A polar radical pair pathway to assemble the pyrimidinone core of the HIV integrase inhibitor raltegravir potassium

Article abstract of DOI:10.1002/anie.200703681

(Chemical Equation Presented) Break up to make up: Combined experimental and computational studies provide evidence that the key step in the synthesis of a novel anti-HIV drug involves an unprecedented stepwise radical pair rearrangement mechanism in which radical fragments are held together by strong electrostatic forces (see scheme); this is favored over alternative mechanisms involving concerted pericyclic rearrangement.

Full text of DOI:10.1002/anie.200703681

Communications
DOI: 10.1002/anie.200703681
Mechanistic Studies
A Polar Radical Pair Pathway To Assemble the Pyrimidinone Core of
the HIV Integrase Inhibitor Raltegravir Potassium**
Philip J. Pye,* Yong-Li Zhong,* Gavin O. Jones, Robert A. Reamer, Kendall N. Houk,* and
David Askin
The UN and WHO currently estimate that 40 million people
are living with HIV/AIDS worldwide.[1] Combination thera-
pies using reverse transcriptase and protease inhibitors have
been highly successful, but the emergence of drug resistance
prompts a search for alternative treatments. A rapidly
expanding area of HIV/AIDS research targets a third
enzyme, integrase, the catalyst responsible for the integration
of proviral DNA into the host cell chromosome.[2] Research at
Merck has led to the identification of raltegravir potassium
(1),[3,4] a potent and well-tolerated HIV-1 integrase inhibitor
Scheme 1. Synthesis of raltegravir potassium. Cbz=benzyloxycarbonyl.
that targets strand transfer, the second of two catalytic cycles
mediated by the integrase enzyme.[5]
Raltegravir potassium (1) consists of a central hydroxy-
pyrimidinone heterocyclic core that is rapidly assembled by a
two-component coupling reaction between amidoxime 2 and
dimethyl acetylenedicarboxylate (DMAD) to give 3Z/3E,
followed by a thermal rearrangement to give product 4
(Scheme 1).[6] Amidoxime–DMAD adduct isomers 3Z and
3E are converted to hydroxypyrimidinone 4 at significantly
different rates: 3Z, the major isomer, reacts at a temperature
of 1258C within 2 h, while 3E only rearranges at an elevated
temperature of 1358C within 5 h. Plausible mechanisms for
this reaction involving [1,3] and [3,3] shifts[5a–b] are shown in
Scheme 2. Plausible [1,3]- and [3,3]-rearrangement mechanisms.
Scheme 2. 15N-enriched rearrangement precursors 3Z*/3E*
(65:35) were synthesized. Thermolysis of 3Z*/3E* resulted in
the formation of hydroxypyrimidinone 4’*, in which the 15N
label was unexpectedly found to be exclusively at the position
ortho to the ester substituent. Since this outcome is not
consistent with a [3,3]-sigmatropic rearrangement mechanism
(4’’*), a formal [1,3] rearrangement must have occurred.
An in-depth experimental and computational study of the
mechanism has now shown that the apparently attractive
concerted [3,3]-pericyclic route is eschewed in favor of a
[*] Dr. P. J. Pye, Dr. Y.-L. Zhong, R. A. Reamer, Dr. D. Askin
Department of Process Research
Merck Research Laboratories
À
direct N O cleavage to form a polar radical pair (PRP) with a
substantial preference for recombination. Quantum mechan-
ical calculations were used to determine how models 6Z and
6E, with CO2Me replacing CBz, undergo transformation to 5,
a model for pyrimidinone 4. B3LYP density functional
theory,[7] as implemented in Gaussian03,[8] was used to
compute optimized geometries of intermediates and transi-
tion states. Energies were obtained from a calibration
Rahway, NJ 07065 (USA)
Fax: (+1)732-594-5170
E-mail: philip_pye@merck.com
yongli_zhong@merck.com
Dr. G. O. Jones, Prof. K. N. Houk
Department of Chemistry and Biochemistry
University of California, Los Angeles
Los Angeles, CA 90095-1569 (USA)
Fax: (+1)310-206-1843
involving
a smaller computational model where G3-
(MP2)B3,[9] a high-accuracy method, and B3LYP energetics
were compared (other functionals and methods involving
CCSD(T) were also compared, see Supporting Information
for details).
E-mail: houk@chem.ucla.edu
[**] We are grateful to the NSF (CHE-0548209 to K.N.H.) and the NIH
(GM36700 to K.N.H.) for financial support. Computational re-
sources from NSF-PACI and the UCLA ATS are appreciated. We
thank T. J. Novak for mass spectrometric assistance, and P. Bulger
for helpful discussions.
As summarized in Scheme 3, the tautomerization of
amidoxime adducts 6Z/E to 7Z/E is endothermic by
ꢀ 12 kcalmolÀ1. Acid-base catalysis is required to form 7,
since concerted [1,3]-hydrogen shifts have prohibitively high
barriers. Although the activation energies of the allowed
[3,3] shifts of 7Z/E are 25 kcalmolÀ1 (cf. TS5), these rear-
Supporting information for this article (experimental details and
B3LYP/6-31+G(d,p) Cartesian coordinates, G3(MP2)B3 results
and full reference list) is available on the WWW under http://
www.angewandte.org or from the author.
4134
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 4134 –4136

Angewandte
Chemie
Scheme 3. Corrected B3LYP energies and relevant transition structures and intermediates in the possible rearrangement mechanisms of
amidoximes 6Z and 6E. All energies [kcalmolÀ1] are referenced to 6Z and 6E. Pictures are for structures in the stepwise rearrangement of 6E.
(N.D.=not determined.)
rangements do not occur because the tautomeric reactants are
not formed in toluene. The activation energies for the
[1,3] shifts of the amidino fragments (TS3, TS4) of 6Z/E
and 7Z/E are 40–46 kcalmolÀ1, typical of thermally forbidden
radical pair (see below) are consistent only with the radical
pair mechanism.
Alternative stepwise mechanisms, involving the formation
of 1,2-oxazetidine[11] or 1,2,4-oxadiazole intermediates fol-
lowed by rearrangement, can potentially lead to [1,3]-
rearrangement products as shown in Scheme 4. These nucle-
ophilic mechanisms were found to be significantly disfavored
compared to the polar radical pair pathway (see Supporting
Information for details).
À
[1r,3s]-sigmatropic shifts, and are nearly the same as N O
homolysis activation barriers of 6Z/E.[10] The bond dissocia-
tion energies for the complete separation of the amidino and
vinyloxy fragments in 6Z/E are ꢀ 42 kcalmolÀ1, several
kcalmolÀ1 lower than the barriers for [1,3] shifts.
Transition structures were found for diradical formation
(TS1) from 6Z/E. Significantly, TS1–6Z is 3 kcalmolÀ1 lower
in energy than TS1–6E, in excellent agreement with the
experimentally favored reactivity of 3Z in comparison with
3E. PRP is a tightly hydrogen-bound radical pair due to the
polar nature of each radical fragment. It is 8 kcalmolÀ1 more
stable than TS1–6E, and the transition structure for collapse
of the radical pair to product is essentially identical in energy.
The geometries of TS1 and TS3 are quite similar, being
dissociative in nature; these differ most significantly in the
orientation of the migrating amidino moiety in relation to the
vinoxyl moiety (see Supporting Information for further
details). Given the significant error bars associated with the
computed barriers, the calculations do not definitively differ-
entiate between reaction via TS3 and radical pair formation
for one of these substrates (6E). However, the relative
reactivities computed for 6Z/6E, which are similar to the
experimental results, and the intramolecular trapping of a
An experimental probe for the presence of radical
intermediates resulting from thermally induced homolytic
À
cleavage of the N O bond was derived by incorporating an
Scheme 4. a)–c) Plausible stepwise nucleophilic mechanisms.
Angew. Chem. Int. Ed. 2008, 47, 4134 –4136ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
4135

Communications
alkene into a model substrate to act as a potential intra-
of the anti-HIV drug raltegravir potassium. It is likely that
these intermediates, which are held together by strong
electrostatic attractions, rapidly collapse leading to formation
of the heterocyclic product.
molecular radical trap (Scheme 5).[12] When vinyl amidoxime
mixture 10Z/10E (6:1) was heated in o-xylene at 1258C in the
presence of a hydrogen atom source, 11 was formed along
Received: August 11, 2007
Revised: February 26, 2008
Published online: April 24, 2008
Keywords: density functional calculations · heterocycles ·
.
inhibitors · pericyclic reaction · radicals
[1] Aids epidemic Update: December 2006: “UNAIDS/06.29E”.
www.unaids.org.
[2] a) S. M. Hammer, et al.; see Supporting Information J. Am. Med.
Assoc. 2006, 296, 827; b) D. J. Hazuda, et al.; see Supporting
Information Proc. Natl. Acad. Sci. USA 2004, 101, 11233; c) D . J.
Hazuda, et al.; see Supporting Information Science 2004, 305,
528; d) D. J. Hazuda, et al.; see Supporting Information Science
2000, 287, 646.
[3] V. Summa, paper presented at the 23rd ACS National Meeting,
San Francisco, CA, 10 September 2006.
[4] Raltegravir has been approved by the FDA for the treatment of
HIV/AIDS under the trademark ISENTRESS.
Scheme 5. Trapping of proposed radical intermediate.
[5] M. Markowitz, et al.; see Supporting Information J. Acquir.
Immune Def. Syndr. 2006, 43, 509.
with 16, the product of hydrolysis of 15. This supports the
À
presence of the radical pair resulting from N O cleavage in
[6] a) N. D. Heindel, M. C. Chun, Tetrahedron Lett. 1971, 12, 1439;
b) N. D. Heindel, M. C. Chun, J. Chem. Soc. Chem. Commun.
1971, 664; c) T. P. Culbertson, J. Heterocycl. Chem. 1979, 16,
1423; d) A. A. Santelli, A. C. Scotese, J. Heterocycl. Chem. 1979,
16, 213; e) A. Guzman, S. Romero, E. M. Guadalupe Urquiza,
J. M. Muchowski, J. Heterocycl. Chem. 1980, 17, 1101; f) V.
Summa, A. Petrocchi, V. G. Matassa, M. Taliani, R. Laufer, R. D.
Franasco, S. Altamura, P. Pace, J. Med. Chem. 2004, 47, 5336;
g) I. Stansfield, S. Avolio, S. Colarusso, N. Gennari, F. Narjes, B.
Pacini, S. Ponzi, S. Harper, Bioorg. Med. Chem. Lett. 2004, 14,
5085; h) E. Wagner, L. Becan, E. Nowakowska, Bioorg. Med.
Chem. Lett. 2004, 12, 265; i) Y.-L. Zhong, H. Zhou, D. R.
Gauthier, Jr., D. Askin, Tetrahedron Lett. 2006, 47, 1315.
[7] a) A. D. Becke, J. Chem. Phys. 1993, 98, 5648; b) C. Lee, W.
Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785; c) A. D. Becke, J.
Chem. Phys. 1993, 98, 1372; d) P. J. Stephens, F. J. Devlin, C. F.
Chabalowski, M. J. Frisch, J. Phys. Chem. 1994, 98, 11623; e) P. J.
Stephens, F. J. Devlin, C. S. Ashvar, K. L. Bak, P. R. Taylor, M. J.
Frisch, ACS Symp. Ser. 1996, 629, 105; f) W. J. Hehre, L. Radom,
P. von R. Schleyer, J. A. Pople, Ab Initio Molecular Orbital
Theory, Wiley, New York, 1986.
the reaction mixture. Reaction of the corresponding reactant
with the styrenyl group reduced gave a reduced product
analogous to 11 in 66% yield.
Evidence consistent with the polar radical pair mechanism
was provided by a crossover experiment (Scheme 6).[12] A 1:1
mixture of labeled 3Z***/3E*** and unlabeled 3Z/3E was
heated in o-xylene at 1258C for 2 h and at 1358C for 4 h to
[8] Gaussian 03 (Revision C.02): M. J. Frisch et al.; see Supporting
Information.
Scheme 6. 15N- and 13C-labeled vinyloxyamidine rearrangement.
[9] A. G. Baboul, L. A. Curtiss, P. C. Redfern, K. Raghavachari, J.
Chem. Phys. 1999, 110, 7650.
[10] J. Y. Choi, C. K. Kim, C. K. Kim, I. Lee, J. Phys. Chem. A 2002,
106, 5709.
[11] M. Yokoyama, Y. Menjo, M. Ubukata, M. Irie, M. Watanabe, H.
Togo, Bull. Chem. Soc. Jpn. 1994, 67, 2219.
[12] a) M. Newcomb, Tetrahedron 1993, 49, 1151; b) M. Newcomb, N.
Tanaka, A. Bouvier, C. Tronche, J. H. Horner, O. M. Musa, F. N.
Martinez, J. Am. Chem. Soc. 1996, 118, 8505; c) J. H. Horner,
O. M. Musa, A. Bouvier, M. Newcomb, J. Am. Chem. Soc. 1998,
120, 7738.
afford hydroxypyrimidinones 4*** and 4, which were identi-
fied by high-resolution mass spectrometry. No crossover
between the labeled and unlabeled fragments was observed.
This result reinforces the computational results discussed
previously: PRP (Scheme 3) recombines to give product
within the solvent cage.
Experimental and computational investigations have
provided evidence for the intermediacy of a novel polar
radical-pair in the assembly of the hydroxypyrimidinone core
4136
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 4134 –4136