Target-directed synthesis of antibacterial drug candidate GSK966587

Article abstract of DOI:10.1021/ol101235f

(Equation Presented). An efficient enantioselective total synthesis of the potent antibiotic GSK966587 was accomplished. Highlights of the synthesis include two innovative Heck reactions, a highly selective zincate base directed ortho-metalation, Sharpless asymmetric epoxidation, and a fully convergent final step fragment coupling.

Full text of DOI:10.1021/ol101235f

ORGANIC
LETTERS
2010
Vol. 12, No. 15
3422-3425
Target-Directed Synthesis of
Antibacterial Drug Candidate GSK966587
Eric A. Voight,*,†,‡ Hao Yin,† Susan V. Downing,† Stacie A. Calad,†
Hayao Matsuhashi,† Ilaria Giordano,§ Alan J. Hennessy,§ Richard M. Goodman,†
and Jeffery L. Wood†
GlaxoSmithKline, Synthetic Chemistry Department, 709 Swedeland Road, P.O. Box
1539, King of Prussia, PennsylVania 19406, and Antibacterial DPU, Infectious
Diseases CEDD, Gunnels Wood Road, SteVenage, SG1 2NY, U.K.
eric.a.Voight@abbott.com
Received May 28, 2010
ABSTRACT
An efficient enantioselective total synthesis of the potent antibiotic GSK966587 was accomplished. Highlights of the synthesis include two
innovative Heck reactions, a highly selective zincate base directed ortho-metalation, Sharpless asymmetric epoxidation, and a fully convergent
final step fragment coupling.
In a study aiming to identify novel antimicrobial agents,
GSK966587 (1, Figure 1) was chosen for further develop-
ment due to its potent activity against Gram-positive and
Gram-negative bacteria. The existing synthetic route to
prepare compound 1 was lengthy and challenging to scale.
An efficient and inexpensive catalytic asymmetric synthesis
was needed to support material needs for a rapid project
timeline. From a synthetic standpoint, the complex hexacyclic
structure of GSK966587 presented ample opportunity for the
discovery and development of novel organic reactions.
Our strategy to GSK966587 (1) is shown in Scheme 1.
To maximize convergency, a disconnection that separated
the fully elaborated side chain (2) from the tricyclic core
spiro-epoxide (3) was chosen. The greater nucleophilicity
of the piperidine nitrogen over the alkylamine would ideally
circumvent the need for a protecting group. Diamine 2 would
be derived from reductive amination of readily prepared
aldehyde 51 with commercially available 1-Boc-4-aminopi-
peridine (4). Spiro-epoxide 3 could be prepared from allylic
alcohol 6 via Sharpless asymmetric epoxidation2 followed
by cyclization. We envisioned the preparation of alcohol 6
from a directed ortho-metalation reaction to install the C-8
substituent,3 requiring 7-fluoro[1.5]naphthyridinone (7). Com-
pound 7 could arise from an acrylate Heck reaction4 of
commercially available 2-chloro-5-fluoro-3-pyridinamine (8)5
followed by cyclization.6
Figure 1. GSK966587 (1).
To begin the synthesis, the Heck reaction of chloropyridine
8 was investigated (Scheme 2). Although significant progress
has been made regarding Heck reactions of aryl chlorides,7
there are few examples with heteroaryl chloride substrates.8
Initial screening of ligand, base, and solvent combinations
† Synthetic Chemistry Department, Chemical Development.
‡ Current Address: Abbott Laboratories.
§ Antibacterial Discovery Performance Unit, Infectious Diseases Center
for Excellence in Drug Discovery.
(1) Cailleau, N.; Davies, D. T.; Hennessy, A. J.; Jones, G. E.; Miles,
T. J.; Pearson, N. D. Patent WO 2007/071936 A1.
10.1021/ol101235f  2010 American Chemical Society
Published on Web 07/15/2010

Scheme 1. Retrosynthetic Analysis
identified S-Phos9 as an optimal ligand, affording Heck
product 9 in 69% yield after chromatography. The coupling
product could be treated with NaOEt in EtOH to produce
the desired naphthyridinone 7 in 84% yield.6a
With a reliable route to fluoronaphthyridinone 7 in hand,
its conversion to methoxynaphthyridine 11 was investigated
(Scheme 3). A vast array of direct methylation conditions
gave primarily N-methylation, so an indirect path via
chloronaphthyridine 10 was employed. Both POCl3 and
SOCl2/DMF accomplished chlorination cleanly.12 Basic
conditions were originally used to convert chloronaphthy-
ridine 10 to methoxynaphthyridine 11, but competitive
fluorine displacement was a persistent problem. As an
alternative, the conversion of naphthyridinone 7 to chlo-
ronapthyridine 10 was followed by MeOH addition. The
excess HCl that was generated promoted the transformation
to methoxynaphthyridine 11.13 Demethylation back to naph-
thyridinone 7 was the major side reaction (6%), but this
byproduct could be washed away efficiently with 4 N
aqueous NaOH.
Scheme 2
.
One-Pot Heck/Cyclization Reaction of
2-Chloropyridine 8
Scheme 3. Preparation of
7-Fluoro-8-iodo-2-methoxy[1.5]naphthyridine (13)
Although the two-step process was sufficient for supplying
material for early development work, it was observed that
desired naphthyridinone 7 was a major byproduct in reactions
with tri-tert-butylphosphine as ligand in aromatic solvents.10
Further optimization around the direct preparation of naph-
thyridinone 7 identified the current best conditions (Scheme
2) and afforded the product in 76% yield. The mechanism
of the cyclization reaction remains unclear, but a palladium
hydride addition/cyclization/elimination pathway seems to
be the most likely rationale.11
(2) Katsuki, T.; Sharpless, K. B. J. Am. Chem. Soc. 1980, 102, 5974.
(3) For related quinoline metalation strategies, see: (a) Bannacef, I.;
Tymciu, S.; Dhilly, M.; Mongin, F.; Que´guiner, G.; Lasne, M.; Barre´, L.;
Perrio, C. J. Org. Chem. 2004, 69, 2622. (b) Lefebvre, O.; Marull, M.;
Schlosser, M. Eur. J. Org. Chem. 2003, 2115. (c) Marull, M.; Schlosser,
M. Eur. J. Org. Chem. 2003, 1576.
At this point in the synthesis, a regioselective directed
ortho-metalation3 of naphthyridine 11 was required (Table
1). Traditional alkyllithium or lithium amide bases gave poor
(4) Ziegler, C. B., Jr.; Heck, R. F. J. Org. Chem. 1978, 43, 2941.
(5) 2-Chloro-5-fluoro-3-pyridinamine: $2295/kg from Archimica.
(6) For a similar strategy to a [1.5]naphthyridinone, see: (a) Sakamoto,
T.; Kondo, Y.; Yamanaka, H. Chem. Pharm. Bull. 1985, 33, 4764. (b) Li,
J.; Zheng, L.; King, I.; Doyle, T. W.; Chen, S. Curr. Med. Chem. 2001, 8,
121.
(11) Intermediate 9 was observed by HPLC analysis during the one-
step Heck cyclization, and exposure of isolated 9 to those reaction conditions
provided cyclized product 7. However, in the presence of only Cy2NMe
and xylene at 135 °C, the reaction from isolated 9 to napthyridinone 7 was
very slow, suggesting that Pd catalyst is needed for the conversion.
(12) Dibutylformamide was later used in place of dimethylformamide
to catalyze the SOCl2 reaction in order to alleviate safety concerns around
the potential formation of dimethylcarbamoyl chloride (DMCC), a known
potent carcinogen. See: Levin, D. Org. Process Res. DeV. 1997, 1, 182.
(13) The mono-HCl salt of 10 did not react with MeOH; however, even
a catalytic amount of HCl could convert the mono-HCl salt to 11. Thanks
to Huan Wang (GSK) for this suggestion.
(7) Littke, A. F.; Fu, G. C. Org. Synth. 2005, 81, 63, and references
cited therein.
(8) (a) Littke, A. F.; Fu, G. C. J. Am. Chem. Soc. 2001, 123, 6989. (b)
Basu, B.; Frejd, T. Acta Chem. Scand. 1996, 50, 316.
(9) Wolfe, J. P.; Singer, R. A.; Yang, B. H.; Buchwald, S. L. J. Am.
Chem. Soc. 1999, 121, 9550.
(10) A similar path to a [1,8]naphthyridinone has been reported in 42%
yield: Singh, B.; Bacon, E. R.; Robinson, S.; Fritz, R. K.; Lesher, G. Y.;
Kumar, V.; Dority, J. A.; Reuman, M.; Kuo, G.; Eissenstat, M. A.; Pagani,
E. D.; Bode, D. C.; Bentley, R. G.; Connell, M. J.; Hamel, L. T.; Silver,
P. J. J. Med. Chem. 1994, 37, 248.
Org. Lett., Vol. 12, No. 15, 2010
3423

results and required cryogenic conditions, since methox-
ynapthyridine 11 was prone to form dianions (entry 1) and/
or suffered nucleophilic displacement of the C-7 fluorine.
While LDA gave promising results under carefully controlled
conditions (entry 2), we also became interested in zincate
base methodology.14 Uchiyama’s zincate, TMPZn(t-Bu)2Li,
gave predominantly undesired monoiodide 14 (entry 3).15
When (i-Pr)2NZn(t-Bu)2Li was used, monoiodide 13 was the
major product, but the deprotonation was sluggish (entry 4).
Since ZnEt2 and LDA are both commercially available and
inexpensive, we attempted to prepare and use (i-
Pr)2NZnEt2Li.16 This zincate reagent gave rapid deprotona-
tion of naphthyridine 11 at -10 °C in THF, providing the
desired monoiodide 13 with excellent selectivity after iodine
quench (entry 5).17 At least 3 equiv of iodine was needed,
since EtI was also formed in the reaction. After addition of
IPA, filtration, and a 1 N NaOH slurry, key intermediate 13
was isolated in 85% yield from naphthyridine 11. In practice,
the conversion of naphthyridinone 7 to iodide 13 was
performed as a through process using a crude toluene stream
of naphthyridine 11 (Scheme 3). This procedure gave isolated
iodide 13 in 74% yield for the two steps. The novel
heterocycle deprotonation demonstrated here complements
the pioneering pyridine metalation techniques introduced by
Schlosser and co-workers18 and should be applicable to other
challenging directed ortho-metalation reactions.
alcohol was investigated (Scheme 4, eq 1). R-Regioselective
allyl alcohol Heck reactions of aryl bromides are known to
occur in ionic liquid solvents.19 After extensive ligand and
base screening, and optimization of temperature and reagent
stoichiometry, a 71% yield of allylic alcohol 6 was obtained
using [bmim][BF4] as the solvent. While the ionic liquid
conditions gave acceptable results, we also chose to explore
traditional solvents. Ethylene glycol proved to perform as
well as [bmim][BF4], and after further optimization, a 77%
yield of allyl alcohol 6 could be obtained following flash
column chromatography.20
Scheme 4. Allyl Alcohol Coupling Reactions
As an alternative to the allyl alcohol Heck coupling, a
Negishi coupling with zincate 12, obtained from selective
deprotonation with (i-Pr)2NZnEt2Li (Scheme 3), was also
explored. The zincate base was added to naphthyridine 11
to form zincate 12, and this solution was transferred to a
preformed mixture of 2-bromopropen-3-ol, LDA, THF,
Pd2dba3·CHCl3, and trifurylphosphine. The process from
naphthyridinone 7 to allyl alcohol 6 via Negishi coupling
provided the product in 68% yield and only required a single
purification (Scheme 4, eq 2). Compared to the Heck
coupling strategy, this process is one step shorter since
formation of iodide 13 is no longer necessary.
To introduce the single asymmetric center of GSK966587,
a Sharpless asymmetric epoxidation of allylic alcohol 6 was
used (Scheme 5).2 This reaction took place at 0 °C with 10
mol % of Ti(Oi-Pr)4, 15 mol % of L-DIPT, 2.5 equiv of
cumene hydroperoxide, and 100 wt % of 4 Å molecular
sieves. Under these conditions, epoxy alcohol 16 could be
isolated in 81% yield and 90% ee.
Table 1. Preparation of Iodide 13a
time
(min)
temp
(°C)
13/14/15
base
equiv
(area %)b
1
2
3
4
5
LDA
LDA
1.5
1.1
1.1
1.0
1.2
15
5
30
120
30
-70
-60
23
23
-10
42/8/40
83/8/1.5
26/34/23
45/5/0
TMPZn(t-Bu)2Li
(i-Pr)2NZn(t-Bu)2Li
(i-Pr)2NZnEt2Lic
96/4/0
a All reactions in THF; quenched with excess I2. b Area percentage by
HPLC analysis. Remaining HPLC area was primarily starting material.
c When this reaction was repeated in toluene, the ratio was 88/0/12.
(16) (i-Pr)2NZnEt2Li has not shown any propensity toward benzyne
formation in our studies, even though (TMP)ZnMe2Li is known to react
with haloarenes to form benzynes: Uchiyama, M.; Miyoshi, T.; Kajihara,
Y.; Sadamoto, T.; Otani, Y.; Ohwada, T.; Kondo, Y. J. Am. Chem. Soc.
2002, 124, 8514.
After considerable experimentation with Negishi and
Suzuki coupling approaches to prepare allylic alcohol 6, the
Heck reaction between iodonaphthyridine 13 and allyl
(17) A report of the unsuccessful use of this reagent in a screen of various
zincate bases has appeared: Gauthier, D. R., Jr.; Limanto, J.; Devine, P. N.;
Desmond, R. A.; Szumigala, R. H., Jr.; Foster, B. S.; Volante, R. P. J.
Org. Chem. 2005, 70, 5938.
(14) For a review, see: Mulvey, R. E.; Mongin, F.; Uchiyama, M.;
Kondo, Y. Angew. Chem., Int. Ed. 2007, 47, 3802.
(18) (a) Marzi, E.; Bobbio, C.; Cottet, F.; Schlosser, M. Eur. J. Org.
Chem. 2005, 2116. (b) Schlosser, M.; Mongin, F. Chem. Soc. ReV. 2007,
36, 1161.
(15) This regiochemical discrepancy has been observed in deprotonation
of pyridines: Imahori, T.; Uchiyama, M.; Sakamoto, T.; Kondo, Y. Chem.
Commun. 2001, 2450. The regiochemistry of the undesired mono-iodide
has not been confirmed.
(19) Pei, W.; Mo, J.; Xiao, J. J. Organomet. Chem. 2005, 690, 3546.
(20) A 65-70% yield was obtained via EtOAc/cyclohexane crystal-
lization; however, this procedure remains in need of optimization.
3424
Org. Lett., Vol. 12, No. 15, 2010

Aqueous workup, concentration, and carbamate deprotection
with AcCl/MeOH gave the tri-HCl salt of 2, which was
basified with aq NaOH and extracted with DCM. This
procedure gave fully elaborated side chain 2 in 77% overall
yield.
Scheme 5. Preparation of Tricyclic Diol 18
Scheme 7. Total Synthesis of GSK966587
Encouraging results were obtained when epoxide 16 was
opened with 4-Boc-aminopiperidine or fully elaborated side
chain 2 (Scheme 1), but concerns about intramolecular
fluorine displacement quickly led us to consider an alternative
route (Scheme 5). When unpurified epoxy alcohol 16 was
treated with concentrated HCl, chloro diol 17 was formed.
After aqueous workup, the solution of chloro diol 17 in
butyronitrile was heated to 100 °C for 1.5 h, causing
cyclization and demethylation to occur in one pot.21 Cooling
to room temperature completed the crystallization of tricyclic
diol 18 in 63% overall yield from allylic alcohol 6 with no
deterioration of ee.22 The high water solubility of tricyclic
diol 18 precluded any reaction conditions requiring an
aqueous workup, so this protocol was well-suited for efficient
isolation.
To complete the synthesis of GSK966587 (Scheme 7),
spiro-epoxide 3 was prepared for final fragment coupling.
Diol 18 was first converted to spiro-epoxide 3 using Et3N
and perfluorobutanesulfonyl fluoride.23 After 1 h at room
temperature and filtration through silica gel, epoxide 3 (98%
solution yield) was treated with side chain 2 (1.5 equiv) at
room temperature for 14 h. GSK966587 precipitated directly
from the reaction mixture in 76% isolated yield from 18 (98.7
HPLC area %, 98% ee).
An efficient new route to GSK966587 was developed,
converting 2-chloro-5-fluoro-3-pyridinamine (8) to GSK966587
(1) in eight steps and 25% overall yield with three isolated
intermediates. The route showcased two innovative Heck
reactions, novel applications of zincate base methodology
in regioselective heterocycle deprotonation and one-pot
Negishi cross-coupling, a Sharpless asymmetric epoxidation,
and a fully convergent final coupling without requiring
protecting groups. Further development of this route could
provide the foundation for a long-term manufacturing-scale
supply of GSK966587.
Scheme 6. Preparation of Fully Elaborated Side Chain 2
Acknowledgment. Thanks to Carolyn Grady (GSK) for
technical sourcing, Josephine Vega (GSK) for analytical
support, and Kevin Leach (GSK) for NMR support.
The preparation of fully elaborated side chain 2 is shown
in Scheme 6. Commercially available N-Boc-4-aminopip-
eridine was condensed with aldehyde 5,1 followed by sodium
borohydride addition to complete the reductive amination.
Supporting Information Available: Experimental pro-
cedures and spectroscopic and analytical data for the
products. This material is available free of charge via the
Internet at http://pubs.acs.org.
OL101235F
(21) Chloromethane is generated in the conversion of 17 to 18 and safety
precautions (proper ventilation) were taken.
(22) With the opposite enantiomers of 16, 17, and 18, an overall yield
of 71% was obtained when epoxy alcohol 17 was isolated.
(23) Klar, U.; Neef, G.; Vorbruggen, H. Tetrahedron Lett. 1996, 37,
7497.
Org. Lett., Vol. 12, No. 15, 2010
3425