RCM macrocyclization made practical: An efficient synthesis of HCV protease inhibitor BILN 2061

Article abstract of DOI:10.1021/ol800183x

We report here that dramatic improvement of the key RCM reaction in the synthesis of HCV protease inhibitor BILN2061 can be achieved by N-substitution of the diene substrate with an electron-withdrawing group. Mechanistic studies using ¹H NMR spectroscopy showed an unprecedented switch of the initiation sites and the correlation between such switch and the results of RCM, from the unmodified to the modified substrates. We also provided theoretical evidence that such modification may also increase the thermodynamic preference of the macro cyclic product over the diene substrate.

Full text of DOI:10.1021/ol800183x

ORGANIC
LETTERS
2008
Vol. 10, No. 6
1303-1306
RCM Macrocyclization Made Practical:
An Efficient Synthesis of HCV Protease
Inhibitor BILN 2061
Chutian Shu,*,† Xingzhong Zeng,† Ming-Hong Hao,‡ Xudong Wei,† Nathan K. Yee,†
Carl A. Busacca,† Zhengxu Han,† Vittorio Farina,†,§ and Chris H. Senanayake†
Department of Chemical DeVelopment and Department of Medicinal Chemistry,
Boehringer Ingelheim Pharmaceuticals, 900 Ridgebury Road,
Ridgefield Connecticut 06877
chutian.shu@boehringer-ingelheim.com
Received January 25, 2008
ABSTRACT
We report here that dramatic improvement of the key RCM reaction in the synthesis of HCV protease inhibitor BILN2061 can be achieved by
N-substitution of the diene substrate with an electron-withdrawing group. Mechanistic studies using 1H NMR spectroscopy showed an
unprecedented switch of the initiation sites and the correlation between such switch and the results of RCM, from the unmodified to the
modified substrates. We also provided theoretical evidence that such modification may also increase the thermodynamic preference of the
macrocyclic product over the diene substrate.
Macrocycles are prevalent structural motifs in natural
products, and they are increasingly popular scaffolds in drug
design.1 For example, new potent hepatitis C protease
inhibitors containing a 15-membered macrocyclic lactam
have been recently shown to be clinically active.2 Thus, the
development of macrocyclization strategies,3 without the use
of impractical high-dilution conditions, is of great interest
to both industrial and academic scientists.4 A useful concept
that has been employed to describe the difficulty of forming
large rings is that of “effective molarity” (EM), which
depends on the size of the ring to be formed and also on
any conformational constraints that may affect the system.5
(2) (a) Tsantrizos, Y. S.; Bolger, G.; Bonneau, P.; Cameron, D. R.;
Goudreau, N.; Kukolj, G.; LaPlante, S. R.; Llina`s-Brunet, M.; Nar, H.;
Lamarre, D. Angew. Chem., Int. Ed. 2003, 42, 1356. (b) Faucher, A.-M.;
Bailey, M. D.; Beaulieu, P. L.; Brochu, C.; Duceppe, J.-S.; Ferland, J.-M.;
Ghiro, E.; Gorys, V.; Halmos, T.; Kawai, S. H.; Poirier, M.; Simoneau, B.;
Tsantrizos, Y. S.; Llina`s-Brunet, M. Org. Lett. 2004, 6, 2901. For another
macrocycle-based HCV inhibitor, see: (c) Venkatraman, S.; Njoroge, F.
G.; Girijavallabhan, V. M.; Madison, V. S.; Yao, N. H.; Prongay, A. J.;
Butkiewicz, N.; Pichardo, J. J. Med. Chem. 2005, 48, 5088.
(3) Parenty, A.; Moreau, X.; Campagne, J.-M. Chem. ReV. 2006, 106,
911.
(4) (a) Namba, K.; Kishi, Y. J. Am. Chem. Soc. 2005, 127, 15382. (b)
Fraunhoffer, K. J.; Prabagaran, N.; Sirois, L. E.; White, M. C. J. Am. Chem.
Soc. 2006, 128, 9032.
(5) (a) Illuminati, G.; Mandolini, L. Acc. Chem. Res. 1981, 14, 95. (b)
Galli, C.; Mandolini, L. Eur. J. Org. Chem. 2000, 3117. For a recent study
of EM in RCM, see: (c) Conrad, J. C.; Eelman, M. D.; Silva, J. A. D.;
Monfette, S.; Parnas, H. H.; Snelgrove, J. L.; Fogg, D. E. J. Am. Chem.
Soc. 2007, 129, 1024.
† Department of Chemical Development.
‡ Department of Medicinal Chemistry.
§ Current Address: Johnson & Johnson Pharmaceutical Research and
Development, Turnhoutseweg 30, B-2340 Beerse, Belgium.
(1) (a) Weissjohann, L. A.; Ruijter, E.; Garcia-Rivera, D.; Brandt, W.
Mol. DiVersity 2005, 9, 171. (b) Roxburgh, C. J. Tetrahedron 1995, 51,
9767. (c) Craik, D. J.; Cemazar, M.; Daly, N. L. Curr. Opin. Drug DiscoVery
DeV. 2006, 9, 251.
10.1021/ol800183x CCC: $40.75
© 2008 American Chemical Society
Published on Web 02/23/2008

substituent at the C-4 position of the hydroxyproline moiety
had a small but detectable effect on the RCM rate and
therefore on the EM. This small effect was tentatively
ascribed to subtle conformational factors. Second, we noticed
that, when the initiation of the reaction was monitored using
a stoichiometric amount of Grubbs catalyst 3, carbene
transfer occurred to a large extent (96%) at the vinylcyclo-
propane moiety, where the Ru may be stabilized by chelation.
Such stabilization, in turn, may reduce the concentration of
the active Ru catalyst in the reaction and negatively affect
the rate of the RCM step.7c
Scheme 1. Approach to 1 by RCM Macrocyclization
To exploit these clues, we decided to prepare a number
of derivatives in which the amide bond had been protected
with various removable groups. We expected that such
substitution may interrupt the coordinative stabilization by
the ester group through steric interaction (A1,3 strain) and
therefore lead to initiation at the nonenoic acid moiety, which
may be beneficial to the RCM.8 Gratifyingly, our first insight
found experimental confirmation (Scheme 2).
In recent years, the ring-closing metathesis (RCM) reaction
has emerged as a very attractive approach to the synthesis
of macrocycles, due to its neutral reaction conditions, broad
functional group tolerance, and simple reaction procedure.6
Nonetheless, implementation of RCM macrocyclizations
in an industrial setting is extremely problematic and very
uncommon. A recent review highlights the fact that, in the
large number of macrocyclizations that have been effected
by RCM, the substrate concentration range was 0.2-8.5 mM,
and the catalyst loading 2-10 mol %.6c Our recently
disclosed process for BILN 2061, (1, Scheme 1) which was
scaled to >100 kg, suffered from high dilution (10 mM) and
large catalyst loading (3-5 mol %), thus rendering the
approach unsuitable for production-scale manufacturing.7
We are pleased to report here our solution to this problem,
that is, the development of an RCM macrocyclization that
can operate at “acceptable” plant concentrations (>0.2 M)
and at low catalyst loadings (e0.1 mol %), thus making this
the first example of a truly practical RCM macrocyclization.
Our approach stems from various initial observations on
our early RCM process: First, we noticed that the remote
The resting state of the Grubbs catalyst (3) was readily
determined by 1H NMR spectroscopy as already reported.7c
Indeed, whereas N-benzyl substitution (4c) had a relatively
minor effect on the site of initiation, acylation of the N atom
led to inhibition of carbene transfer to that position. Instead,
carbene transfer took place completely (>98%) at the
nonenoic acid moiety (5b and 5d were produced in >90%
conversion).
With these initial observations on hand, we proceeded to
attempt the RCM of substrate 4b under standard conditions
(10 mM, toluene, 60 °C), but employing a second-generation
Ru catalyst 7.9 The desired RCM took place with an initial
rate that was 3-4 times faster than that for substrate 4a under
identical conditions, leading to the desired product in
quantitative yield (>98%), without formation of dimers. This
is a significant improvement over the RCM reaction of 4a
(Vide infra). Reactions with our key substrates were then
Scheme 2. RCM Initiation with Modified Substrates
1304
Org. Lett., Vol. 10, No. 6, 2008

tested, under what we presume to be thermodynamic con-
ditions,10 on the basis of our previous work (Scheme 3 and
Table 1).
Table 1. Effect of Concentration on RCM of 4a-da
concn
(M)
temp
(oC)
yield
(convn)
entry
substrate
7, mol %
1
2
3
4
5
6
7
8
4a
4a
4a
4a
4b
4b
4b
4b
4b
4b
4b
4b
4c
4c
4d
4d
0.01
0.02
0.05
0.10
0.01
0.02
0.05
0.10
0.05
0.10
0.20
0.40
0.01
0.10
0.01
0.10
1
1
1
1
1
1
1
1
0.1
0.1
0.1
0.1
1
1
1
0.1
60
60
60
60
60
60
60
60
110
110
110
110
60
110
60
110
82 (95)
70 (91)
52 (80)
35 (72)
98 (98)
97 (98)
87 (92)
80 (87)
97 (100)
95 (100)
93 (100)
80 (99)
85 (95)
42 (77)
99 (100)
89 (98)
Scheme 3. RCM Reaction of Modified Substrates.
9
10
11
12
13
14
15
16
a Reactions were run on 1 mmol scale. Yields were determined by
quantitative HPLC assay. Conversion refers to starting material consumed
(HPLC assay). See Supporting Information for details.
As entries 1-4 show, in the case of 4a, the yield of the
RCM product 8 decreases as the substrate concentration
increases. 4b, on the other hand, operating at a 10-fold higher
concentration (0.10 M, entry 8), provided essentially the same
yield as the best result obtained with 4a. Furthermore, by
running the reaction at higher temperatures,11 both the yield
and the EM increased, and we were able to lower the catalyst
loading to 0.1%. Entry 11 demonstrates that we haVe
achieVed our goal of operating with e0.1 mol % catalyst,
as well as under standard concentrations (g0.2 M). At even
higher concentration (0.4 M, entry 12), up to 20% yield of
dimer was detected, therefore limiting the yield of the
RCM.12 The N-benzyl derivative 4c behaved similarly to 4a,
whereas the acetyl derivative 4d mirrored 4b. These data
indicate that the increase of EM by N-substitution with an
electron-withdrawing group was closely related to the shift
of initiation from 6 to 5.13
The reversible nature of metathesis and high reactivity of
7, however, suggest that at least part of the effect of the
electron-withdrawing substituent is to increase the thermo-
dynamic EM, possibly by reducing the ring strain. Indeed,
both amide and the cyclopropane ring represent trans
elements, which strain the ring system in 8a. It is possible
that removal of the enforced planarity in 8a by N-Boc
substitution may reduce ring strain and improve the ther-
modynamic EM. Thus, a theoretical analysis was carried out
to (1) calculate the strain energy of the macrocycle and (2)
gain an understanding of the conformational characteristics
of the open chain and ring molecules.
The strain energy content of the macrocycle was calculated
by first determining the conformational energy change
between the open chain molecules with and without Boc
substitution, 4a and 4b, and then comparing with the energy
change between ring molecules 8a and 8b with the same
chemical modifications. The difference between these two
energy changes, that is, ∆∆E, is the contribution of Boc
substitution to the strain energy of the molecule. Because
we were only interested in the strain energy of the core
structure, we omitted the PNB group from our calculations
for simplicity. The conformational energy of the open chain
and ring molecules was calculated by molecular mechanics
(MM) and quantum chemical (QM) DFT methods. MM
methods are computationally efficient. Using the powerful
torsion-scan/low-mode search14 MC algorithm from the
Macromodel program,15 we identified the unique energy
(6) (a) Grubbs, R. H., Ed. Handbook of Metathesis; Wiley-VCH:
Weinheim, 2003. (b) Fu¨rstner, A.; Langemann, K. J. Org. Chem. 1996, 61,
3942. (c) Gradillas, A.; Pe´rez-Castells, J. Angew. Chem., Int. Ed. 2006, 45,
6086. (d) Fu¨rstner, A.; Davies, P. W. Chem. Commun. 2005, 42, 2307. (e)
Kulkarni, A. K.; Diver, S. T. J. Am. Chem. Soc. 2004, 126, 8110.
(7) (a) Yee, N. K.; Farina, V.; Houpis, I. N.; Haddad, N.; Frutos, R. P.;
Gallou, F.; Wang, X.-J.; Wei, X.; Simpson, R. D.; Feng, X.; Fuchs, V.;
Xu, Y.; Tan, J.; Zhang, L.; Xu, J.; Smith-Keenan, L. L.; Vitous, J.; Ridges,
M. D.; Spinelli, E. M.; Johnson, M.; Donsbach, K.; Nicola, T.; Brenner,
M.; Winter, E.; Kreye, P.; Samstag, W. J. Org. Chem. 2006, 71, 7133. (b)
Nicola, T.; Brenner, M.; Donsbach, K.; Kreye, P. Org. Process Res. DeV.
2005, 9, 513. (c) Zeng, X.; Wei, X.; Farina, V.; Napolitano, E.; Xu, Y.;
Zhang, L.; Haddad, N.; Yee, N. K.; Grinberg, N.; Shen, S.; Senanayake,
C. H. J. Org. Chem. 2006, 71, 8864.
(8) Change of initiation site is known to affect RCM. See: (a) Wallace,
D. J. Angew. Chem., Int. Ed. 2005, 44, 1912. (b) Hoye, T. R.; Jeffrey, C.
S.; Tennakoon, M. A.; Wang, J.; Zhao, H. J. Am. Chem. Soc. 2004, 126,
10210.
(9) Michrowska, A.; Bujok, R.; Haruytyuyan, S.; Sashuk, V.; Dolgonos,
G.; Grela, K. J. Am. Chem. Soc. 2004, 126, 9318. We are fully aware of
the fact that initiation with the Grela catalyst may be different from that
with catalyst 3. On the other hand, only catalysts containing two strong
donor ligands give rise to observable intermediates, and those with IMes
and SIMes ligands cyclize too quickly to make the carbene transfer products
observable.
(10) Our previous studies on diene 4a showed the reaction to be readily
reversible under these conditions. See ref 7a.
(11) Yamamoto, K.; Biswas, K.; Gaul, C.; Danishefsky, S. J. Tetrahedron
Lett. 2003, 44, 3297.
(12) The main cyclic dimeric product was isolated, and separately
converted to 8b under standard RCM conditions. See supporting information
for details on the assignment of the structure as well as the conversion of
8b to 1.
(13) Visser, M. S.; Heron, N. M.; Didiuk, M. T.; Sagal, J. F.; Hoveyda,
A. H. J. Am. Chem. Soc. 1996, 118, 4291.
(14) Kolossvary, I.; Guida, W. C. J. Comput. Chem. 1999, 20, 1671.
(15) Macromodel, v 9.1; Schro¨dinger, Inc.: New York, 2006.
Org. Lett., Vol. 10, No. 6, 2008
1305

minima and the lowest-energy conformations of the relevant
molecules. However, since the parametrizations of current
force fields are not fully validated for the complex, polar-
lipophilic macrocycles like the present system, the calculated
energies obtained with the MM method may not reflect the
true strain energy well. Therefore, we further used QM DFT
methods to calculate the energies of the most representative
conformation for the ring and open chain molecules. Table
2 summarizes the calculated strain energy reduction in the
with the experimental results. By analyzing the QM DFT
optimized structures, we determined that the zero point
energy of the molecules does not make a net contribution to
the ∆G change in the Boc-substituted macrocycle, thus
minimizing entropic contributions.
In summary, we have developed a practical synthesis of
BILN 2061 by introducing the first example of practical
RCM macrocyclization. The origins of the “N-Boc-effect”17
seem to be grounded in favorable kinetic and thermodynamic
effects. We have shown that strategic introduction of
removable groups on the RCM linker can direct the initiation
site and have a remarkable effect on the RCM, thus
complementing the known relay strategy,8 and can also affect
the strain content in the product, thus dramatically increasing
the thermodynamic EM. Generalization of these important
effects could lead to widespread utilization of the RCM and
other macrocyclizations in the industrial plant.
Table 2. Calculated Reduction of Strain Energy by Boc
Substitution on Molecule 8b in Comparison to 8a
method
OPLS01a MM3a MMFFsa DFT/B3LYPb
-3.33 -1.99 -1.10 -2.18
energy change
(kcal/mol)
a Calculated from Boltzmann average of the energy minima of the chain
and ring molecules sampled by Macromodel program in water. Solvation
energy was calculated by the generalized Born model provided in the
Macromodel program. b Calculated from full geometry optimization of
single structure for each molecule in gas phase with the Jaguar program.16
The basis set used in the calculation was 6-31G*. The calculated gas-phase
strain energy change is -4.26 kcal/mol. Calculation with Jaguar program
using continuous solvation model gave a difference of solvation energy of
2.08 kcal/mol. The sum of gas-phase strain energy and solvation energy is
shown in the table.
Acknowledgment. We thank Prof. Scott Denmark of the
Department of Chemistry of the University of Illinois for
useful suggestions.
Supporting Information Available: Experimental Pro-
cedures for all compounds described in the paper, the con-
version from 8b into 1, and detailed information on calcula-
tions. This material is available free of charge via the Internet
at http://pubs.acs.org.
OL800183X
macrocycle due to Boc substitution. Our calculations using
a number of theoretical methods invariably indicate that Boc
substitution at the P1/P2 N atom reduces the strain energy
of the ring molecule by ∼2 kcal/mol, which is consistent
(16) Jaguar, v 6.5; Schro¨dinger, Inc.: New York, 2006.
(17) We prefer the Boc group over other acyl groups like the acetyl
because Boc groups are easily removed from the product, thus making the
N-Boc a “protecting-activating moiety” in this case.
1306
Org. Lett., Vol. 10, No. 6, 2008