Synthesis of Grazoprevir, a Potent NS3/4a Protease Inhibitor for the Treatment of Hepatitis C Virus

Article abstract of DOI:10.1021/acs.orglett.8b03173

An efficient synthesis of grazoprevir is reported. Starting from four readily available building blocks, grazoprevir is prepared in 51% overall yield and >99.9% purity for pharmaceutical use.

Full text of DOI:10.1021/acs.orglett.8b03173

Letter
pubs.acs.org/OrgLett
Cite This: Org. Lett. XXXX, XXX, XXX−XXX
Synthesis of Grazoprevir, a Potent NS3/4a Protease Inhibitor for the
Treatment of Hepatitis C Virus
Feng Xu,* Jungchul Kim, Jacob Waldman, Tao Wang, and Paul Devine
Department of Process Research and Development, MRL, Merck & Co., Inc., Rahway, New Jersey 07065, United States
S
* Supporting Information
ABSTRACT: An efficient synthesis of grazoprevir is reported. Starting from four readily
available building blocks, grazoprevir is prepared in 51% overall yield and >99.9% purity for
pharmaceutical use.
backbone of the macrocyclic ring, we envisioned exploring
several alternative protocols for C−C bond coupling with
building block 2, fitting our goal to develop a cost-effective and
environmentally responsible process for the efficient prepara-
tion of grazoprevir. To this end, we recently reported a
practical synthesis of the synthetically challenging trans-
cyclopropoxy building block 2, with a suitable functional
group of choice in the cyclopropoxy ring,3 which enabled us to
explore various C−C coupling strategies including sp2−sp3,
sp2−sp2, and sp2−sp couplings and olefin metathesis, as
outlined in Scheme 1. As such, a series of synthetic strategy
permutations to form the macrocyclic ring, via C−C coupling
or amidation, were accessible for investigation. This article
describes the successful realization of the most practical
strategy to assemble the described building blocks, with
minimal functional group manipulation/transformation, result-
ing in an efficient and high-yielding synthesis of grazoprevir
that achieves the high purity standards in the pharmaceutical
industry for commercialization.
The strategy for closure of the grazoprevir macrocycle was
ultimately responsible for the sequence of events and types of
reactions employed to assemble building blocks 2, 3, and 4.
The chemistry applied to discover grazoprevir utilized a ring
closing metathesis (RCM) reaction to form the macrocyclic
skeleton (Scheme 2).2b The RCM process suffered a low yield
(25%) due to the binding affinity of the catalyst with substrate
6 to form a stable chelated Ru complex.4 Further investigation
of the RCM strategy showed that this catalyst sequestering
could be overcome using substrate 8 with an extended,
pendant alkene chain off the quinoxaline moiety, affording a
significant improvement of the RCM yield (71%).4 However,
on the basis of an overall cost consideration, the application of
RCM to construct grazoprevir failed to provide the necessary
benefits suitable for large scale preparation.
ince the discovery of the hepatitis C virus (HCV) in
1
S1989, scientific advances in understanding of the HCV
life cycle have led to the evolution of a series of direct-acting
antiviral HCV therapies, resulting in significant improvement
on virus cure rate and shortening treatment duration for
diverse patient populations. Grazoprevir (1),2 a potent NS3/4a
protease inhibitor, in combination with the HCV NS5a
inhibitor elbasvir, was recently approved as the novel
therapeutic Zepatier for the treatment of HCV by the FDA
and EMA.
Grazoprevir (1) features an 18-membered macrocycle and
seven stereogenic centers. Retrosynthetic analysis of grazopre-
vir, employing multiple bond disconnection strategies, logically
led to four building blocks, as shown in Scheme 1. Taking
advantage of commercially available N-Boc-hydroxyproline
ester 4, it appeared synthetically attractive to establish the
ether linkage between the hydroxyproline moiety and the
grazoprevir skeleton via an SNAr displacement of chloroqui-
noxaline 3 (X = Cl), while building blocks 2 and 5 could both
be introduced by amidation reactions. To construct the
Scheme 1. Retrosynthetic Analysis of Grazoprevir
Received: October 4, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.8b03173
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
Scheme 2. Key Steps in the Early Syntheses of Grazoprevir
Scheme 3. Through-Process to Proline Chloroquinoxaline
13 and Macrocyclization via sp2−sp3 Coupling
The synthesis developed to support the early safety and
clinical studies of grazoprevir relied upon a macrolactamization
to form the macrocycle 11 (Scheme 2).5 While the initial
strategy for the preparation of precursor 10 through a
Sonogashira coupling followed by hydrogenation proved
effective, the liability of this synthesis was the incompatible
reactivity of the electron-rich quinoxaline ring in 10 with the
various conditions attempted for the requisite removal of the
Boc group. Numerous impurities were formed through
reaction of the quinoxaline moiety with cationic species
generated under the deprotection conditions, adversely
impacting the synthetic efficiency, yield, and overall process
robustness. Given this observation, it was evident that removal
of the Boc group, introduced from commercially available
proline ester 3, should be carried out earlier in the synthesis to
suppress these side reactions, following formation of the SNAr
product 12 (Scheme 3), in which the quinoxaline moiety
maintained its electron-deficient nature. In addition, we
envisioned that access to the unprotected proline intermediate
13 from 12 would provide the opportunity to explore more
promising synthetic permutations to prepare grazoprevir, as
the building block carboxylic acid 2 could be incorporated in
the core macrocyclic ring via amidation followed by a
transition metal catalyzed C−C coupling with the chloroqui-
noxaline moiety or vice versa.
Interestingly, this early Boc deprotection approach created a
new set of challenges. While the initial SNAr reaction
proceeded smoothly to form chloroquinoxaline 125 in 88%
yield and 95:5 regioselectivity, it was observed that the
electron-deficient nature of the quinoxaline ring in the desired
product 13 rendered it susceptible to competitive SNAr
reactivity at the remaining chloride position. When the Boc
deprotection of 12 was carried out under acidic aqueous
conditions or in alcohol solvents, SNAr substitution of the
chloride in the quinoxaline ring by the nucleophilic solvent
(water or alcohols) to form the corresponding quinoxalinol or
ether became significant.6 Ultimately, employment of MeSO3H
in the non-nucleophilic solvent, acetonitrile, proved crucial to
developing a high-yielding Boc deprotection and afforded the
salt 13 that crystallized directly from the reaction mixture as an
acetonitrile solvate. With this result, a robust, two-step
through-process of an SNAr reaction followed by Boc
deprotection was developed, from which the salt 13 was
isolated in 78% yield7 with good rejection of the regioisomer
(98:2), while the chloro-substituted, electron-deficient qui-
noxaline ring remained intact.
With salt 13 in hand, we explored the feasibility of
constructing the backbone of the 18-membered macrocycle
via intramolecular sp2−sp3 coupling.8,9 Selective hydroboration
of amide 14 with pinacolborane in the presence of a catalytic
amount of [Ir(cod)Cl]2 /dppe afforded the boronate 15, which
was subjected to Suzuki−Miyaura coupling conditions to
afford macrocycle 11 in 22% yield. Alternatively, treatment of
9a
boronate ester 15 with KHF2 yielded the corresponding
potassium salt 16 that underwent cyclization via Molander’s
photoredox conditions in only 10% yield. Unfortunately,
attempts to improve the cyclization yield by either pathway
through extensive screening of Pd and Ni sources and ligands
under numerous conditions proved unsuccessful.
Given the lack of progress with sp2−sp3 couplings, we turned
our attention to applying an sp2−sp coupling to install building
block 2 (R = HCC(CH2)2) on quinoxaline 13 (Scheme
4).10 With the nitrogen in the proline moiety unprotected, the
thermostability of free base 13, which was generated in situ
under the basic conditions required for sp2−sp coupling,
proved concerning. In fact, attempts to apply the Sonogashira
conditions5 developed for the preparation of compound 10
resulted in complete decomposition of 13. Further studies
showed that use of an alcohol solvent was vital to develop a
mild and efficient Sonogashira reaction addressing the thermo-
instability of free base 13, presumably the reduction
capability11 of alcohol solvent improved the catalyst stability
and reactivity, which was evidenced by the significantly
increased catalytic turnover number at mild temperature
(Table 1, entry 2).11 Carrying out the Sonogashira coupling
in other solvents under various conditions resulted in an
unsatisfactory yield and/or poor conversion. With methanol as
solvent, we were able to decrease the reaction temperature to
B
DOI: 10.1021/acs.orglett.8b03173
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
impurities that would require rejection in the final crystal-
lization. Early efforts for the coupling of acid 19 with amine 5
were carried out in acetonitrile in the presence of EDC·HCl
and pyridine.5 However, this process was demonstrated to lack
operational robustness and impurity control due to a series of
significant competitive side reactions. In particular, to ensure
isolating quality grazoprevir, a large excess of EDC·HCl (1.65
equiv) was required to achieve >99% conversion, because
unconsumed 19 was poorly rejected under the isolation
conditions. Furthermore, while the use of excess EDC·HCl
resulted in high conversions, these conditions likewise led to
the rapid formation of several byproducts that were also poorly
rejected during the isolation. It was therefore crucial to quench
the reaction as soon as the desired conversion had been
achieved. Upon failing to do so, the quantities of the
byproducts increased through continuous reaction of product
1 with EDC and resulted in concomitantly poor quality
grazoprevir. This delicate balance between the side reactions vs
rejection of impurities vs conversion rendered the control of
product quality more challenging and necessitated additional
studies to understand the pathways by which the multiple
impurities were forming.
Scheme 4. Construction of Macrocyclic Core via
Sonogashira Coupling−Macrolactamization
Table 1. Selected Results of Sonogashira Coupling of
Chloroquinoxaline 13 with Alkyne Acid 2 (R = HC
C(CH2)2)
These studies exploring the competitive side reactions of
EDC with grazoprevir led to identification of byproducts 20,
23, and 24, the structures of which were unambiguously
elucidated using NMR spectroscopy and LC/MS techniques. A
plausible pathway for the formation of 23 and 24 via a novel
rearrangement is depicted in Scheme 5. The EDC adduct
a
a
entry
1
catalysts
solvents
MeCN
cond
conv
yield
3 mol % (Ph3P)2PdCl2,
6 mol % CuI
50 °C,
5 h
>98%
85%
98%
78%
−
2
3
4
0.1 mol % (Ph3P)2PdCl2, MeOH
0.5 mol % CuI
35 °C,
>99%
84%
5 h
1 mol % XantPhos-Pd-
G2, 30 mol % CuI
AcNMe2 35 °C,
18 h
Scheme 5. Endgame and Plausible Pathways of Byproducts
4 mol % (Ph3P)2PdCl2,
30 mol % CuI
AcNMe2 35 °C,
<20%
20 h
a
Determined by HPLC analysis.12
35 °C and the catalyst loading to as low as 0.05 mol % of air
stable (Ph3P)2PdCl2, one of the least expensive palladium
catalysts, together with 0.1 mol % CuI. Under the optimized
conditions, the Sonogashira coupling of 13 and 2 (R = HC
C(CH2)2) proceeded smoothly in the presence of i-Pr2NEt to
yield the desired product 17 in 98% assay yield.
Without isolation of alkyne 17, the Sonogashira reaction
crude stream was solvent switched to THF/AcNMe2 and
treated with HATU to afford the desired macrolactam 18 in
84% isolated yield and >97% purity without any epimerization
observed.13 The corresponding regioisomer carried from 13
was rejected to <0.2%. Hydrogenation of alkyne 18 in the
presence of 5% Pd(OH)2/C in THF at 50 psi gave the desired
11 in >99% conversion. Upon removal of the Pd catalyst and
concentration, macrocycle 11 was isolated from aqueous i-
PrOH in 87% yield (Scheme 4, Option A). Alternatively, for
Option B, which featured the reverse sequence of hydro-
genation followed by macrolactamization, it was essential to
treat the crude stream 17 with charcoal prior to hydrogenation
to remove impurities poisoning the Pd catalyst in order to
achieve the desired complete reduction (≥99%)14 of alkyne 17
to the corresponding alkane. Although the subsequent
macrolactamization proceeded well, Option A provided a
more robust process and was selected as the macrocyclization
process en route to grazoprevir.
guanidine 20 was shown to form in ∼30% yield when the
reaction mixture was aged overnight. [1,3]-Carbonyl migra-
tion15 would lead to the formation of intermediate 21.
Subsequent intramolecular nucleophilic attack would yield
the electron-deficient guanidine 22 that could undergo
fragmentation via a nucleophilic amine attack to afford 23
and 24.16
After hydrolysis of ester 11 to the corresponding acid 19, it
was critical to identify the optimized conditions for the final
amide coupling, while understanding degradation pathways
that would both compromise the overall yield and form
C
DOI: 10.1021/acs.orglett.8b03173
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
ORCID
With this mechanistic insight in hand, a simple solution to
suppress the formation of 20 and subsequent 23 and 24 was
identified. By switching from the polar solvent acetonitrile to a
much less polar solvent THF, which was believed to attenuate
the nucleophilicity and/or acidity of the acyl sulfonamide
moiety in 1,17 the nucleophilic addition to form adduct 20 was
dramatically suppressed. Indeed, even aging the EDC reaction
mixture in THF at ambient temperature for 1 week resulted in
<2% of impurities 20, 23, and 24.
Further optimization of the EDC coupling conditions was
governed by the stability/reactivity of sulfonamide 5. Unlike
the stable pTSA salt 5, the free base 5 proved to be unstable at
ambient temperature,18 with ∼40% observed decomposition
overnight (Scheme 6).12 This instability of the free base 5 was
Feng Xu: 0000-0003-1949-324X
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We gratefully acknowledge the following colleagues of Merck
Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc.,
Kenilworth, NJ, USA, M. Maust, J. Hill, and M. Weisel of the
High Pressure Laboratory and K. Belyk for experimental
assistance, L. DiMichele, R. Reamer, and Dr. A. Buevich for
assistance with NMR studies, Dr. J. Jo for HRMS analysis, and
Dr. R. T. Ruck and D. Andrews for useful discussions.
Scheme 6. Decomposition of Sulfonamide 5
REFERENCES
■
(1) Choo, Q. L.; Kuo, G.; Weiner, A. J.; Overby, L. R.; Bradley, D.
W.; Houghton, M. Science 1989, 244, 359−362.
(2) (a) Lawitz, E.; Gane, E.; Pearlman, B.; Tam, E.; Ghesquiere, W.;
Guyader, D.; Alric, L.; Bronowicki, J. P.; Lester, L.; Sievert, W.;
Ghalib, R.; Balart, L.; Sund, F.; Lagging, M.; Dutko, F.; Shaughnessy,
M.; Hwang, P.; Howe, A. Y.; Wahl, J.; Robertson, M.; Barr, E.; Haber,
B. Lancet 2015, 385, 1075−1086. (b) Harper, S.; McCauley, J. A.;
Rudd, M. T.; Ferrara, M.; DiFilippo, M.; Crescenzi, B.; Koch, U.;
Petrocchi, A.; Holloway, M. K.; Butcher, J. W.; Romano, J. J.; Bush, K.
J.; Gilbert, K. F.; McIntyre, C. J.; Nguyen, K. T.; Nizi, E.; Carroll, S.;
S; Ludmerer, S. W.; Burlein, C.; DiMuzio, J. M.; Graham, D. J.;
Stahlhut, M. W.; Olsen, D. B.; Monteagudo, E.; Cianetti, S.; Giuliano,
C.; Pucci, V.; Trainor, N.; Fandozzi, C. M.; Rowley, M.; Coleman, P.
J.; Vacca, J. P.; Summa, V.; Liverton, N. ACS Med. Chem. Lett. 2012,
3, 332−336.
overcome by lowering the reaction temperature to 0−5 °C and
by charging pyridine to a slurry of azeotropically dried acid 19
and 5 pTSA salt in THF. Subsequent treatment of the resulting
homogeneous solution with EDC gave the desired 1 with an
excellent purity profile, while the decomposition of 5 was
suppressed to a minimal level (<1%). Of note is the extreme
ease of operation of this endgame process at large scale without
over-reaction/stability concerns. Upon aqueous citric acid
workup, grazoprevir monohydrate was crystallized from
aqueous acetone as its desired pharmaceutical form19 in 94%
yield and >99.8% purity with no impurities (>0.1%).20
In summary, an efficient synthesis of grazoprevir (1) has
been described. The building blocks 2−5 are installed in an
optimal and efficient sequence, eliminating the incompatible
reactivity of the intermediates encountered in the alternative
syntheses. The application of an effective Sonogashira coupling
using the air-stable (Ph3P)2PdCl2−CuI catalyst mixture served
to construct the key macrocyclic backbone. The development
of an optimal endgame process was accomplished with
mechanistic understanding providing the requisite insight
into helping suppress the competitive undesired reaction
pathways. Starting from dichloroquinoxaline 3 (X, Y = Cl),
with four isolated intermediates, grazoprevir was prepared in
51% overall yield and >99.8% purity for pharmaceutical use. By
combining all the key elements required for a manufacturing
process, this route has been successfully implemented for
commercialization of grazoprevir.20
(3) Xu, F.; Zhong, Y.-L.; Li, H.; Qi, J.; Desmond, R.; Song, Z. J.;
Park, J.; Wang, T.; Truppo, M.; Humphrey, G. R.; Ruck, R. T. Org.
Lett. 2017, 19, 5880−5883.
(4) Williams, M. J.; Kong, J.; Chung, C. K.; Brunskill, A.; Campeau,
L.-C.; McLaughlin, M. Org. Lett. 2016, 18, 1952−1955.
(5) Kuethe, J.; Zhong, Y.-L.; Yasuda, N.; Beutner, G.; Linn, K.; Kim,
M.; Marcune, B.; Dreher, S. D.; Humphrey, G.; Pei, T. Org. Lett.
2013, 15, 4174−4177.
(6) Presumably, protonation of the chloroquinoxaline nitrogen
increased its electron-deficient nature and sufficiently lowered the
barrier for SNAr reactivity. For the same reason, acidic aqueous
workup of 13 proved to be undesirable.
(7) MeCN solvate 13 underwent desolvation upon drying.
(8) The use of intermolecular sp2−sp3 coupling is not attractive, due
to the instability of the corresponding coupling product and free base
13 under these coupling conditions (vide infra in the text).
(9) (a) Primer, D. N.; Karakaya, I.; Tellis, J. C.; Molander, G. A. J.
Am. Chem. Soc. 2015, 137, 2195−2198. (b) Li, H.; Scott, J. P.; Chen,
C.; Journet, M.; Belyk, K.; Balsells, J.; Kosjek, B.; Baxter, C. A.;
Stewart, G. W.; Wise, C.; Alam, M.; Song, Z. J.; Tan, L. Org. Lett.
2015, 17, 1533−1536.
(10) Alternative strategies via intramolecular Heck or Sonogashira
couplings resulted in unsatisfactory results, due to selectivity
challenges and/or cost of ligands. Kong, J.; McLaughlin, M.;
Williams, M.; Song, Z. J.; Chen, Y.; Jin, Y. PCT Int. Appl. WO
2015095437, 2015.
(11) (a) Xu, F.; Kosjek, B.; Cabirol, F. L.; Chen, H.; Desmond, R.;
Park, J.; Gohel, A. P.; Collier, S. J.; Smith, D. J.; Liu, Z.; Janey, J. M.;
Chung, J. Y. L.; Alvizo, O. Angew. Chem., Int. Ed. 2018, 57, 6863−
6867. (b) Zacuto, M. J.; Xu, F. J. Org. Chem. 2007, 72, 6298−6300.
(12) For details, see the Supporting Information.
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.or-
glett.8b03173.
Experimental procedure, characterization data, and
discussion (PDF)
(13) Intermolecular amidation was inevitable; however, it was
significantly suppressed under the optimal lactamization conditions.
The resulting oligomeric byproducts could be rejected to a
nondetectable level in the subsequent isolated intermediates.
(14) Due to poor rejection of partially reduced alkene intermediate.
AUTHOR INFORMATION
Corresponding Author
■
*E-mail: feng_xu@merck.com.
D
DOI: 10.1021/acs.orglett.8b03173
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
(15) Graubaum, H.; Csunderlik, C.; Glatt, H.-H.; Bacaloglu, R.;
Malurea-Munteanu, M. Z. Chem. 1984, 24, 57−58.
(16) Formation of 23 was observed when amide 27 was treated with
EDC.
(17) (a) Reichardt, C.; Welton, T. Solvent Effects in Organic
Chemistry, 4th ed.; Wiley-VCH: Weinheim; 2011. (b) This solvent
effect was also confirmed by varying the ratio of AcNMe2 in THF.12
(18) Amine 5 has been used for amidation on large scale; however,
studies on its stability/reactivity were lacking. Flick, A. C.; Ding, H.
X.; Leverett, C. A.; Kyne, R. E., Jr.; Liu, K. K.-C.; Fink, S. J.;
O’Donnell, C. Bioorg. Med. Chem. 2016, 24, 1937−1980.
(19) In the previous process,5 it was necessary to recrystallize the
initially isolated grazoprevir to obtain the desired pharmaceutical
form.
(20) The entire process starting from 3 has been successfully carried
out on large scale (>100 kg).
E
DOI: 10.1021/acs.orglett.8b03173
Org. Lett. XXXX, XXX, XXX−XXX