Synthesis of (-)-PNU-286607 by asymmetric cyclization of alkylidene barbiturates

Article abstract of DOI:10.1021/ja808014h

PNU-286607 is the first member of a promising, novel class of antibacterial agents that act by inhibiting bacterial DNA gyrase, a target of clinical significance. Importantly, PNU-286607 displays little cross-resistance with marketed antibacterial agents and is active against methicillin-resistant staphylococcus aureus (MRSA) and fluoroquinoline-resistant bacterial strains. Despite the apparent stereochemical complexity of this unique spirocyclic arbituric acid compound, the racemic core is accessible by a two-step route employing a relatively obscure rearrangement of vinyl anilines, known in the literature as the "tert-amino effect." After a full investigation of the stereochemical course of the racemic reaction, starting with the meso cis-dimethylmorpholine, a practical asymmetric variant of this process was developed.

Full text of DOI:10.1021/ja808014h

Published on Web 03/04/2009
Synthesis of (-)-PNU-286607 by Asymmetric Cyclization of
Alkylidene Barbiturates
J. Craig Ruble,†,| Alexander R. Hurd,*,‡, Timothy A. Johnson,§,# Debra A. Sherry,‡
Michael R. Barbachyn,‡, Peter L. Toogood,‡, Gordon L. Bundy,† David R. Graber,†
and Gregg M. Kamilar†,O
Infectious Diseases Medicinal Chemistry, Pharmacia Corporation, 301 Henrietta Street,
Kalamazoo, Michigan 49001, Antibacterial Chemistry, Pfizer Global Research and DeVelopment,
2800 Plymouth Road, Ann Arbor Michigan 48105, and Antibacterial Chemistry, Pfizer Global
Research and DeVelopment, Eastern Point Road, Groton, Connecticut 06340
Received October 10, 2008; E-mail: hurd@lycera.com
Abstract: PNU-286607 is the first member of a promising, novel class of antibacterial agents that act by
inhibiting bacterial DNA gyrase, a target of clinical significance. Importantly, PNU-286607 displays little
cross-resistance with marketed antibacterial agents and is active against methicillin-resistant staphylococcus
aureus (MRSA) and fluoroquinoline-resistant bacterial strains. Despite the apparent stereochemical
complexity of this unique spirocyclic barbituric acid compound, the racemic core is accessible by a two-
step route employing a relatively obscure rearrangement of vinyl anilines, known in the literature as the
“tert-amino effect.” After a full investigation of the stereochemical course of the racemic reaction, starting
with the meso cis-dimethylmorpholine, a practical asymmetric variant of this process was developed.
Introduction
PNU-286607 was identified during a screening effort at
Pharmacia and Upjohn for compounds possessing whole cell
antibacterial activity.1 A reverse chemical genomics approach
led to the identification of bacterial type II topoisomerase
enzymes (DNA gyrase and topoisomerase IV) as the mechanistic
targets for this compound.2 The fluoroquinolones and novobiocin
have provided ample precedent that these enzymes are essential
in bacteria and are appropriate antibacterial targets. Further,
PNU-286607 displays little cross-resistance with marketed
antibacterial therapies.2 Since there is a dire need for new
molecular entities and antibacterial agents with novel mecha-
nisms of action to counter bacterial resistance, PNU-286607
represents a promising opportunity.3,4
Figure 1. Structure of PNU-286607 and the active enantiomer (-)-1 as
determined by NMR analysis and X-ray crystal structure analysis.
Extensive NMR studies and an X-ray crystal structure deter-
mination revealed that PNU-286607 had the structure shown
in Figure 1.2 PNU-286607 presented several significant synthetic
challenges due to its structural complexity. In addition to the
unusual spirocyclic barbituric acid moiety, the compound
possesses three stereogenic centers around the fused morpholine
portion of the tetrahydroquinoline core. Diastereocontrol of these
stereogenic centers was of particular concern. Further, since the
antibacterial activity resides solely in the (-)-enantiomer of
PNU-286607 ((-)-1),2 an asymmetric synthesis was given high
priority.5
Upon attempts to resynthesize the originally assigned structure
of PNU-286607, it became clear that the structure was incorrect.
† Pharmacia Corporation.
‡ Pfizer Global Research and Development, Ann Arbor, Michigan.
§ Pfizer Global Research and Development, Groton, Connecticut.
| Current address: Lilly Research Laboratories, Eli Lilly and Company,
Indianapolis, IN 46285.
Current address: Lycera Corp., 46701 North Commerce Center Dr.,
Plymouth, MI 48170.
A careful review of the literature revealed an obscure, but
seemingly well-studied set of reactions influenced by what has
been termed the “tert-amino effect.” Although this reaction has
been reviewed,6 perhaps the most authoritative body of work
# Current address: Pfizer Veterinary Medicine Research and Development,
333 Portage St., Kalamazoo, MI 49007.
Current address: AstraZeneca Pharmaceuticals LP, 35 Gatehouse Dr.,
Waltham, MA 02451.
O Current address: Pfizer Research Technology Center, Cambridge, MA
02139.
(1) Barbachyn, M. R.; Bundy, G. L.; Dobrowolski, P. J.; Hurd, A. R.;
Martin, G. E.; McNamara, D. J.; Palmer, J, R.; Romero, D. L.; Romero,
A. G.; Ruble, J. C.; Sherry, D. A.; Thomasco, L. M.; Toogood, P. L.
U.S. Patent 7,208,490 B2, April 24, 2007 (filed October 2, 2003).
(2) Miller, A. A. et al Antimicrob. Agents Chemother. 2008, 52 (8), 2806.
(3) Spellberg, B.; Powers, J. H.; Brass, E. P.; Miller, L. G.; Edwards, J.
E, Jr. Antimicrob. Res. DeV. 2004, 38, 1279.
(4) (a) Levy, S. B. AdV. Drug DeliVery ReV. 2005, 57, 1446. (b) Peterson,
L. R. Clin. Microbiol. Infect. 2005, 11 (5), 4. (c) Silver, L. L.; Bostian,
K. A. Antimicrob. Agents Chemother. 1993, 37, 377.
(5) The absolute configuration of (-)-1 was determined by X-ray analysis
of a camphanic amide of the reduced nitro derivative. See ref 2.
(6) (a) Meth-Cohen, O.; Suschitzky, H. AdV. Heterocyl. Chem. 1972, 14,
211. (b) Quintela, J. M. Recent Res. DeVel. Org. Chem. 2003, 7, 259.
9
10.1021/ja808014h CCC: $40.75
2009 American Chemical Society
J. AM. CHEM. SOC. 2009, 131, 3991–3997 3991

A R T I C L E S
Ruble et al.
Scheme 1. tert-Amino Cyclizations
Scheme 2. Cyclization of Aldehyde 5 to PNU-286607
Scheme 3. Cyclization of Malononitrile Adducts of Aldehyde 8
outlining the scope and limitations of this reaction comes from
work of Verboom and Reinhoudt.7 Additionally, some recent
developments of a variant of this reaction have been reported
by the Sames and Chatani laboratories.8 The tert-amino effect,
generically illustrated in Scheme 1, involves the 1,5-hydrogen
shift of a 2-vinyl N-alkyl aniline (2) followed by cyclization of
the newly generated stabilized anion on to the iminium ion of
3, producing 4. Typically, very strong electron withdrawing
moieties are required for the initial [1,5]-shift. At the start of
our investigations, there were very few published examples
involving diastereoselective tert-amino reactions wherein the
resident stereogenic centers were located on the pendant
azacycle,7d and none with a stereogenic center located adjacent
to the migrating hydrogen.9
position.1,2,10 However, a clear understanding of the stereo-
chemical course of the reaction was lacking
The first indication of the complexities of the reaction arose
when we studied the reaction of malononitrile with aldehyde
8, Scheme 3. Instead of the expected reaction product with
the all-equatorial arrangement of substituents around the mor-
pholine ring (10), the ring juncture stereochemistry was inverted;
bis-nitrile 9 was the observed product. Upon stirring of 9 in 4
N HCl/dioxane for 1 h, bis-nitrile 10 was formed. The
stereochemical assignments of 9 and 10, were easily confirmed
by 1H NMR analysis. In particular, the coupling constants
between hydrogens H(C4) and H(C4a) were diagnostic (J ) 3
Hz for compound 9, J ) 9 Hz for compound 10).
The observation of an initially formed kinetic product in the
malononitrile series led us to consider whether the same pathway
is operational in the barbituric acid series as well. We undertook
Results and Discussion
Preliminary studies with meso aldehyde 5 (Scheme 2)
revealed that simple mixing and heating of 5 with barbituric
acid in methanol provided good yields of PNU-286607, with
all the substituents around the morpholine ring in an equatorial
1
a time course H NMR analysis, and a deuterium labeling
experiment aimed at outlining the stereochemical course of this
reaction.
(7) (a) Verboom, W.; Reinhoudt, D. V.; Visser, R.; Harkema, S. J. Org.
Chem. 1984, 49, 269. (b) Nijhuis, W. H. N.; Verboom, W.; Reinhoudt,
D. N. Synthesis 1987, 7, 641. (c) Nijuis, W. H. N.; Verboom, W.;
El-Fadl, A. A.; Harkema, S.; Reinhoudt, D. N. J. Org. Chem. 1989,
54, 199. (d) Nijhuis, W. H. N.; Verboom, W.; El-Fadl, A. A.; van
Hummel, G. J.; Reinhoudt, D. N. J. Org. Chem. 1989, 54, 209. (e)
Groenen, L. C.; Verboom, W.; Nuhuis, W. H. N.; Reinhoudt, D. N.;
Van Hummel, G. J.; Feil, D. Tetrahedron 1988, 44, 4637.
(8) (a) Pastine, S. J.; McQuaid, K. M.; Sames, D. J. Am. Chem. Soc. 2005,
127, 12180. (b) Tobisu, M.; Chatani, N. Angew. Chem., Int. Ed. 2006,
45, 1683.
(9) After the completion of our study, a cyclization of a 3,5-dimethylpi-
peridine derivative was reported.; Deeva, E. V.; Glukhareva, T. V.;
Zybina, N. A.; Morzherin, Y. Y. Russ. Chem. Bull., Int. Ed. 2005, 54,
1537.
(10) Our initial racemic synthesis appeared in the published patent literature
(WO 2004/031195 A1, submitted on October 2, 2003 and published
on April 15, 2004). Subsequently, two papers were submitted and
published describing the synthesis of some compounds similar to PNU-
286607: D’yachenko, E. V.; Glukhareva, T. V.; Dyudya, L. V.; Eltsov,
O. V.; Morzherin, Y. Y. Molecules 2005, 10, 1101. Krasnov, K. A.;
Kartsev, V. G. Russ. J. Org. Chem. 2005, 41, 901.
9
3992 J. AM. CHEM. SOC. VOL. 131, NO. 11, 2009

Asymmetric Cyclizations of Alkylidene Barbiturates
A R T I C L E S
1
Figure 2. Time course H NMR spectra of the conversion of 5 to PNU-286607 via 11 in DMSO-d6 (aromatic region only).
1
Because of the technical challenges of the experiment, namely
the faster rate of the barbituric acid cyclization and unknown
solubility of all the potential intermediates in this pathway,
aldehyde 5 and barbituric acid were combined in DMSO-d6 and
1H NMR spectra were obtained at various time intervals and
temperatures (Figure 2, only the aromatic region is shown for
clarity). After 20 min at room temperature, only the starting
aldehyde 5 was observed. After 4 days at room temperature an
intermediate accumulated that was shown by independent
synthesis to be the alkylidene intermediate 6. Upon heating to
60 °C and then to 80 °C, a new, previously unidentified species
accumulated. Further heating of the mixture at 100 °C resulted
in conversion of this intermediate to the known product, PNU-
286607 (1), which is presumably the thermodynamically more
stable isomer. We speculate that the previously unidentified
intermediate is the kinetically formed isomer 11. A definitive
assignment, usually accomplished by examination of the
aliphatic region of the H NMR spectra, was complicated by
overlap of resonances from multiple species.
Next, we studied the [1,5]-shift using a nonsymmetrically
deuterated morpholine. Reinhoudt and co-workers have inves-
tigated the [1,5]-shift with pyrrolidine-2,2,5,5-d4.7e In this study
the isotope effect was determined to be 3.0 ( 0.3 at 91.2 °C in
DMSO-d6, providing significant support for hydrogen migration
in the rate-determining step. We were interested in studying a
nonsymmetrically labeled morpholine for deeper mechanistic
insight and ultimately to examine the possibility of developing
an asymmetric process. First, chiral nonracemic, doubly labeled
morpholine 14 was prepared by adaptation of methods previ-
ously reported by our co-workers (Scheme 4). (S)-1-Amino-
propan-2-ol (12) was benzylated via reduction of the corre-
sponding imine,11 which was followed by an acylation-alkylation
procedure with bromopropionic acid and LiAlD4 reduction.
During this sequence, considerable epimerization occurred,
9
J. AM. CHEM. SOC. VOL. 131, NO. 11, 2009 3993

A R T I C L E S
Ruble et al.
Scheme 4. Preparation and Cyclization of Deuterium-Labeled PNU-286607
Scheme 5. Isomerization of the trans Isomer 20 to PNU-286607
resulting in the need for chromatographic separation of the
disastereomers and accounting for the low yield of the desired
cis-dimethylmorpholine. Amine deprotection followed by ad-
dition to 2-fluoro-5-nitrobenzaldehyde (15) furnished the cy-
clization precursor (16). To achieve a cyclization, aldehyde 16
was combined with barbituric acid and heated to 80 °C in
methanol. The resulting products were isolated as a 3:1 mixture
of 17 and 18. The products were separated by CSP-HPLC and
cis product (PNU-286607) was observed during the isolation
of trans-dimethylmorpholine product (20), derived from alde-
hyde 19, Scheme 5. While this observation could have been
explained by the presence of a cis-dimethylmorpholine con-
taminant in 19, PNU-286607 was consistently observed despite
rigorous purification of aldehyde 19. This result indicated that
both the C(4) and C(4a) centers were stereochemically labile
and that the configuration of the desired product may be
controlled by a single persistent stereocenter at C(2). An
enantioselective synthesis of (-)-1 was designed around this
observation. Whereas the racemic target, PNU-286607, was
prepared from the meso morpholine aldehyde 5, the synthesis
of (-)-1 was planned using chiral, nonracemic trans-dimeth-
ylmorpholine intermediate (+)-19, Figure 3.
1
the structures were assigned by H NMR analysis (Supporting
Information). The migration of the deuterium was stereoselective
in compound 18. The ratio of product derived from a hydrogen
migration (17) to product derived from deutero migration (18)
was ∼3:1, in line with the previously reported deuterium isotope
effect.7e We were pleased to detect only two distinct products
and not products derived from double deuterium migration,
which would indicate a reversible 1,5-hydrogen migration. This
synthesis of 17 represents the first asymmetric synthesis of PNU-
286607, albeit in deuterated form. Further, the absolute con-
figuration of the active enantiomer ((-)-1) was reconfirmed
since 17 and (-)-1 displayed equal antibacterial activity; 18
was devoid of activity.
Capitalizing on the observed isomerization, to develop a
robust, scalable, and practical procedure was anticipated to be
a considerable challenge. It was unclear whether the energetic
differences between the two compounds (20 and PNU-286607,
Scheme 5) were sufficient to drive the equilibration to syntheti-
Next, we turned our attention to the development of a
synthetic route to nondeuterated asymmetric products. We had
already determined that the C(4a) stereocenter (Figure 2) was
stereochemically labile and was dependent upon the configu-
ration of the adjacent methyl group at C(4) (vide supra). The
doubly equatorial arrangement of the methyl groups in the 2,6-
dimethyl morpholine ring was thought to be the major factor
that led to the high levels of diastereoselectivity that we observed
in the synthesis of PNU-286607. However, examination of the
C(4a) isomerization mechanism (Figure 2) led us to consider
whether the C(4) methyl-bearing stereocenter was labile as well,
via the putative iminium ion intermediate 7, Figure 2. A
breakthrough observation occurred when a small amount of the
(11) A similar procedure has been reported for the synthesis of the
enantiomer of the aminoalcohol: Mickelson, J. W.; Belonga, K. L.;
Jacobsen, E. J. J. Org. Chem. 1995, 60, 4177.
Figure 3. Synthetic design for the asymmetric synthesis of (-)-1.
9
3994 J. AM. CHEM. SOC. VOL. 131, NO. 11, 2009

Asymmetric Cyclizations of Alkylidene Barbiturates
A R T I C L E S
Figure 4. 1H NMR spectra (methyl region) of the isomerization of (+)-20 to (-)-1 in DMSO-d6.
Scheme 6. Asymmetric Synthesis of (-)-1, the Active Enantiomer of PNU-286607
cally acceptable levels. Further, we anticipated that the experi-
mental conditions necessary to achieve the isomerization might
be quite vigorous and we were concerned about the durability
of the products under these conditions.
established as an ∼8:1 mixture of cis/trans isomers. As a proof
of concept, this ratio was deemed to be acceptable.
Practically, protic solvents were found to be optimal, provid-
ing adequate temperature control and favorable reaction rates.
Aldehyde (+)-19, prepared by the addition of chiral morpholine
2112 to aldehyde 15, was treated with barbituric acid and MeOH
and heated to 65 °C (eq 1, Scheme 6). This effected the
conversion of (+)-19 to trans product (+)-20 in 48% yield. The
trans product 20 then was isomerized in 72% yield to (-)-1 by
We began our investigation into the asymmetric synthesis
1
by first monitoring the isomerization using H NMR spectros-
1
copy. Figure 4 shows the methyl region of the H NMR time
course in DMSO-d6 for the isomerization of (+)-20 to (-)-1.
After 2 h at 140 °C, two sets of new resonances appeared which
became more abundant at 5 h. By 22 h, equilibrium was
(12) (2R,6R)-2,6-Dimethylmorpholine was purchased from BASF.
9
J. AM. CHEM. SOC. VOL. 131, NO. 11, 2009 3995

A R T I C L E S
Ruble et al.
Scheme 7. Mechanism of [1,5]-Shift, Cyclization Producing (+)-20 and Isomerization to (-)-1
simple heating in n-butanol at 117 °C. Further, the process was
reduced to a one-step procedure by dissolution of aldehyde (+)-
19 and barbituric acid in n-butanol and heating to reflux for
24 h (Scheme 6). The ratio of crude products was 81:14:3:2,
but the desired product (-)-1 was isolated in 74% yield and
>99:1 er, providing a two-step asymmetric synthesis of (-)-1,
the active enantiomer of PNU-286607.
Extension of this mechanistic analysis to the isomeric
morpholine (+)-19 is illustrated in the lower portion of Scheme
7. Alkylidene 22, derived from barbituric acid condensation with
aldehyde (+)-19, undergoes the [1,5]-shift producing zwitterion
23. Presumably, the product of kinetic control 24 arises via axial
attack of the enolate on the iminium ion. Although product 24
was never identified and characterized, there were four isomers
produced in the asymmetric cyclization. The least abundant
isomer is presumed to be 24, the kinetically formed trans
product.13 Reversion to zwitterion 23 and subsequent cyclization
occurs via a higher energy transition state structure with
equatorial attack, or alternatively via a higher energy morpholine
ring conformation to provide (+)-20, the thermodynamically
more stable trans isomer.
When sufficient energy is applied to the system, zwitterion
23 isomerizes to zwitterion 25, giving access to the cis manifold.
Attack of the enolate produces either the cis product from kinetic
control (11), or the cis product from thermodynamic control
(-)-1. Because of the all-equatorial arrangement of the sub-
stituents around the morpholine ring, (-)-1 is the lowest energy
and most abundant of the four isomers formed. In the above
stereochemical analysis, the kinetic versus thermodynamic
designation refers to the relationship between the stereogenic
centers at C(4) and C(4a) and the cis and trans designation refers
to the relationship of the two methyl groups across the
morpholine system (Figure 2 and Scheme 7). Heating of
isomerically pure (-)-1 or (+)-20 to 117 °C in n-butanol yields
the identical ratio of diastereomers further indicating that the
process is under thermodynamic control. However, the [1,5]-
Mechanism. The proposed mechanism of the [1,5]-hydride
shift, cyclization sequence in the meso series to the kinetic
product 11 is illustrated in Figure 2. The first step, formation
of the alkylidene (6) occurs readily at room temperature and is
usually accompanied by a color change from the bright yellow
of the aldehyde to the deep red color of the alkylidene.
Alkylidene 6 has been isolated and characterized, and is
competent in the thermal conversion to PNU-286607. The
deuterium-labeling experiment lends support to the intramo-
lecular, stereoselective hydride migration being the key step,
although the experiment fails to indicate which hydride (deu-
terium) migrates, axial or equatorial, because of epimerization
at the newly formed methine bridgehead position (C4a).
However, it is reasonable to conclude that the axial hydride
migrates due to preferential orbital overlap with the coplanar
vinyl aryl system. The product of kinetic control 11 arises by
axial attack of the enolate on the iminium ion. In the subsequent
isomerization, 11 thermally reverts via a retro-Mannich type
process to zwitterion 7 followed by closure to PNU-286607,
either via a higher energy transition state structure with
equatorial attack of the enolate on the imminium ion, or via
axial attack on a higher energy morpholine conformation. Either
mechanism gains access to the more thermodynamically stable
product with the all-equatorial arrangement of substituents
around the morpholine ring.
(13) Attempts to characterize 24 were hampered by its low abundance and
poor stability when isolated in enriched form.
9
3996 J. AM. CHEM. SOC. VOL. 131, NO. 11, 2009

Asymmetric Cyclizations of Alkylidene Barbiturates
A R T I C L E S
support for the intermediacy of enamine 26 in the isomer-
ization process.14
Conclusion
We have developed a practical two-step asymmetric synthesis
of the (-)-enantiomer of PNU-286607, a promising member
of a novel class of antibacterial agents currently under develop-
ment. We have highlighted mechanistic aspects of the reaction
that permit stereochemical control over the reaction course and
the isolation of products of high enantiomeric purity. Our
mechanistic understanding of the tert-amino effect as it applies
to PNU-286607 allowed for the use of chiral trans-dimethyl-
morpholine as the source of asymmetry to obtain a product
containing an embedded cis-dimethylmorpholine. Although only
a single methyl stereocenter around the morpholine ring persists
during the process, the trans arrangement is necessary for
introduction of chirality.
Figure 5. Putative enamine intermediate 26 and deuterium incorporation
product 1-d1.
Scheme 8
Acknowledgment. The authors thank Gary Martin for extensive
NMR analysis of PNU-286607, 9, and 10 and Fusen Han for the
X-ray crystal structure of PNU-286607. We also acknowledge
Jeremy Starr for assistance with analytical data.
shift appears to be irreversible, as there is no erosion of the
enantiomeric ratio of recovered (-)-1.
Presumably, the isomerization of zwitterion 23 to 25
proceeds via the enamine 26, Figure 5. Although enamine
26 could not be detected, we present two pieces of evidence
that support the intermediacy of 26. First, heating aldehyde
5 with dibenzylamine and p-toluenesulfonic acid produces
enamine 27, itself an interesting product derived from the
[1,5]-shift of a hydride onto the dibenzyliminium ion, Scheme
8. Product 27 lends support for the intermediacy of the
morpholino-enamine substructure 26. Second, when the [1,5]-
shift, cyclization, isomerization sequence (Scheme 5) was
conducted in n-butanol-d1, deuterium incorporation occurs
at the C4 position affording 1-d1 (Figure 5), providing further
Supporting Information Available: General experimental
details; full experimental details including spectroscopic and
analytical data for compounds (-)-1, 8-10, 13, 14, 16-20, and
27; X-ray crystal structure of PNU-286607. This material is
available free of charge via the Internet at http://pubs.acs.org.
JA808014H
(14) Additionally, a significant literature precedent exists for a similar
methyl group isomerization in Sapogenin to iso-Sapogenin. Woodward,
R. B.; Sondheimer, F J. Am. Chem. Soc. 1958, 40, 6693.
9
J. AM. CHEM. SOC. VOL. 131, NO. 11, 2009 3997