HIV integrase inhibitor MK-1376
1
Me
Me
supplies of MK-1376. Several retrosynthetic routes were
envisioned (Scheme 2). Our initial strategy was to introduce
the C-10 amine center asymmetrically, followed by cycliza-
tion to the oxepane, and finally, asymmetric introduction of
the C-6 center through either a hydroamination (route A) or
a -allylation approach (route B). The C-10 amine could also
be installed later via displacement of the corresponding
mesylate (route C). Alternatively, an approach was envi-
sioned with oxidative installation of the C-10 center after
having installed the C-6 center via the -allylation (route
D).
Our attempts at intramolecular hydroamination of a
representative propargylic ether 1 to form the seven-mem-
bered azepane were unsuccessful (Scheme 2, route A).
There was a strong preference for cyclization to form the
eight-membered ring, providing only traces of the desired
azepane 2 (Scheme 3).
Given this result, we turned our attention (Scheme 2,
route B) to the intramolecular -allyl cyclization of allylic
acetate 6, which is readily available from inexpensive cis-
butene diol 4 (Scheme 4). Since earlier studies had suggest-
ed that the C-10 stereocenter of intermediates 5, 6, and 7
was prone to epimerization, the desired stereochemistry
would need to be established later in the synthesis. Initially
the -allyl cyclization was run in a racemic sense to allow
rapid exploration of downstream chemistry, yielding exclu-
sively the trans isomer.7 We hoped that the C-10 stereo-
O
N
O
OH
R
OH
N
N
O
O
N
R
2
3
minor
major
Scheme 3 Alkyne hydroamination
chemistry could be inverted through a sequence similar to
that used for the preparation of MK-0396 (Figure 1) involv-
ing oxidation to the chloroamine followed by elimination to
an imine/enamine intermediate5b and subsequent syn-se-
lective reduction. Computations suggested a preference for
the anti-addition transition state of sodium triacetoxyboro-
hydride to imine to give syn 9a by 14 kcal/mol, providing
hope for a stereoselective reduction. While selectivity var-
ied based on the hydride source employed (Scheme 4, bot-
tom), the use of sodium borohydride and sodium triace-
toxyborohydride both offered >20:1 syn-selectivity for the
desired product. Unfortunately, upon reinvestigation of the
-allyl cyclization to render the route asymmetric, no suc-
cessful conditions were identified. We hypothesized that
the significant steric bulk imparted by the Boc-methyl-
amine substituent was partially responsible for the poor se-
lectivity of the -allyl cyclization, leading us to investigate
installation of this group at a later stage.
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K

C
P. E. Maligres et al.
Special Topic
Synthesis
OH
OTIPS
OH
OTIPS
1) DMAD, MeOH
then o-xylene, 145 °C
2) PhSO2Cl, Pyr
1) TIPSCl, KOt-Bu
2) BrCH2COOt-Bu, NaH
3) DIBAL-H
1) TIPSCl, KOtBu
2) BrCH2COOtBu, NaH
3) DIBAL-H
1) DMAD, MeOH
then o-xylene, 145 °C
4) TMSCN
5) aq NH2OH, MeOH
2) TBAF
NOH
4) KCN, MeNH2, then
Boc2O
5) aq NH2OH, MeOH
3) aq HCl
4) Ac2O, Pyr
NOH
3) Ac2O, pyr
OH
O
OH
O
NH2
NH2
32%
61%
38%
56%
4
4
OH
NMeBoc
5
10
OAc
1) π-allyl
cyclization
2) H2, Pd
1) π-allyl
Et
O
OAc
O
N
cyclization
1) H2, Pd
2) MeNH2
O
N
OBs
O
N
2) aq LiOH
OMs
N
N
OBs
O
OAc
HN
3) PFBA
58%
O
O
HN
3) MsCl
Et3N
COOMe
58%
64%
N
COOMe
O
O
COOMe
NHPFB
NMeBoc
OMs
anti
NMeBoc
7
6
1:1 syn/anti
OAc
12
11
Bs = SO2Ph
Et
O
N
Et
Et
O
O
N
1) HCl
OH
reductant
O
2) t-BuOCl
OH
N
N
N
O
O
O
O
3) DBU
4) MeNH2
N
COOMe
COOMe
NHPFB
NHMe
postulated quinone methide
23%
NHMe
desired
syn only
type intermediate
8
9a
13
Et
Et
O
N
O
Scheme 5 Stereoselective C-10 displacement with MeNH2
OH
OH
N
O
O
+
O
O
N
N
Quinone-methide
NHPFB
NHPFB
NHMe
NHMe
syn-9a
Syn
Anti
anti-9b
syn-9a /anti-9b
Reductant
NaB(OAc)3H
NaBH4
NaBH3CN
BH3·t-BuNH2
BH3-Pyr
>20:1
>20:1
11:1
2:1
PFB =
ΔΔG‡ = 1.7 kcal/mol
ΔΔG‡ = 0.0 kcal/mol
1:5
F
BH3·Et3N
1:5
Figure 2 Transition states leading to syn and anti addition of methyl-
amine to 13
Scheme 4 C-10 Amine oxidation/stereoselective reduction
To this end, we focused our efforts on C-10 oxygenated
systems featuring less steric hindrance at C-10 instead of
protected C-10 amines (Scheme 2, route C). We envisioned
activation of the C-10 center via conversion into a leaving
group, taking advantage of the latent phenol to reveal a qui-
none methide system, which eliminates the need for stereo-
control at the originating sp3 C-10 center (Scheme 5).8 Trap-
ping of the quinone methide by methylamine could afford
exclusively the syn isomer through this stereoconvergent
approach. While the hydride addition was shown to occur
from the anti face, initial transition-state modeling of the
quinone methide and methylamine suggested that the tran-
sition state leading to the desired product syn-9a was fa-
vored by 1.7 kcal versus the corresponding anti transition
state (Figure 2). We envisioned that the C-6 enantioen-
riched mesylate 12 could be accessed via a cyanohydrin-
based route involving asymmetric allylation (Scheme 5).
Initial screens of intramolecular asymmetric allylation
of acetate 11 to form oxepane 14 revealed some surprising
results (Table 1). While reactions with the naphthyl-Trost
ligand 15 and Pd2dba3 in DCM (CH2Cl2) gave no C-6 enantio-
selectivity under typical conditions, the addition of Bu4NBr
resulted in high enantioselectivity and modest diastereose-
lectivity at moderate conversions (entries 1 vs. 2). It was
also found that the activity was highest with a large ratio of
the ligand to metal (2:1), despite the use of a bidentate
phosphine.9 In fact, at a ratio of 1:1, no conversion was ob-
served (entry 3). Reactions in other solvents showed either
low reactivity or poor enantioselectivity (entries 4–7). Low-
ering the temperature showed a modest increase in selec-
tivity, but a concomitant decrease in conversion, while rais-
ing the temperature showed the converse (entries 8 and 9).
Upon extended aging, reactions reached moderately higher
conversion, but showed severe degradation of ee (entry 10).
To understand the effect of Bu4NBr, other halides (Cl, I),
tetrafluoroborate, and tetrabutylphosphonium ion were in-
vestigated, but all were inferior (entries 11–15). Higher lev-
els of Bu4NBr resulted in slight improvements in ee, but de-
creased reactivity (entries 16 and 17). Higher catalyst load-
ing was effective in achieving higher conversion but led to
decreased ee and cis/trans ratio, likely due to product epi-
merization.10
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K

D
P. E. Maligres et al.
Special Topic
Synthesis
Table 1 Optimization of Allylic Amination on C-10 Acetoxy Intermedi-
ate 11
reactivity could be leveraged in a more efficient route in-
volving a benzylic oxidation at the C-10 position (Scheme 2,
route D). Pyrimidinone silyl ether 17 was prepared in four
steps from inexpensive 1,4-cis-butenediol and isolated as a
crystalline solid (Scheme 6). The ester functionality could
either be retained or converted into the corresponding am-
ide 18. Products 21 and 22 were obtained after a sequence
involving benzenesufonylation, desilylation, and O-acyla-
tion.
OAc
Pd2(dba)3 (3 mol%)
ligand 15 (12 mol%)
O
N
O
OAc
OAc
Bu4NBr (1 equiv)
DCM (0.17 M)
20 °C
N
HN
O
O
COOMe
N
COOMe
OAc
OAc
O
O
11
14
Ph2P
NH HN
PPh2
OH
OTBS
1) KOt-Bu, TBSCl
THF
(R,R)-Naph-Trost
ligand 15
DMAD
MeOH
2)
cat. DBU
CN
3) 50% aq, NH2OH
MeOH
NOH
NH2
then
o-xylene
145 °C
Entry Change from standarda
Conv
(%)
cis/trans cis
trans
OH
O
ee (%) ee (%)
47% from 4
4
16
1
2
none
40
100
<2
57:43
41:59
ND
92
<5
ND
ND
54
<5
<5
95
89
68
62
33
91
73
<5
96
98
91
86
<5
ND
ND
44
<5
<5
93
53
33
61
28
78
53
<5
87
93
77
OTBS
BsCl
Et3N
no Bu4NBr
O
3
1:1 L/Pd
OH
R
HN
cat. DMAP
MeCN
88%
4
MeCN
<2
ND
O
N
5
MeCN, no Bu4NBr
THF, no Bu4NBr
toluene, no Bu4NBr
10 °C
100
100
79
39:61
37:63
55:45
58:42
48:52
37:63
42:58
41:59
55:45
47:53
39:61
57:43
58:42
47:53
PFBA
MeOH
94%
17: R = COOMe
18: R = CONHPFB
Bs = PhSO2
6
7
OAc
OTBS
1. cat. aq HCl
2. Ac2O, Pyr
8
25
O
N
O
OBs
OBs
R
9
40 °C, 8 h
64
HN
HN
86% from 19
10
11
12
13
14
15
16
17
18
40 °C, 22 h
75
O
O
N
R
Bu4NI
37
19: R = COOMe
21: R = COOMe
22: R = CONHPFB
Bu4NCl
82
20: R = CONHPFB
Bu4PBr
51
Scheme 6 -Allyl substrate preparation
Bu4PI
44
Bu4NBF4
91
It was envisioned that compounds 21 and 22 could un-
dergo a similar sequence as with acetate 11 involving asym-
metric allylic amination and alkene hydrogenation to give
bicycles 23 and 24, respectively, bearing no substitution at
the C-10 position (Scheme 7). Subsequent oxidation could
intercept the chemistry in Scheme 5 at the quinone
methide intermediate (13).
0.5 equiv Bu4NBr, 2 mol% Pd
2 equiv Bu4NBr, 2 mol% Pd
2 equiv Bu4NBr, 5 mol% Pd
46
29
88
a Standard conditions: Naph-Trost ligand 15 (12 mol%), Pd2dba3 (3 mol%),
(2:1 L/M), Bu4NBr (1 equiv), substrate 11 (0.17 M), 20 °C, 14–22 h.
This surprising observation hinted at unprecedented re-
activity by which the allylation step was reversible. Materi-
al that was >90% (S)-configuration at C-6 was carried for-
ward to mesylate 12 as a ca. 1:1 mixture of C-10 epimers.
Despite this starting mixture, deacylation and activation as
the mesylate 12, leading to quinone methide 13 gave exclu-
sively the syn product upon attack by MeNH2. Ultimately,
while this approach delivered the desired material, chal-
lenges related to ee erosion at high conversion of the -al-
lylation led us to investigate an alternative approach.
OAc
HN
O
O
alkene
hydrogenation
allylic
amidation
OBs
R
OBs
R
N
O
67% from 21
O
N
N
21: R = COOMe
22: R = CONHPFB
23: R = COOMe
24: R = CONHPFB
Et
O
Et
O
N
C-10 benzylic
oxidation
OBs
OBs
R
N
N
O
O
N
R
The previous route provided proof of concept that a -
allyl cyclization could be employed to set the C-6 stereocen-
ter, and this group could direct attack of methylamine on a
quinone methide to afford the desired syn selectivity. This
X
25: R = COOMe
26: R = CONHPFB
X = OH, =O, Cl, Br, I, NR2
Scheme 7 -Allyl cyclization and benzylic oxidation
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K

E
P. E. Maligres et al.
Special Topic
Synthesis
Applying what was learned from C-10 acetate 11, opti-
mal asymmetric allylation conditions were developed for
unsubstituted allylic acetate 21 after further screens involv-
ing 0.75% mol Pd2dba3, 3.15% mol (R,R)-Naph-Trost ligand
(15), and 10% mol Bu4NBr at 0.17 M in DCM and 22 °C. In
only 7 h, these conditions gave 98.5% conversion and 96.2%
ee, with 2.6% of elimination by-product 27 (Scheme 8).11
However, as before, prolonged reaction times were found to
lead to erosion of ee from reversible ring closure, and, in
this system, also resulted in formation of diene 27 from
elimination. Since the pyrimidinone can act as a leaving
group, the allylic product is also susceptible to formation of
the -allyl Pd species. To our knowledge, this is the first ex-
ample of reversibility in a -allylation reaction. To suppress
the epimerization and elimination processes post reaction,
a series of additives were explored, but most showed little
effect. Leveraging our previous observation that a low L/M
ratio suppressed catalysis, we found that the addition of 1.5
mol% Pd(OAc)2 immediately halted further reaction (pro-
ductive and unproductive).
Table 2 Screen of Hydrogenation of 23a
Catalyst
Unreacted SM 23 Product 25
(%) (%)
n-Butyl ether 28
(%)
none
90
4
6
19
24
7
5% Pt/C
9
3
72
73
93
4
5% Pt/alumina
5% Pd(S)/C
IrO2
0
88
59
4
8
2% Ir/C
28
81
79
13
15
20
5% Pd/BaSO4
5% Pt(S)/C
1
a Reaction conditions: Crude allylic amination stream in i-PrOAc treated
with Pd(OAc)2. Added 5 wt% of heterogeneous catalyst relative to olefin 23,
40 psi H2, 16 h, 25 °C.
complex mixtures of by-products, probably due to the pres-
ence of an additional benzylic center, while employment of
ester 25 gave encouraging results with NBS, Br2, NIS and
Pyr2INO3. The best results were obtained in HOAc with par-
tial conversion into exclusively the syn-C-10 halide.
At this point, evidence from the Pyr2INO3 halogenations
indicated an auxiliary oxidant could be beneficial in driving
the halogenation to completion. Optimal conditions were
identified involving NBS and HNO3 in HOAc, resulting in
complete conversion into a mixture of C-10 syn-mono- and
dibrominated products 30a and 29, respectively (anti-
monobromide 30b was undetectable) (Scheme 9). A deter-
mination needed to be made whether the mono- or dibro-
minated product would offer a more effective path forward,
although obtaining selectivity for either species alone
proved challenging. The chemistry of the dibromide 29 was
Me
Pd2(dba)3
Naph Trost
ligand
O
O
N
H2
cat.
OBs
OBs
N
N
21
O
O
Bu4NBr
CH2Cl2
N
COOMe
COOMe
23
25
Me
O
O
N
OBs
OBs
HN
HN
O
O
N
COOMe
COOMe
28
27
n-butyl ether from reduction
of elimination by-product
elimination
by-product
Scheme 8 Optimized -allyl cyclization and hydrogenation
Our goal was to identify effective hydrogenation condi-
tions on the crude reaction mixture to produce ethyl-sub-
stituted product 25, rather than isolating vinyl intermediate
23, to ensure configurational stability of the C-6 stereocen-
ter.12 Initial attempts to perform hydrogenation with the re-
maining Pd catalyst led to only 4% product, and formation
of 6% of a new n-butyl ether by-product 28, resulting from
reduction of the diene 27 (Table 2). A screen of hydrogena-
tion conditions revealed that reduction of the desired
alkene took place in the presence of sulfided palladium on
carbon, while minimizing ring opened by-products and
avoiding epimerization. After hydrogenation, the optical
purity can be upgraded from 94% ee to >99% ee through
crystallization of the much less soluble racemate from i-
PrOAc. After filtration from the racemate, recrystallization
from isopropanol results in 74% isolated yield of compound
25 over two steps from acetate 21.
NBS
HNO3
HOAc
Et
O
N
OBs
N
25
O
COOMe
(EtO)2POH, NMM
MeOH, THF, 78%
from 25
Br
Br
29
+
Et
Et
O
O
OBs
OBs
N
N
O
O
N
COOMe
N
COOMe
syn Br
Br
anti
30a
30b
PFBA, MeOH
95% HPLC assay yield
Et
O
Et
O
OBs
OBs
N
N
+
O
O
NHPFB
NHPFB
N
N
Extensive screening of C-10 oxidation on ester 25 and
amide 26 was performed. A variety of reagents and condi-
tions were investigated, targeting a benzylic halide, alcohol,
ketone, or amine derivative.13 Use of amide 26 gave rise to
O
O
Br
Br
31b
31a
Scheme 9 Bromination of 25
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K

F
P. E. Maligres et al.
Special Topic
Synthesis
explored to determine whether a path involving an enam-
ine or ketone was viable. Reactions of dibromide with me-
thylamine were messy, while attempts to hydrolyze to the
ketone resulted mostly in ring-opened product. The dibro-
mide was also found to readily lose bromine (even oxidizing
HBr to Br2) to give mixtures of monobromide diastereo-
mers. This serendipitous observation revealed the possibili-
ty that the dibromide could be selectively debrominated
and offer a more robust path to the monobromide.
A screen of reducing agents showed diethyl phosphite to
be the most effective reducing agent, chemoselectively re-
ducing dibromide 29 to monobromides 30a and 30b with-
out overreduction to starting material 25. The 2:1 mixture
of monobromide esters 30a and 30b, respectively, could
then be converted into the corresponding PFB amides 31a
and 31b via direct Me3Al mediated ester amidation in the
presence of the benzenesulfonate and bromide (Scheme 9).
As seen previously (Scheme 9), reaction of MeNH2 with
the electrophilic C-10 center of amide 31a and 31b gave the
syn isomer, independent of the original stereochemistry at
C-10, consistent with the intermediacy of a quinone
methide. After further optimization, it was found that the
reaction of MeNH2 with the mixture of PFB amides 31a and
31b gave a 97:3 syn/anti ratio of the corresponding methyl-
amines, with concomitant cleavage of the benzenesulfon-
ate. Crystallization of 9a as the p-toluenesulfonic salt pro-
vided the syn product phenol in 80% overall yield from 31
with 99:1 syn/anti ratio (no loss of ee). The final coupling
was accomplished via a mixed anhydride intermediate, pro-
viding MK-1376 in 90% yield from penultimate 9a with
>99% dr, >99% ee, and >99% purity by HPLC (Scheme 10).
stitution by the pyrimidinone nitrogen to form a seven-
membered heterocycle, which is reversible, requiring con-
trol through a combination of modified reaction conditions
and an unusual post-reaction quench using Pd(OAc)2. A
benzylic bromination, followed by displacement with me-
thylamine via a quinone methide intermediate, serves as a
critical transformation to drive a stereoconvergent route to
the syn oxepane of MK-1376. This novel strategy offered
100-fold greater productivity and 50-fold less waste than
the initial medicinal chemistry route and laid the ground-
work for critical scale-up efforts.14
Assay yields were obtained using analytical standards prepared by re-
crystallization or preparative chromatography. All isolated yields re-
flect correction for purity based on quantitative HPLC assays. Solka-
floc was obtained from Seidler Chemical Co., Newark, NJ. All reagents
and solvents were used as received without further purification un-
less otherwise stated. Unless otherwise noted, all manipulations were
carried out under an inert atmosphere of nitrogen gas. In general,
glassware was not specially dried prior to use. Solvent switch refers to
constant volume distillation with continuous addition of the new sol-
vent. Ratios and percentages of solvents used for chromatography,
TLC, HPLC and workups are by volume. High-pressure liquid chroma-
tography (HPLC) analysis was performed with a Hewlett Packard se-
ries 1100 HPLC system; HPLC method for reported retention times:
Zorbax Eclipse plus C18 50 × 4.6mm 1.8m column eluted at
1.5 mL/min gradient 10→95% MeCN in 0.1% aq H3PO4 over 5 min then
hold isocratic with detection at 210 nm. Analytical TLC was per-
formed using Merck Kieselgel G60 F254 precoated plates (0.25 mm)
followed by visualization with UV light (254 nm), staining with iodine
vapor or staining with a solution of 14% ammonium molybdate and
0.5% ceric sulfate in 10% aqueous sulfuric acid then heat. Flash col-
umn chromatography was performed using silica gel (Merck, 70–230
mesh ASTM). Optical rotations were obtained with a Perkin–Elmer
241 polarimeter at 20 °C. Melting points were obtained with a Barn-
stead/Thermolyne MEL-TEMP® apparatus and are uncorrected. 1H
and 13C NMR chemical shifts are reported in ppm; coupling constants
are reported in Hz. 1H and 13C NMR spectra were recorded with
Bruker AM and AMX systems. Elemental analyses were obtained from
Shanghai Institute of Organic Chemistry Chinese Academy of Sciences
and Quantitative Technologies Inc, Whitehouse, NJ.
MeNH2
MeOH
then TsOH
Et
Et
O
O
OH
O
OBs
NHPFB
N
N
O
O
80%
NHPFB
N
N
O
Br
NHMe
31a + 31b
1:1 syn/anti
9a
9a 97:3 syn/anti crude
9a·TsOH 99:1 syn/anti
O
OH
(Z)-3-((4-((tert-Butyldimethylsilyl)oxy)but-2-en-1-yl)oxy)pro-
panenitrile
Et
O
N
Me2N
OH
O
N
A 75 L three-necked RB flask was charged with THF (27.5 L) and t-
BuOK (95wt%, 6.10 kg, 51.6 mol) at r.t. to give a white slurry. cis-2-
Butene-1,4-diol (4; 4.55 kg, 51.6 mol) was added as neat liquid over
0.5 hour maintaining the temperature below 20 °C. The resulting slur-
ry was aged at r.t. for 30 minutes and TBS-Cl (7.00 kg, 46.5 mol) dis-
solved in THF (3.5 L) was added over 0.5 hour, keeping the tempera-
ture below 30 °C. The reaction mixture was aged at r.t. for 1–2 hours
and quenched into a stirred mixture of water (25 L) and t-BuOMe (25
L). The layers were separated, the organic layer was washed with 5%
aq NH4Cl (18 L) and 5% aq NaCl (2 × 18 L). The t-BuOMe layer was con-
centrated to an oil in a 50 L three-necked RB flask. Heptane (20 L) was
added and the mixture was concentrated to an oil to afford the cis-2-
butene-1,4-diol mono-TBS ether,15 which was used without further
purification in the next step. 1H NMR spectroscopic analysis indicated
this crude oil contained 88 mol% of the mono-TBS ether (54.4 mol)
O
NHPFB
O
PivCl, NMM
91%
O
NMe
Me2N
O
MK-1376
Scheme 10 Stereoselective methylamine displacement
In conclusion, a concise asymmetric synthesis of MK-
1376 was developed, providing 11% overall yield from cis-
1,4-butenediol and inexpensive, readily available starting
materials and reagents, without the need for any chromato-
graphic purifications. The key feature of this synthesis is an
intramolecular asymmetric -allyl palladium complex sub-
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K

G
P. E. Maligres et al.
Special Topic
Synthesis
and 12 mol% of the bis-TBS ether. Acrylonitrile (68.1 mol, 4.46 L) and
DBU (0.684 L, 4.54 mol) were added at r.t. and the reaction mixture
was heated to 60 °C for 16 hours. The reaction mixture was cooled to
r.t., diluted with t-BuOMe (25 L), transferred into a 100 L extractor
and washed with 5% aqueous KH2PO4 (20 L). The layers were separat-
ed, the organic phase was washed with water (2 × 18 L) and concen-
trated to 10 L. The solution was solvent switched to MeOH to afford
the coupled nitrile (10.5 kg, 91% assay yield from cis-2-butene-1,4-
diol mono-TBS ether) as a solution in MeOH (30 L total volume),
which was used as is in the next step. A sample of the crude solution
was evaporated to provide a colorless oil.
3.70 (d, J = 1.25 Hz, 2 H), 3.67 (d, J = 1.38 Hz, 0.5 H), 3.58–3.66 (m,
2 H), 2.33–2.45 (m, 2 H), 2.02 (d, J = 1.25 Hz, 1 H), 1.24 (d, J = 7.15 Hz,
0.5 H), 0.88 (s, 9 H), 0.05 (s, 6 H).
LC-MS (EI, 70 eV): m/z [M + H]+ calcd for C19H35N2O7Si: 431.2; found:
431.3.
(Z)-Methyl 2-(2-((4-((tert-Butyldimethylsilyl)oxy)but-2-en-1-
yl)oxy)ethyl)-5-hydroxy-6-oxo-1,6-dihydropyrimidine-4-carbox-
ylate (17)
A 100 L three-necked RB flask was charged with o-xylene (55 L) and
heated to 143–145 °C (heating mantle, gentle reflux). The crude
DMAD adduct in xylene (7.40 kg assay, 16.7 mol, 50 wt% o-xylene
solution) from the previous reaction was added over 2 hours to the
hot o-xylene solution. The reaction mixture was aged 3 additional
hours at the end of the addition, allowed to cool slightly to 130 °C and
o-xylene (15 L) distilled under partial reduced pressure (internal temp
120–130 °C). The reaction mixture was allowed to cool to r.t. to give a
tan mobile slurry of crystallized pyrimidinone. The slurry was aged
overnight at r.t. Heptane (45 L) was added over 2 hours and then aged
for 2 hours. The slurry was filtered, and the cake was washed with 1:1
toluene/heptane (15 L) and then with heptane (10 L). The cake was
dried under vacuum and a stream of nitrogen for three days at r.t. to
give the TBS-pyrimidinone 17 (3.75 kg, >97 A%, 85 wt%, 47% yield) as
a white crystalline solid.
HPLC: tR = 4.92 min.
1H NMR (400 MHz, CDCl3):  = 5.68–5.80 (m, 1 H), 5.51–5.65 (m, 1 H),
4.25 (d, J = 6.02 Hz, 2 H), 4.13 (d, J = 6.27 Hz, 2 H), 3.65 (t, J = 6.40 Hz,
2 H), 2.61 (t, J = 6.40 Hz, 2 H), 0.91 (s, 9 H), 0.08 (s, 6 H).
13C NMR (100 MHz, CDCl3):  = 133.5, 126.1, 117.9, 67.0, 64.7, 59.6,
25.9, 19.0, 18.4, 5.12.
HRMS (ESI): m/z [M + H]+ calcd for C13H26NO2Si: 256.1733; found:
256.1732.
(E)-3-(((Z)-4-((tert-Butyldimethylsilyl)oxy)but-2-en-1-yl)oxy)-N′-
hydroxypropanimidamide (16)
A 75 L three-necked RB flask was charged with a crude solution of the
coupled nitrile from the previous step (10.42 kg, 40.8 mol in 30 L of
MeOH). Hydroxylamine (50% aqueous, 2.75 L, 44.9 mol) was added at
r.t. in one portion and the resulting solution was heated to 60 °C for 6
hours (99.8 A% conversion). The reaction mixture was cooled to r.t.
and transferred into an extractor containing t-BuOMe (35 L) and wa-
ter (35 L). The layers were separated (pH ~9) and the organic phase
was washed with 5% brine (2 × 18 L). The aqueous portion was back
extracted with t-BuOMe (15 L). The combined organic layers were
concentrated to an oil and solvent switched to MeOH (3 L/kg solution)
to afford a solution of the amidoxime, which was used as is in the
next step. A sample of the crude solution was evaporated to provide
16 as a colorless oil.
Rf = 0.15 (EtOAc/EtOH/HOAc, 15:4:1); mp 132–134 °C; HPLC: tR = 4.11
min.
1H NMR (400 MHz, CDCl3):  = 11.15 (br, 1 H), 10.68 (br, 1 H), 5.67–
5.73 (m, 1 H), 5.53–5.58 (m, 1 H), 4.20 (d, J = 6.53 Hz, 2 H), 4.12 (d, J =
6.40 Hz, 2 H), 4.01 (s, 3 H), 3.76 (t, J = 5.58 Hz, 2 H), 2.93 (t, J = 5.65 Hz,
2 H), 0.87 (s, 9 H), 0.04 (s, 6 H).
13C NMR (100 MHz, CDCl3):  = 169.73, 158.60, 150.14, 149.66,
133.45, 126.08, 125.94, 66.99, 66.82, 59.43, 53.51, 35.22, 25.87, 18.27,
–5.22.
HRMS (ESI): m/z [M + H]+ calcd for C18H31N2O6Si: 399.1951; found:
399.1944.
HPLC: tR = 2.60 min.
1H NMR (400 MHz, CDCl3):  = 5.60–5.65 (m, 1 H), 5.48–5.53 (m, 1 H),
4.94 (br, 2 H), 4.16 (d, J = 6.02 Hz, 2 H), 3.99 (d, J = 6.27 Hz, 2 H), 3.55
(t, J = 5.71 Hz, 2 H), 2.32 (t, J = 5.65 Hz, 2 H), 0.83 (s, 9 H), 0.00 (s, 6 H).
(Z)-Methyl 2-(2-((4-((tert-Butyldimethylsilyl)oxy)but-2-en-1-
yl)oxy)ethyl)-6-oxo-5-((phenylsulfonyl)oxy)-1,6-dihydropyrimi-
dine-4-carboxylate (19)
13C NMR (100 MHz, CDCl3):  = 153.67, 132.87, 126.55, 68.33, 66.82,
59.50, 31.91, 31.32, 25.94, –5.15.
HRMS (ESI): m/z [M + H]+ calcd for C13H29N2O3Si: 289.1947; found:
289.1944.
TBS pyrimidinone (7.20 kg, 85 wt%, 15.4 mol) was charged to a 100 L
RB flask. Pyridine (10.5 kg) was added at r.t., followed by the addition
of benzenesulfonyl chloride (3.70 kg, 20.9 mol) over 30 minutes while
maintaining the temperature at 25–30 °C. The solution was stirred at
25–30 °C for 2 hours. The batch was cooled to 20 °C and MeOH (21 L)
was added followed by water (17.5 L). The slurry was stirred at 28–
30 °C for 30 minutes. Water (17.5 L) was added at 20–25 °C. The slurry
was aged 30 minutes and filtered. The filter cake washed with
MeOH/water (3:5 v/v, 30 L). The cake was dried in a nitrogen stream
on the filter pot overnight to give TBS besylate 19 (10.3 kg, 71wt%,
88% yield) as a white crystalline solid.
Dimethyl 2-(((E)-(1-Amino-3-(((Z)-4-((tert-butyldimethylsilyl)-
oxy)but-2-en-1-yl)oxy)propylidene)amino)oxy)but-2-enedioate
A 75 L three-necked RB flask was charged with crude amidoxime 16
solution from the previous step (10.95 kg, 38 mol in 27 L of MeOH).
The reaction mixture was cooled to –20 °C and dimethyl acetylenedi-
carboxylate (DMAD, 4.67 L, 38 mol) was added over 45 minutes keep-
ing the temperature below 5 °C. The reaction mixture was concen-
trated, and solvent switched to o-xylene to make a 50 wt% solution of
the DMAD adduct (16.4 kg, 100% assay yield). A sample of the crude
solution was evaporated to provide a colorless oil containing the two
isomers in 1:1 ratio.
Rf = 0.52 (t-BuOMe); mp 112–115 °C; HPLC: tR = 4.93 min.
1H NMR (400 MHz, CDCl3):  = 11.19 (br, 1 H), 7.92 (d, J = 7.2 Hz, 2 H),
7.57–7.61 (m, 1 H), 7.45–7.49 (m, 2 H), 5.59–5.63 (m, 1 H), 5.42–5.45
(m, 1 H), 4.11 (d, J = 5.96 Hz, 2 H), 4.03 (d, J = 5.96 Hz, 2 H), 3.74 (s,
3 H), 3.67 (t, J = 5.65 Hz, 2 H), 2.83 (t, J = 5.65 Hz, 2 H), 0.78 (s, 9 H),
0.05 (s, 6 H).
13C NMR (100 MHz, CDCl3):  = 163.10, 158.83, 157.57, 146.04,
136.66, 135.41, 134.40, 133.80, 129.00, 128.61, 125.62, 67.12, 66.17,
59.45, 53.19, 35.27, 25.90, 18.30, –5.21.
HPLC: tR = 5.02 and 5.07 min.
1H NMR (400 MHz, CDCl3):  = 5.81 (d, J = 1.38 Hz, 0.2 H), 5.60–5.74
(m, 3 H), 5.48–5.59 (m, 1 H), 5.28 (br, 0.3 H), 4.21 (d, J = 6.02 Hz, 2 H),
4.03–4.14 (m, 2 H), 3.88 (d, J = 1.25 Hz, 0.5 H), 3.83–3.85 (m, 2 H),
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K

H
P. E. Maligres et al.
Special Topic
Synthesis
HRMS (ESI): m/z [M + H]+ calcd for C24H35N2O8SSi: 539.1883; found:
539.1874.
13C NMR (100 MHz, CDCl3):  = 162.99, 161.14, 158.17, 143.69,
136.55, 134.53, 134.15, 131.67, 129.08, 128.62, 118.77, 71.45, 66.69,
57.07, 53.26, 40.63.
HRMS (ESI): m/z [M + H]+ calcd for C18H19N2O7S: 407.0913; found:
(Z)-Methyl 2-(2-((4-Acetoxybut-2-en-1-yl)oxy)ethyl)-6-oxo-5-
((phenylsulfonyl)oxy)-1,6-dihydropyrimidine-4-carboxylate (21)
407.0907.
TBS besylate 19 (9.90 kg, 71wt%, 13.1 mol) and acetonitrile (18 L)
were charged to a 50 L RB flask. Concd HCl (95 mL, 1.14 mol) was add-
ed and the solution was stirred at 15–18 °C for 1 hour. The mixture
was washed with heptane (2 × 9 L) to remove TBS-OH. The acetoni-
trile layer containing the desilylated intermediate (HPLC: tR = 2.68
min) was charged to a 72 L RB flask and pyridine (190 mL, 2.36 mol)
was added. The solution was concentrated to approximately 15 L and
flushed with acetonitrile (18 L) by distillation under reduced pressure
(50 °C). Acetic anhydride (4.5 kg, 44.1 mol) was added and the solu-
tion was warmed to 75 °C for 3 hours. i-PrOH (30 L) was added and
the batch was concentrated under reduced pressure (50 °C) to remove
20 L solvent. i-PrOH was added to obtain a final volume of 55 L at
50 °C. The solution was treated with Darco-G60 (activated carbon,
900 g), stirred at 50 °C for 1 hour and filtered warm through a small
solka-floc pad into a 100 L RB flask. The filter cake was washed with i-
PrOH (10 L) at 50 °C. The combined filtrates were stirred, cooled to
30 °C and seeded with product 21 (1 g). The slurry was cooled to 20 °C
over 1 hour and then cooled to 2 °C and aged 1 hour. The slurry was
filtered and washed with i-PrOH (10 L) and heptane (10 L). The cake
was dried to give the acetate-besylate 21 (5.2 kg, >98wt%, 86% yield)
as a white crystalline solid.
(S)-Methyl 6-Ethyl-4-oxo-3-((phenylsulfonyl)oxy)-6,7,9,10-tetra-
hydro-4H-pyrimido[1,2-d][1,4]oxazepine-2-carboxylate (25)
5% Pd (S)/C (828 g) was added to the crude CH2Cl2 solution of allylated
product 23 from the previous step (2.03 kg, 5.00 mol assuming 100%
yield for the allylation). The slurry was charged to a 10 gallon stirred
autoclave. The reaction vessel was pressurized with hydrogen to 45
psi and heated at 30 °C until hydrogen uptake showed complete con-
version. This was repeated on a second batch. The two hydrogenation
batches were combined, t-BuOMe (2.5 L) was added and the mixture
was then stirred with MgSO4 (0.8 kg) and K2HPO4 (0.8 kg) for 30 min-
utes. The mixture was filtered through silica (6 kg). The silica pad was
washed with 9:1 CH2Cl2/t-BuOMe (25 L). The combined filtrates were
concentrated, and solvent switched to isopropyl acetate (13 L). The
mixture was cooled to 22 °C over 1 hour. The resulting slurry was fil-
tered on a pad of silica (0.25 kg) and the filter cake of racemate was
washed with i-PrOAc (4 L) to give, on drying, the racemic ethyl oxe-
panopyrimidinone (0.17 kg, 10–20% ee). The filtrates were concen-
trated, and solvent switched to i-PrOH (13 L). The mixture was treated
with Darco G60 (0.3 kg) at 80 °C for 1 hour. The mixture was filtered
hot through a pad of solka-floc and washed with hot i-PrOH (4 L) and
acetone (1.3 L). The combined filtrates were solvent switched to i-
PrOH at 60–80 °C and cooled to 23 °C over 3 hours. The slurry was
cooled to 10 °C over 2 hours and filtered. The cake was washed with i-
PrOH (3 L) at 10 °C. The cake was dried under nitrogen stream to give
the oxepanopyrimidinone 25 (3.186 kg, 95wt% purity, 74% from allyl-
ic acetate 21) as a white crystalline solid.
Rf = 0.20 (t-BuOMe); mp 71–73 °C; HPLC: tR = 3.31 min.
1H NMR (400 MHz, CDCl3):  = 11.39 (br, 1 H), 8.03 (t, J = 5.20 Hz, 2 H),
7.66–7.73 (m, 1 H), 7.53–7.61 (m, 2 H), 5.64–5.78 (m, 2 H), 4.62 (t, J =
2.80 Hz, 2 H), 4.18 (t, J = 2.80 Hz, 2 H), 3.83 (s, 3 H), 3.75–3.82 (m,
2 H), 2.96 (t, J = 5.80 Hz, 2 H), 2.07 (s, 3 H).
13C NMR (100 MHz, CDCl3):  = 171.25, 163.19, 158.85, 158.19,
146.26, 136.70, 135.32, 134.51, 129.98, 129.14, 128.67, 127.53, 66.69,
60.05, 53.28, 35.39, 21.03.
Rf = 0.40 (t-BuOMe); mp 99–103 °C; []D 57.7° (c = 1 in CHCl3); HPLC:
tR = 3.51 min.
1H NMR (400 MHz, CDCl3):  = 7.98–8.05 (m, 1 H), 7.65–7.72 (m, 1 H),
7.51–7.61 (m, 2 H), 5.12–5.17 (m, 1 H), 4.05–4.21 (m, 2 H), 3.83 (s,
3 H), 3.42–3.66 (m, 3 H), 3.14 (dd, J = 15.81, 4.39 Hz, 1 H), 1.97–2.10
(m, 1 H), 1.84–1.95 (m, 1 H), 0.90 (t, J = 7.47 Hz, 3 H).
13C NMR (100 MHz, CDCl3):  = 163.05, 160.63, 158.50, 143.33,
136.64, 134.48, 134.22, 129.07, 128.60, 70.84, 66.73, 57.80, 53.21,
40.95, 22.42, 10.59.
HRMS (ESI): m/z [M + H]+ calcd for C20H23N2O9S: 467.1124; found:
467.1115.
Methyl 4-Oxo-3-((phenylsulfonyl)oxy)-6-vinyl-6,7,9,10-tetrahy-
dro-4H-pyrimido[1,2-d][1,4]oxazepine-2-carboxylate (23)
A solution of the acetate besylate 21 (2.50 kg, 5.36 mol) and tetrabu-
tylammonium bromide (173 g, 0.536 mol) in CH2Cl2 (20 L) was de-
gassed with nitrogen. Degassed CH2Cl2 (6.8 L) was added to a mixture
of Pd2dba3 (37 g, 0.040 mol) and (R,R)-naphthyl Trost ligand (1R,2R)-
(+)-1,2-diaminocyclohexane-N,N′-bis(2-diphenylphosphino-1-naph-
thoyl) (15; 134 g, 0.169 mol). After stirring for 10 minutes, the cata-
lyst solution was transferred to the stirred reaction vessel, and the re-
action mixture was aged for about 8 h. Solid Pd(OAc)2 (60 g, 0.27 mol)
was added to quench the reaction. The resultant solution of allylated
product 23 was carried forward to the next step. A second batch was
run at the same scale. A sample of the crude solution was washed
with water, filtered through a plug of silica gel and crystallized from i-
PrOAc to provide 23 as a white crystalline solid.
HRMS (ESI): m/z [M + H]+ calcd for C18H21N2O7S: 409.1069; found:
409.1052.
(S)-Methyl10,10-Dibromo-6-ethyl-4-oxo-3-((phenylsulfonyl)oxy)-
6,7,9,10-tetrahydro-4H-pyrimido[1,2-d][1,4]oxazepine-2-carbox-
ylate (29) and (6S,10S)-Methyl 10-Bromo-6-ethyl-4-oxo-3-((phe-
nylsulfonyl)oxy)-6,7,9,10-tetrahydro-4H-pyrimido[1,2-
d][1,4]oxazepine-2-carboxylate (30a)
Oxepanopyrimidinone 25 (3.35 kg, 8.20 mol) was suspended in acetic
acid (24.6 L) in a 100 L four-neck RB flask. NBS (4.38 kg, 24.6 mol) was
added followed by concentrated nitric acid (2.46 L, 38.5 mol). The
mixture was warmed to 75 °C over 1 h and the resulting homogenous
orange mixture was maintained at 72–80 °C for 3 hours. The mixture
was concentrated under vacuum (50 Torr at 75–45 °C) to remove 6 L
orange distillate, leaving a pale-yellow residual solution. The mixture
was cooled to 35 °C, CH2Cl2 (12.3 L) and water (16.4 L) were added.
The organic layer was separated and the aqueous phase was back-ex-
tracted with CH2Cl2 (4.1 L). The combined organic phases were
washed with water (16.4 L) and 1.67 M aqueous K2HPO4 (24.5 L). The
Rf = 0.38 (t-BuOMe); mp 90–94 °C; HPLC: tR = 3.46 min.
1H NMR (400 MHz, CDCl3):  = 7.99–8.07 (m, 2 H), 7.65–7.75 (m, 1 H),
7.51–7.62 (m, 2 H), 6.07 (ddd, J = 17.41, 10.76, 3.70 Hz, 1 H), 5.88 (dd,
J = 3.51, 2.38 Hz, 1 H), 5.36 (dd, J = 10.79, 2.51 Hz, 1 H), 5.05 (dd, J =
17.44, 2.26 Hz, 1 H), 4.33 (dd, J = 13.93, 4.14 Hz, 1 H), 3.85 (s, 3 H),
4.04–4.16 (m, 1 H), 3.60–3.73 (m, 2 H), 3.46–3.58 (m, 1 H), 3.07 (dd,
J = 15.43, 4.89 Hz, 1 H).
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K

I
P. E. Maligres et al.
Special Topic
Synthesis
combined CH2Cl2 extracts were dried over MgSO4 (100 g), filtered and
concentrated. The residue was solvent switched to 4.1 L THF contain-
ing a 4:3 mixture of di-bromo/mono-bromo products 29 and 30a,
which was used directly in the next step. A sample of the crude solu-
tion was evaporated and purified by flash column chromatography on
silica gel (EtOAc/n-heptane = 1:4), which gave the dibromide 29 (97%
LCAP) as a white crystalline solid.
1 H), 4.01 (dd, J = 14.18, 1.38 Hz, 1 H), 3.77 (s, 3 H), 3.60 (d, J =
14.18 Hz, 1 H), 2.43 - 2.57 (m, 1 H), 1.95–2.11 (m, 1 H), 0.95 (t, J =
7.40 Hz, 3 H).
13C NMR (100 MHz, CDCl3):  = 162.53, 158.68, 157.63, 142.87,
136.45, 134.60, 134.57, 129.13, 128.51, 72.35, 71.25, 60.35, 53.16,
52.47, 22.85, 10.80.
HRMS (ESI): m/z [M + H]+ calcd for C18H20BrN2O7S: 487.0175; found:
487.0166.
Rf = 0.75 (t-BuOMe); mp 143–147 °C; []D = 181.5° (c = 1 in CHCl3);
HPLC: tR = 4.50 min.
1H NMR (400 MHz, CDCl3):  = 7.99–8.09 (m, 2 H), 7.69–7.77 (m, 1 H),
7.55–7.65 (m, 2 H), 5.04–5.12 (m, 1 H), 4.71 (d, J = 13.93 Hz, 1 H), 4.37
(dd, J = 14.18, 2.89 Hz, 1 H), 4.12 (d, J = 14.05 Hz, 1 H), 3.80 (s, 3 H),
3.68 (d, J = 14.18 Hz, 1 H), 2.40–2.48 (m, 1 H), 1.81–1.92 (m, 1 H), 0.96
(t, J = 7.47 Hz, 3 H).
13C NMR (100 MHz, CDCl3):  = 162.14, 158.72, 155.18, 141.98,
136.55, 134.58, 134.23, 129.14, 128.55, 80.59, 70.79, 66.39, 60.93,
53.15, 23.46, 10.90.
(6S)-10-Bromo-6-ethyl-2-((4-fluorobenzyl)carbamoyl)-4-oxo-
6,7,9,10-tetrahydro-4H-pyrimido[1,2-d][1,4]oxazepin-3-yl Ben-
zenesulfonate (31a and 31b)
CH2Cl2 (10.5 L) was charged in a 50 L round-bottom flask. 4-Fluoro-
benzylamine (578 g, 4.62 mol) was added and the slightly yellow
solution was degassed with N2 for 1 h. Me3Al (2 M, 2.3 L, 4.6 mol) was
added slowly to the benzylamine solution over 1 h. The solution was
aged at r.t. for 1 h. The diastereomeric mixture of syn/anti-monobro-
mides 30a and 30b (1.5 kg, 3.08 mol) dissolved in CH2Cl2 (6 L) was
added to the amine-Al complex solution over 20 min at 20–25 °C. The
reaction mixture was aged at r.t. for 1 h and was pumped into a 50 L
cylindrical vessel containing 1N HCl (15 L) at 10 °C. The relatively
large exotherm and rapid gas evolution (methane) was controlled by
the slow addition of the mixture to the aqueous HCl. The organics
were washed with 1N HCl (2 × 15 L), saturated NaHCO3 (15 L) and sat-
urated brine (15 L) at r.t. The organic layer containing the amide prod-
ucts 31a and 31b (1.69 kg, 95% HPLC assay yield) was concentrated
and solvent switched to methanol. The final volume of the MeOH
solution was approximately 25 L and the solution of the amides was
used in next step without further purification. A second batch was
run at the same scale. A sample of the crude solution was evaporated
and crystallized from i-PrOAc to provide a 2:1 mixture of 31a and 31b
as a white crystalline solid.
HRMS (ESI): m/z [M + H]+ calcd for C18H19Br2N2O7S: 566.9259; found:
566.9243.
(6S)-Methyl syn- and anti-10-Bromo-6-ethyl-4-oxo-3-((phenylsul-
fonyl)oxy)-6,7,9,10-tetrahydro-4H-pyrimido[1,2-d][1,4]oxaze-
pine-2-carboxylates (30a and 30b)
To the crude 4:3 mixture of di-bromo/mono-bromo oxepanopyrimid-
inones 29 and 30a in THF from the previous step was added MeOH
(16.4 L), NMM (631 g, 5.74 mol) and diethyl phosphite (741 g, 5.74
mol). The mixture was stirred for 1 hour at 25–40 °C. Water (19 L) was
added dropwise over 2 hours while the mixture was gradually cooled
to 17 °C. The resulting slurry was filtered, and the filter cake was
washed with 1:1 MeOH/water (16.4 L) and dried to provide the 2:1
diastereomeric mixture of syn/anti-monobromides 30a and 30b re-
spectively (3.16 kg, 78% from 25) (mp 135–143 °C). A sample of the
syn/anti-mixture was purified by flash chromatography on silica gel
(EtOAc/n-heptane = 1:4), which gave the syn monobromide 30a and
the anti monobromide 30b both as white crystalline solids. Both 30a
and 30b undergo isomerization over time to ultimately give a 95:5
mixture of 30a and 30b, respectively.
Mp 117–129 °C; []D 70.3° (c = 1 in CHCl3); HPLC: tR = 4.60 and 4.66
min.
1H NMR (400 MHz, DMSO-d6):  = 9.20 (t, J = 6.09 Hz, 1 H), 7.93 (dd,
J = 8.41, 1.13 Hz, 2 H), 7.77–7.85 (m, 1 H), 7.62–7.70 (m, 2 H), 7.30
(dd, J = 8.60, 5.71 Hz, 2 H), 7.10–7.20 (m, 2 H), 5.57 (d, J = 1.00 Hz,
1 H), 4.81–4.95 (m, 1 H), 4.13–4.31 (m, 4 H), 4.04–4.12 (m, 1 H), 3.79
(d, J = 13.80 Hz, 1 H), 2.23–2.41 (m, 1 H), 1.94–2.03 (m, 1 H), 0.83 (t,
J = 7.47 Hz, 3 H).
13C NMR (100 MHz, DMSO-d6):  = 161.74 (d, J = 241 Hz, 1 C), 161.53,
160.53, 158.80, 157.99, 146.12, 136.49, 135.28, 135.22, 132.53,
129.90 (d, J = 7 Hz, 1 C), 128.58, 115.46 (d, J = 21 Hz, 1 C), 71.95, 70.49,
59.75, 54.37, 42.01, 22.75, 22.75, 10.82.
syn-Monobromide 30a
Rf = 0.60 (t-BuOMe); mp 143–145 °C; []D 188.8° (c = 1 in CHCl3);
HPLC: tR = 4.08 min.
1H NMR (400 MHz, CDCl3):  = 8.02 (d, J = 7.78 Hz, 3 H), 7.67–7.74 (m,
1 H), 7.53–7.63 (m, 2 H), 5.29–5.36 (m, 1 H), 5.00 (br, 1 H), 4.36 (dd,
J = 14.24, 3.20 Hz, 1 H), 4.24 (dd, J = 14.12, 2.82 Hz, 1 H), 4.02 (dd, J =
14.18, 1.25 Hz, 1 H), 3.60 (d, J = 14.31 Hz, 1 H), 3.81 (s, 3 H), 2.45–2.61
(m, 1 H), 2.04–2.10 (m, 1 H), 0.97 (t, J = 7.47 Hz, 3 H).
19F NMR (376 MHz, DMSO-d6):  = –115.84.
HRMS (ESI): m/z [M + H]+ calcd for C24H24BrFN3O6S: 580.0553; found:
13C NMR (100 MHz, CDCl3):  = 162.55, 158.69, 157.59, 142.92,
136.51, 134.67, 134.55, 129.11, 128.54, 72.39, 71.31, 60.35, 53.19,
52.38, 22.86, 10.80.
580.0548.
(6S,10S)-6-Ethyl-N-(4-fluorobenzyl)-3-hydroxy-10-(methylami-
no)-4-oxo-6,7,9,10-tetrahydro-4H-pyrimido[1,2-d][1,4]oxazepine-
2-carboxamide 4-Methylbenzenesulfonate (9a·TsOH)
HRMS (ESI): m/z [M + H]+ calcd for C18H20BrN2O7S: 487.0175; found:
487.0161.
The methanol solution of the crude amides 31a and 31b from the pre-
vious step (3.0 kg, 5.17 mol) was charged to a 100 L flask and cooled
to 5 °C. Methylamine solution in EtOH (33 wt%, 8 M, 3.23 L, 41.4 mol)
was added over 15 minutes. The solution was aged 30 minutes at 5–
10 °C, warmed to 20 °C and aged for about 4 hours until reaction was
complete. A solution of toluenesulfonic acid (3.93 kg, 20.7 mol) in wa-
ter (20 L) was added to the batch. The batch was seeded with salt
product 9a·TsOH (1 g), and aged 30 minutes. Water (9 L) was added
over 30 minutes, aged for 30 minutes and additional water (60 L) add-
anti-Monobromide 30b
Rf = 0.64 (t-BuOMe); mp 140–142 °C; []D 193.2° (c = 1 in MeOH);
HPLC: tR = 4.10 min.
1H NMR (400 MHz, CDCl3):  = 7.95–8.06 (m, 2 H), 7.67–7.74 (m, 1 H),
7.51–7.61 (m, 2 H), 5.32 (dd, J = 2.70, 1.44 Hz, 1 H), 4.93–5.02 (m,
1 H), 4.33 (dd, J = 14.24, 3.20 Hz, 1 H), 4.22 (dd, J = 14.18, 2.89 Hz,
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K

J
P. E. Maligres et al.
Special Topic
Synthesis
ed over 1 hour. The slurry was aged at r.t. for 1 hour, filtered, washed
with 3:1 water/MeOH (10 L) and dried on a filter pot for 2 days with
nitrogen flow. The crude dry cake was slurried in EtOAc (30 L) for 2
hours at r.t., filtered and washed with EtOAc (10 L). The cake was
dried overnight to give amine salt 9a·TsOH (2.32 kg, 80%) as a white
crystalline solid.
net and R. Miller for alternate route exploration and scale-up lab
work; G. Black, C. K. Esser, D. Henderson, J. F. Leone, J. Liu, L. Tan and
D. Von Langen for scale-up of original route; M. Biba and T. Brkovic
and Y. Liu for analytical support; R. Reamer for NMR support; S. D.
Dreher, E. N. Guidry, S. Shultz, D. Steinhuebel, D. M. Tellers and M.
Tudge for chemocatalysis, high-throughput screening and automation
support; F. Fleitz for biocatalysis support; E. Choi and C. K. Maguire
for molecular and materials characterization support; K. Emerson, E.
Fisher and I. Lee for calorimetry and engineering support. We thank
Jianjun Duan and Guiquan Liu at WuXi AppTec Co., Ltd. for characteri-
zation of intermediates. We are grateful to R. C. Ruck, P. G. Bulger, D.
M. Tschaen and LC. Campeau for their useful edits, suggestions, and
managerial support.
Mp 202–204 °C; []D 31.6° (c = 1 in CHCl3); HPLC: tR = 4.49 min (TsOH
1.10 min).
1H NMR (400 MHz, DMSO-d6):  = 12.42 (br. s., 1 H), 9.63 (d, J =
5.40 Hz, 1 H), 9.29 (br. s., 1 H), 8.85 (br. s., 1 H), 7.49 (d, J = 7.53 Hz,
2 H), 7.38 (dd, J = 8.28, 5.77 Hz, 2 H), 7.06–7.23 (m, 4 H), 5.10 (br. s.,
1 H), 4.41–4.62 (m, 3 H), 4.00–4.17 (m, 2 H), 3.74–3.94 (m, 2 H), 2.71
(br. s., 3 H), 2.29 (s, 3 H), 2.14 (dt, J = 13.58, 6.70 Hz, 1 H), 1.77–1.94
(m, 1 H), 0.97 (t, J = 7.34 Hz, 3 H).
13C NMR (100 MHz, DMSO-d6):  = 167.82, 161.57 (d, J = 242 Hz, 1 C),
157.89, 147.89, 145.25, 138.07, 134.55, 129.38 (d, J = 8 Hz, 1 C),
128.17, 125.48, 123.81, 115.19 (d, J = 22 Hz, 1 C), 67.91, 64.98, 60.50,
31.78, 22.00, 20.80, 11.24.
Supporting Information
Supporting information for this article is available online at
https://doi.org/10.1055/s-0040-1707994.
S
u
p
p
orit
n
gInformati
o
n
S
u
p
p
orit
n
gInformati
o
n
19F NMR (376 MHz, DMSO-d6):  = –115.56.
HRMS (ESI): m/z [M + H]+ calcd for C19H24FN4O4: 391.1782; found:
391.1775.
References
(1) New address: Arrowhead Biopharmaceuticals, 502 South Rosa
Road, Madison, WI 53719, USA
(2) New address: Colgate Palmolive Company, 909 River Rd, Piscat-
away, NJ 08854, USA
(3) (a) Kallings, L. O. J. Intern. Med. 2008, 263,03 218. (b) Global
AIDS Monitoring 2019 Joint United Nations Programme on
HIV/AIDS (UNAIDS) WHO; UNAIDS/JC2941; ISBN 978-92-9253-
088-4. (c) HIV/AIDS Surveillance Report - 2017; Centers for
Disease Control and Prevention, 2017; Vol. 29.
Anal. Calcd for C26H31FN4O7S: C, 55.51; H, 5.55; F, 3.38; N, 9.96; S,
5.70; Found: C, 55.51; H, 5.60; F, 3.14; N, 9.81; S, 5.53.
MK-1376
A 50 L flask was charged with N,N-dimethyloxamic acid (0.66 kg, 5.64
mol), CH2Cl2 (14 L), and NMM (1.13 kg, 11.2 mol). The mixture was
cooled to 15 °C. Pivaloyl chloride (0.64 kg, 5.31 mol) was added and
aged 2 hours at r.t.. Amine salt 9a·TsOH (2.01 kg, 3.58 mol) was added
in one portion and aged 2 hours. The mixture was quenched with wa-
ter (12 L) and 5 M HCl (0.21 L, 1.05 mol). The lower organic layer was
separated and washed with water (2 × 12 L). The solution was con-
centrated to low volume and i-PrOH (8 L) was added. The solution was
heated to 55–60 °C, seeded with MK-1376 (24 g, crystalline Form II),
solvent switch continued at 55–60 °C by adding i-PrOH to maintain
the volume at approximately 12 L. The slurry was aged at 55–60 °C for
6 hours, cooled to r.t. and filtered. The cake was washed with i-
PrOH/heptane (10 L) and heptane (10 L) and dried in vacuo at 40 °C to
afford MK-1376 as a white crystalline solid (1.62 kg, >98 wt% purity,
91% yield).
(4) (a) Humphrey, G. R.; Pye, P. J.; Zhong, Y.-L.; Angelaud, R.; Askin,
D.; Belyk, K. M.; Maligres, P. E.; Mancheno, D. E.; Miller, R. A.;
Reamer, R. A.; Weissman, S. A. Org. Process Res. Dev. 2011, 15,
73. (b) For details of the thermal rearrangement mechanism of
DMAD adduct, see: Pye, P. J.; Zhong, Y.-L.; Jones, G. O.; Reamer,
R. A.; Houk, K. N.; Askin, D. Angew. Chem. Int. Ed. 2008, 47, 4134.
(5) (a) Zhong, Y.-L.; Pipik, B.; Lee, J.; Kohmura, Y.; Okada, S.; Igawa,
K.; Kadowaki, C.; Takezawa, A.; Kato, S.; Conlon, D.; Zhou, H.;
King, A. O.; Reamer, R. A.; Gauthier, D. R. Jr.; Askin, D. Org.
Process Res. Dev. 2008, 12, 1245. (b) Zhong, Y.-L.; Krska, S. W.;
Zhou, H.; Reamer, R. A.; Lee, J.; Sun, Y.; Askin, D.; Ferrara, M. B.
Org. Lett. 2009, 11, 369. (c) Ferrara, M.; Crescenzi, B.; Donghi,
M.; Muraglia, E.; Nizi, E.; Pesci, S.; Summa, V.; Gardelli, C. Tetra-
hedron Lett. 2007, 48, 8379.
(6) Isaacs, R. C. A.; Thompson, W. J.; Williams, P. D.; Su, D.-S.;
Venkatraman, S.; Embrey, M. W.; Fisher, T. E.; Wai, J. S.; Dubost,
D. C.; Ball, R. G.; Choi, E. J.; Pei, T.; Trice, S. L.; Campbell, N.;
Maddess, M.; Maligres, P. E.; Shevlin, M.; Song, Z. J.; Steinhuebel,
D. P.; Strotman, N. A.; Yin, J. US 20100087419, 2010.
(7) Subsequent studies using conditions optimized for asymmetric
allylation showed only racemic product with this substrate.
(8) Toteva, M. M.; Richard, J. P. Adv. Phys. Org. Chem. 2011, 45, 39.
(9) Impact of L/M ratio and effect of Bu4NBr in Trost allylation has
been previously reported, see: (a) Trost, B. M.; Crawley, M. L.
Chem. Rev. 2003, 103, 2921. (b) Trost, B. M. Tetrahedron 2015,
71, 5708. (c) Trost, B. M.; Machacek, M. R.; Aponick, A. Acc.
Chem. Res. 2006, 39, 747.
Rf = 0.50 (CH2Cl2/MeOH, 9:1); mp 142–144 °C; HPLC: tR = 3.57 min
(anti isomer 3.72 min).
1H NMR (500 MHz, CDCl3):  = 12.30 (s, 1 H), 9.63 (t, J = 6.3 Hz, 1 H),
7.37 (m, 1 H), 6.97 (m, 1 H), 5.82 (br s, 1 H), 4.62 (m, 1 H), 4.54 (d, J =
6.3 Hz, 2 H), 4.43 (m, 1 H), 4.20 (d, J = 3.1 Hz, 2 H), 4.13 (dd, J = 11.8,
2.7 Hz, 1 H), 3.07 (s, 3 H), 3.02 (s, 3 H), 2.85 (s, 3 H), 2.06 (m, 1 H), 1.83
(m, 1 H), 1.31 (t, J = 7.4 Hz, 3 H).
13C NMR (125 MHz, CDCl3):  = 168.75, 166.74, 165.09, 162.47 (d, JCF
=
244.9 Hz), 159.45, 147.85, 144.38, 134.76 (d, JCF = 3.1 Hz), 130.30 (d,
JCF = 8.0 Hz), 125.39, 115.57 (d, JCF = 21.3 Hz), 67.89, 64.98 (br), 60.96,
54.78 (br), 42.76, 37.45, 34.00, 33.46 (br), 25.07, 11.70.
HRMS (ESI): m/z [M + H]+ calcd for C23H29FN5O6: 490.2102; found:
490.2093.
(10) Even under these optimized conditions for the -allyl cycliza-
tion, Boc-methylamine species 6 still gave only racemic trans-
product.
Acknowledgment
We thank: C. Bazaral, R. S. Hoerner, T. Houck, and M. Weisel from the
high pressure lab for their work on catalytic hydrogenation; M. Jour-
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K

K
P. E. Maligres et al.
Special Topic
Synthesis
(11) Interestingly, when 1 equiv K2CO3 was added to the asymmetric
allylation reaction, the rate increased and the sense of enanti-
oselectivity completely switched from 96% ee (R) to –45% ee (S).
This reversal in selectivity could signal a change in mechanism
from one involving attack of the pyrimidone on Pd followed by
reductive elimination to one involving intermolecular attack of
the pyrimidone anion directly on the Pd-bound allyl fragment.
(12) No hydrogenation reaction was observed in dichloromethane.
(13) During this oxidation screen an interesting unprecedented oxi-
dation was discovered using Tl(NO3)3 in HOAc, which oxidized
the oxepane to the olefin (Scheme 11). This olefin could then be
converted into the dibromide, which gave the vinyl bromide
regioselectively upon base-mediated elimination. Further
pursuit of this method was not attempted due to the toxicity of
thallium; but, it was encouraging to note traces of the desired
C-10 halogenated compound were observed after halogenations
in the presence of Tl(NO3)3 using NBS, NCS and t-BuOCl. At this
point, evidence from the thallium(III) catalyzed and Pyr2INO3
halogenations indicated an auxiliary oxidant may be beneficial
in driving the halogenation to completion.
(14) Process mass intensity (PMI) was decreased from >60,000 to
1200.
(15) Bhunia, N.; Das, B. Synthesis 2015, 47, 1499.
Et
Et
O
N
O
Et
Et
O
N
O
N
Tl(NO3)3
HOAc
Br2
DCM
OBs
OBs
OBs base
COOMe
OBs
N
N
N
N
O
O
O
O
O
COOMe
N
O
COOMe
COOMe
Br
Br
Br
Et
iodination
with I+
OSO2Ph
COOMe
exclusively
NIS/HOAc: 25% conversion
[Pyr2I]+ [NO3]–/HOAc: 40% conversion
N
syn
N
I
Scheme 11
© 2020. Thieme. All rights reserved. Synthesis 2020, 52, A–K