Rearrangements encountered in the attempted syntheses of pyridoazepinone carboxylic acids

Article abstract of DOI:10.1021/ol0481378

(Chemical equation presented) Attempts to synthesize pyridoazepinone carboxylic acids such as 33 by standard methodologies resulted exclusively in unusual and unexpected rearrangement products. Seven-membered ring formation was attempted by ring expansion

Full text of DOI:10.1021/ol0481378

ORGANIC
LETTERS
2004
Vol. 6, No. 24
4527-4530
Rearrangements Encountered in the
Attempted Syntheses of
Pyridoazepinone Carboxylic Acids
William J. Sanders,* Xiaolin Zhang, and Rolf Wagner
Infectious Disease Research, Abbott Laboratories, 200 Abbott Park Road,
Abbott Park, Illinois 60064-6217
will.sanders@abbott.com
Received September 14, 2004
ABSTRACT
Attempts to synthesize pyridoazepinone carboxylic acids such as 33 by standard methodologies resulted exclusively in unusual and unexpected
rearrangement products. Seven-membered ring formation was attempted by ring expansion of a six-membered ring and by aldol ring closure.
In each case, the major product resulted from rearrangement of the starting material without detection of the desired product. Ultimately, an
isomeric pyridoazepinone ethyl ester was prepared; however, attempted saponification resulted in another unusual rearrangement.
Naphthyridones with the general structure 1 (Figure 1), where
R1 ) alkyl, cycloalkyl, or aryl and R7 ) cycloalkylamino,
are well-known antibacterial agents belonging to the qui-
nolone class of gyrase inhibitors.1,2 Recently, it was shown
that similar compounds such as 2, where R1 ) H, are
antibacterial agents that inhibit protein synthesis in a wide
variety of bacterial species.3 The discovery that naphthyri-
dones without a carbon substituent on N-1 have antibacterial
activity without concomitant gyrase inhibition activity launched
an extensive search for structural modifications of 2 that
would improve inhibitory activity against bacterial protein
synthesis. We believed that the identification of 2 as a lead
compound was solely a result of structural bias in the
screening library and further optimization by structural
modification would prove fruitful. In the course of optimiza-
tion, we proposed preparation of seven-membered ring
analogues of generic structure 3 (Figure 1); however,
attempts at the synthesis of several variations of 3 resulted
in unexpected rearrangements without detection of any of
the predicted products.
Seven-membered rings have a long and storied history both
in synthetic chemistry and in drug discovery. Many meth-
odologies have been developed or adapted for the production
of seven-membered rings.4,5 Included among these are the
acyloin condensation and intramolecular aldol reaction and
ring-expansion reactions, including the Beckman rearrange-
ment and numerous fused aziridine- and cyclopropane-
opening reactions.6 More recently, ring-closing metathesis
(1) Andersson, M. I.; MacGowan, A. P. J. Antimicrob. Chemother. 2003,
51 (Suppl. S1), 1-11.
Figure 1. General structure of naphthyridone antibiotics (1),
antibacterial lead compound (2), and proposed seven-membered
ring analogues (3).
(2) The Quinolones, 2nd ed.; Andriole, V. T., Ed.; Academic Press: San
Diego, CA, 1998.
(3) Dandliker, P. J.; Pratt, S. D. et al. Antimicrob. Agents Chemother.
2003, 47, 3831-3839.
10.1021/ol0481378 CCC: $27.50
© 2004 American Chemical Society
Published on Web 11/02/2004

Scheme 1
Scheme 2
reducing agents without significant reduction of either
carbonyl group. This observed electrophilicity at C-2 led us
to choose sulfur ylides as cyclopropanating agents for the
desired transformation. To this end, naphthyridone 8 was
treated with trimethylsulfoxonium iodide/sodium hydride in
DMSO (Scheme 3). Upon treatment with 1 equiv of the ylide,
Scheme 3
has become the method of choice for the synthesis of many
different ring systems.7,8 In the pharmaceutical industry,
discovery of azepine and diazepine peptidomimetics with
wide-ranging biological activities inspired many syntheses
of new seven-membered heterocycles. On the basis of the
wealth of published information, we anticipated that a method
or methods for the preparation of compounds of generic
structure 3 would be revealed.
Synthesis of the naphthyridone ring system is generally
accomplished by intramolecular nucleophilic aromatic sub-
stitution on a 2-chloropyridine as shown in Scheme 1.9
In a one-pot procedure, â-ketoester 4 is converted first to
vinyl ether 5 by reaction with triethyl orthoformate in acetic
anhydride. Removal of solvent followed by treatment of 5
with 2,4-dimethoxybenzylamine affords vinylogous imide 6,
which cyclizes upon addition of sodium hydride to provide
naphthyridone 7. Subsequent substitution at C-7 with pyr-
rolidine and sequential deprotection affords the N-1 unsub-
50% conversion to a single product occurred, with 50%
starting material remaining unreacted. Addition of a second
equivalent of ylide resulted in quantitative conversion to a
single product that was inconsistent with either a cyclopro-
pane or the corresponding ring-expansion product. Mass
spectrometry indicated a molecular weight of 562, consistent
with addition of one methylene group plus 1 equiv of the
sulfur ylide. Analysis of the product by multidimensional
NMR techniques, including COSY, ROESY, HSQC, and
HMBC, led to the identification of sulfur ylide 11 (Scheme
4) as the sole product of the reaction.11
10
stituted naphthyridone 2 of the lead series.
Scheme 2 illustrates an approach to analogues of lead
compound 2 via ring expansion of naphthyridone 8. In the
proposed sequence, simple access to compound 10 was
envisioned via cyclopropanation/ring expansion of naphthy-
ridone 8. Oxidation or isomerization of 10 would then
provide the desired R,â-unsaturated system. We had previ-
ously observed that naphthyridones such as compound 8
underwent selective 1,4-reduction with various borohydride
A proposed mechanism for the formation of 11 is
illustrated in Scheme 4. Apparently, the desired cyclopro-
panation reaction occurs to provide intermediate 9. However,
addition of a second equivalent of the sulfur ylide to the
keto group is more facile than ring expansion, and tetrahedral
intermediate 12 is formed. Under normal circumstances, 12
would be expected to form an epoxide by elimination of
dimethyl sulfoxide;12 however, in the proposed mechanism
(4) Bremner, J. B. Prog. Heterocycl. Chem. 2003, 15, 385-430.
(5) Scriven, E. F. V. Heterocycl. Chem. 1986, 5, 425-54.
(6) Kantorowski, E. J.; Kurth, M. J. Tetrahedron 2000, 56, 4317-4353.
(7) Furstner, A. Angew. Chem., Int. Ed. 2000, 39, 3012-3043.
(8) Grubbs, R. H.; Chang, S. Tetrahedron 1998, 54, 4413-4450.
(9) Chu, D. T. W.; Fernandes, P. B.; Claiborne, A. K.; Gracey, E. H.;
Pernet, A. G. J. Med. Chem. 29, 1986, 2364-2369.
(10) Wagner, R.; Weitzberg, M.; Sanders, W.; Clark, R.; Hinman, M.;
Rosenberg, T.; Motter, C.; Larsen, D.; Bosse, T.; Palazzo, F.; Keyes, R.
Manuscript in preparation.
(11) For references to a similar stable ylide, see: Beautement, K.; Clough,
J. M. Tetrahedron Lett. 1984, 25, 3025-3028.
(12) Trost, B. M.; Melvin, L. S. Sulfur Ylides: Emerging Synthetic
Intermediates; Academic Press: New York, 1975.
4528
Org. Lett., Vol. 6, No. 24, 2004

Scheme 4
Scheme 5
product formed that was unstable to prolonged exposure to
silica gel and to air and was not spectroscopically consistent
with the desired product or with any of the obvious
intermediates in its predicted formation. Careful analysis of
the 1H NMR spectrum combined with the observed molecular
weight of the product as determined by mass spectrometry
resulted in identification of azaindole 18 as the sole product
of the reaction. This identification was confirmed by two-
dimensional NMR analysis.
Scheme 5 illustrates a proposed mechanism for azaindole
formation. Apparently, oxidation of the primary alcohol
proceeds as expected. However, the resultant aldehyde reacts
not with the enol of the â-ketoester but instead undergoes a
Friedel-Crafts-type cyclization to afford intermediate 17.
Rearomatization via intermolecular transfer of malonate to
a reaction byproduct affords a 3-hydroxy-azaindoline that
leads to 18 upon dehydration. While electrophilic aromatic
substitution would not generally be expected to compete with
aldol cyclization, in retrospect the nature of the substrate in
this situation may especially favor Friedel-Crafts cyclization.
To begin with, five-membered ring formation is kinetically
much faster than seven-membered ring formation. Normally,
one would not expect this kinetic advantage to overcome
the thermodynamic barrier of breaking aromaticity necessary
for Friedel-Crafts-type cyclization. However, proposed
cationic intermediate 17 is exceptionally resonance-stabilized
by three nitrogen atoms, and this thermodynamic stabilization
combined with the kinetic advantage of five-membered ring
formation could explain the outcome of the reaction. Ad-
ditionally, loss of carbon dioxide through decomposition of
ethyl malonate may provide an added driving force.
the tetrahedral intermediate breaks down with concomitant
ring-opening to provide a cyclopropyl ester enolate that, after
proton transfer, affords sulfur ylide 13. It is possible that
intermediate 12 eliminates DMSO to form an epoxide as
expected, but in the presence of DMSO solvent, the elimina-
tion is reversible. In this case, irreversible formation of 13
drives the reaction in the observed direction. Intermediate
13 is a very electrophilic “push-pull” cyclopropane13 that
reacts accordingly via ring-opening/proton transfer to afford
the observed product 11. In this sequence, an apparently
straightforward route to a seven-membered ring is circum-
vented by a highly unusual, counterintuitive reaction path-
way. We did not anticipate that 1,2-addition of a sulfur ylide
to the keto group of 9 would compete with 1,4-addition to
naphthyridone 8. Even more remarkable is the elimination
of a cyclopropane ester enolate from intermediate 12 over
the well-precedented epoxidation pathway.
A second synthetic approach to a pyrido[2,3-b]azepinone
16 (Scheme 5) was proposed via one-pot oxidation/intramo-
lecular aldol reaction sequence starting with alcohol 15.
â-Ketoester 4 reacts with 1 equiv of pyrrolidine very
selectively para to the ketoester functionality, providing the
monosubstituted derivative (14) in high yield. Subsequent
displacement of the second chloride was accomplished with
N-benzyl ethanolamine in refluxing acetonitrile to provide
the desired alcohol. Treatment of 15 with the Dess-Martin
periodinane was then expected to result in oxidation of the
primary alcohol to an aldehyde followed by spontaneous
intramolecular aldol cyclization.14 Whether or not the aldol
product would eliminate to provide the desired R,â-unsatur-
ated system was difficult to predict; however, under the acidic
reaction conditions we considered elimination likely. In fact,
when 15 was treated with 1 equiv of the periodinane, a single
Scheme 6 illustrates the successful preparation of a
pyridoazepinone ethyl ester. Pyrido[4,3-c]azepinone 28 was
isolated as a minor byproduct in the synthesis of 5-chlo-
romethylnaphthyridone 27. Cyclization of the vinylogous
imide intermediate formed by reaction of 26 with 2,4-
dimethoxybenzylamine proceeds via competing six- and
seven-membered ring-forming pathways favoring 27 by a
(13) Braun, N. A.; Stumpf, N.; Spitzner, D. Synthesis 1997, 917-920.
(14) Fleming, F. F.; Huang, A.; Sharief, V. A.; Pu, Y. J. Org. Chem.
1999, 64, 2830-2834.
Org. Lett., Vol. 6, No. 24, 2004
4529

Scheme 6
Scheme 7
ratio of approximately 9:1 as determined by LC-MS.
Although the synthetic route to 28 was not optimal, this
represented the first instance in which a seven-membered
ring analogue of lead compound 2 was isolated. Addition of
Boc-protected 3-aminopyrrolidine to 28 afforded compound
29 (Scheme 7) with only trace amounts of the regioisomeric
addition product. Preparation of the first compound of generic
structure 3 required only saponification and acid-catalyzed
deprotection of 29. Ethyl ester 29 was subjected to standard
saponification conditions (aqueous lithium hydroxide in
occurred providing unexpected rearrangement products. The
reactions described above exemplify the difficulties often
encountered in seven-membered ring synthesis. Although not
typical, these examples demonstrate seemingly extraordinary
and circuitous reaction pathways taken to avoid seven-
membered ring formation.
1
dioxane), affording a single product. However, H NMR
showed the presence of an ethyl group in the product and
mass spectrometry indicated identical molecular weights for
starting material and product. Multidimensional NMR analy-
sis of the isolated material led to identification of azaisoindole
32 as the sole reaction product. As illustrated in Scheme 7,
hydroxide apparently undergoes 1,4-addition to the vinylo-
gous imide instead of the expected attack on the ester
carbonyl, resulting in overall hydrolysis of the enamine
functionality. Amine intermediate 31 then undergoes in-
tramolecular imine formation and tautomerization to provide
the observed product 32.
Acknowledgment. The authors would like to thank Prof.
S. Burke and Drs. M. Hinman, G. Sheppard, and S. Davidsen
for reading the manuscript and for helpful comments and
Dr. M. Weitzberg for technical assistance and enthusiastic
discussions.
Supporting Information Available: Experimental pro-
cedures and spectral data for all new compounds and
complete 1H and 13C NMR assignments and two-dimensional
spectra for compounds 11, 14, 18, 29, and 32. This material
is available free of charge via the Internet at http://pubs.acs.org.
In summary, we have described the attempted synthesis
of compounds of generic structure 3 using three standard
synthetic methodologies. In each case, unusual reactions
OL0481378
4530
Org. Lett., Vol. 6, No. 24, 2004