Stereoselective synthesis of a dioxa-bicyclo[3.2.1]octane SGLT2 inhibitor

Article abstract of DOI:10.1021/ol100940w

A promising class of SGLT2 inhibitors bearing a unique dioxa-bicyclo[3.2.1] octane motif was recently disclosed. An improved stereoselective synthesis providing efficient access to one of the most potent and selective compounds from this class is reported. A one-pot deprotection/cyclization was used as the key step to form the dioxa-bicyclo[3.2.1]octane motif with full control of stereochemistry. Using an appropriately substituted aryl group, the route enables the synthesis of any given compound from the class.

Full text of DOI:10.1021/ol100940w

ORGANIC
LETTERS
2010
Vol. 12, No. 13
2940-2943
Stereoselective Synthesis of a
Dioxa-bicyclo[3.2.1]octane SGLT2
Inhibitor
Vincent Mascitti* and Cathy Pre´ville
Pfizer Global Research and DeVelopment, Groton, Connecticut 06340
Vincent.mascitti@pfizer.com
Received April 23, 2010
ABSTRACT
A promising class of SGLT2 inhibitors bearing a unique dioxa-bicyclo[3.2.1]octane motif was recently disclosed. An improved stereoselective
synthesis providing efficient access to one of the most potent and selective compounds from this class is reported. A one-pot deprotection/
cyclization was used as the key step to form the dioxa-bicyclo[3.2.1]octane motif with full control of stereochemistry. Using an appropriately
substituted aryl group, the route enables the synthesis of any given compound from the class.
Type 2 diabetes looms as a growing threat to human health
worldwide, a fact supported by increasingly alarming sta-
tistics.1 A considerable body of research has been carried
out within the pharmaceutical industry to identify new
antidiabetic medicines, but progress has been slow. Recently,
sodium glucose cotransporter 2 (SGLT2) inhibition has
emerged as a very promising approach to the treatment of
Type 2 diabetes. Inhibition of this target, which is responsible
Figure 1. Structure of phlorizin (natural product).
for the bulk of glucose reuptake in the kidney, allows control
of hyperglycemia in a glucose-dependent, insulin-indepen-
dent manner.2
The O-aryl glucoside natural product phlorizin,3,4
a
and C-aryl glucoside SGLT2 inhibitors.5,6 Among these,
several C-aryl glucosides (e.g., dapagliflozin and canagli-
flozin) are now in clinical development (Figure 2). Interest-
ingly, so far the structures from the known agents in
development are mostly C-aryl glucoside derivatives only
differing by their aglycone residue.5 No SGLT2 inhibitor
has reached the market yet, and the pharmaceutical industry
remains intensely focused on discovering differentiated
agents.
nonselective inhibitor of SGLT1 and SGLT2, provided an
interesting lead for the discovery of antidiabetic agents based
on glucose transport inhibition (Figure 1).
Starting with phlorizin, elegant and fruitful lead opti-
mization resulted in the identification of various O-aryl
(1) Diabetes Atlas, 4th ed.; International Diabetes Federation: Brussels,
2009.
(2) Neumiller, J. J.; White, J. R. Jr.; Campbell, R. K. Drugs 2010, 70,
377.
(3) Ehrenkranz, J. R. L.; Lewis, N. G.; Kahn, C. R.; Roth, J. Diabetes
(5) Washburn, W. N. Expert Opin. Ther. Patents 2009, 19, 1485
(6) Washburn, W. N. J. Med. Chem. 2009, 52, 1785
(7) Mascitti, V.; Collman, B. M. WO10023594, 2010
.
Metab. Res. ReV. 2005, 21, 31
.
.
(4) White, J. R., Jr. Clin. Diabetes 2010, 28, 5
.
.
10.1021/ol100940w  2010 American Chemical Society
Published on Web 06/08/2010

Scheme 1. Original Medicinal Chemistry Route
Figure 2. Structures of some C-aryl glucoside SGLT2 inhibitors.
We recently disclosed a new class of such agents based
on a unique dioxa-bicyclo[3.2.1]octane motif (Figure 3).7
Retrosynthetic analysis11 exploiting pseudo C2-symmetry
led to a D-mannose derivative as a readily available starting
chiron,12 formaldehyde and a dithiane intermediate serving
as building blocks (Figure 4).
Figure 3. Dioxa-bicyclo[3.2.1]octane-based SGLT2 inhibitors.
Compound 1 is one of the most potent and selective SGLT2
inhibitors from this class and has demonstrated robust ef-
ficacy in preclinical rodent models.7 It also has the potential
safety advantage of lacking activity in the well-known in
vitro micronucleus assay, a key distinction from certain
SGLT2 inhibitors in development.8 A compound from the
class is currently in phase 2 clinical trials.
The original route to dioxa-bicyclo[3.2.1]octane SGLT2
inhibitors is depicted in Scheme 1. Starting from an acyclic
Weinreb amide as an advanced intermediate,9 the route
capitalized on an unprecedented three-step protocol to
assemble the desired [3.2.1]-bridged ketal system (Scheme
1).7 This versatile synthetic sequence was a useful discovery
tool, which enabled rapid analogue preparation and identi-
fication of compounds with optimal PK/PD10 profiles;
however, the route was not well suited for the synthesis of
specific compounds on a larger scale. For example, it
provided 1 in 0.3% overall yield over 13 linear steps starting
from D-glucose including final HPLC separation of 1 from
its epimer at C-4. These considerations prompted the
development of a new stereoselective route that would
provide efficient access to any specific compound from the
class.
Figure 4. Retrosynthetic analysis.
The new synthesis, exemplified by the preparation of 1,
is summarized in Scheme 2. Diastereoselective addition of
the lithium anion derived from crystalline dithiane 213 to
aldehyde 3 (readily available starting from diacetone-R-D-
mannofuranose)14 at low temperature in THF produced
intermediate 4 as a single diastereomer via Si face addition
to the aldehyde.15
(12) Hanessian, S. The Total Synthesis of Natural Products: The Chiron
Approach; Pergamon Press: Oxford, 1983.
(13) For the synthesis and X-ray crystal structure of 2, see: Samas, B.;
Pre´ville, C.; Thuma, B. A.; Mascitti, V. Acta Crystallogr. 2010, E66, o1386.
(14) Brewster, K.; Harrison, J. M.; Inch, T. D.; Williams, N. J. Chem.
Soc., Perkin Trans. 1 1987, 21.
(15) Other methods, for instance, based on umpolung or biocatalysis,
used to build this R-hydroxy ketone motif starting from a 4-chloro-3-(4-
ethoxybenzyl)benzaldehyde derivative will be reported in due course.
(16) Readily available in two steps from intermediate 4. See Supporting
Information.
(8) In Vitro Mammalian Cell Micronucleus Test (MNvit). OECD
Guideline for Testing of Chemicals No. 487, OECD, Paris, 2010. Available
at: http://www.oecd.org/env/testguidelines. Compound 1 acts neither as an
aneugen or clastogen in the micronucleus assay.
(9) Available in 10 steps starting from D-glucose.
(10) Pharmacokinetics/Pharmacodynamics.
(11) Corey, E. J.; Cheng, X. M. The Logic of Chemical Synthesis; Wiley:
New York, 1989.
(17) Ho, P.-T. Tetrahedron Lett. 1978, 19, 1623.
(18) For an example of dithiane hydrolysis and acid-catalyzed spiroket-
alization, see: Smith, A. B., III.; Duan, J. J.-W.; Hull, K. G.; Salvatore,
B. A. Tetrahedron Lett. 1991, 32, 4855.
Org. Lett., Vol. 12, No. 13, 2010
2941

Scheme 2.
Synthesis of Compound 1a
a The yield of each individual step is indicated in parentheses; the yield from the telescoped process is indicated in red.
The configuration of the newly created stereocenter was
confirmed by single-crystal X-ray diffraction analysis of
derivative 5 (Figure 5).16 The high level of diastereoselec-
HPLC). Reduction of lactol 7 with sodium borohydride
in MeOH at 23 °C produced acyclic advanced intermediate
8, which was primed for validation of the key synthetic
transformation into 1.
When 8 was stirred at 23 °C in a mixture of TFA/H2O
(9/1 vol.) under air, deprotection/cyclization occurred to
cleanly produce 1 as a single isomer in 77% yield. This
reaction is believed to occur by cleavage of the acetonide to
lead to intermediate 9, followed by attack of the unprotected
hydroxyl group(s) onto the unmasked ketone (or a putative
thionium ion) to produce the desired bridged ketal under
thermodynamic control.18,19 Notably, the cyclization occurred
without any epimerization at C-4, in contrast to the original
discovery route. At this point, the next challenge was to
improve the overall yield of the reduction-(deprotection/
cyclization) sequence. We reasoned that the modest isolated
yield of 8 was due to formation of a strong borate complex
that was only partially broken under the original workup
conditions (neutralization with an aqueous solution of
NH4Cl). Thus, rather than isolating 8, the crude material was
exposed directly to the conditions for deprotection/cycliza-
tion, conditions that also would favor complete hydrolysis
of residual borate complex. In practice, telescoping of the
two steps led to a dramatically improved yield (65%).
Compound 1 was purified by flash chromatography over
silica gel and could be isolated as the L-pyroglutamic acid
cocrystal.7
Figure 5. ORTEP representation of the X-ray structure of 5.
tivity and preference for the Si face is consistent with a Cram
chelate-type pretransition state assembly and unfavorable
steric/stereoelectronic effects upon Re face attack of the bulky
nucleophile via a Felkin-Anh pretransition state assembly.
Having installed the stereocenter found at C-4 of the final
product, we next built the tetrasubstituted carbon C-1. This
task was accomplished by first reacting 4 with TBAF in THF
at 0 °C to produce lactol 6 in 75% yield; treatment of a
solution of 6 in MeOH/H2O at 85 °C with excess formal-
dehyde in the presence of K2CO3 provided intermediate 7
in 65% yield.17 The overall yield was 26% over three steps
if each intermediate was isolated. This overall yield could
be dramatically improved by carrying on the crude material
(telescoping) through the three steps to produce intermedi-
ate 7 in both high yield (40%) and purity (>95% by
In conclusion, the stereoselective synthesis described in
this Letter provides 1 in 26% overall yield over five steps
starting from readily available materials. This constitutes
a major improvement over the original route. Noteworthy
features of this improved synthesis include (1) exploitation
of pseudo C2-symmetry to identify the readily available
diacetone-R-D-mannofuranose starting material, (2) a
(19) On the basis of our observations and preliminary investigations,
the presence of a stream of air is believed to help drive the equilibrium
towards the formation of the desired product by scavenging thiol-based
byproducts into a putative 1,2-dithiolane intermediate which then oligo-
merizes. See: Ranganathan, S.; Jayaraman, N. J. Chem. Soc., Chem.
Commun. 1991, 934
.
2942
Org. Lett., Vol. 12, No. 13, 2010

highly stereoselective dithiane anion addition onto an
aldehyde, (3) a deprotection/cyclization sequence to
produce the dioxa-bicyclo[3.2.1]octane-2,3,4-triol motif
in high yield and with exclusive stereocontrol, and (4)
telescoping of several steps to add elements of practicality
(only one intermediate, 7, is isolated). Using an ap-
propriately substituted aryl group, this route also allows
for the efficient synthesis of any given compound from
this class of SGLT2 inhibitors. Lastly, the route provides
rapid access to C-1 and C-5 substituted dioxa-
bicyclo[3.2.1]octane-2,3,4-triol derivatives that could find
additional applications as biologically active compounds.
Acknowledgment. We thank B. Samas for X-ray crystal-
lography determination. We also thank Dr. J. W. Coe, Dr.
R. P. Robinson, and Dr. B. D. Stevens for reviewing this
manuscript. All are from Pfizer Global Research and
Development, Groton, CT, USA.
Supporting Information Available: Experimental pro-
cedures and characterization data for the process shown in
Scheme 2 (PDF). X-ray crystallographic data for 5 (CIF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
OL100940W
Org. Lett., Vol. 12, No. 13, 2010
2943