Commercial route research and development for SGLT2 inhibitor candidate ertugliflozin

Article abstract of DOI:10.1021/op4002802

A practical synthesis of SGLT2 inhibitor candidate ertugliflozin (1) has been developed for potential commercial application. The highly telescoped process involves only three intermediate isolations over a 12-step sequence. The dioxabicyclo[ 3.2.1]octane motif is prepared from commercially available 2,3,4,6-tetra-O-benzyl-D-glucose, with nucleophilic hydroxymethylation of a 5-ketogluconamide intermediate as a key step. The aglycone moiety is introduced via aryl anion addition to a methylpiperazine amide. High chemical purity of the API is assured through isolation of the crystalline penultimate intermediate, tetraacetate 39. A cocrystalline complex of the amorphous solid 1 with L-pyroglutamic acid has been prepared in order to improve the physical properties for manufacture and to ensure robust API quality.

Full text of DOI:10.1021/op4002802

Article
pubs.acs.org/OPRD
Commercial Route Research and Development for SGLT2 Inhibitor
Candidate Ertugliflozin
Paul Bowles,† Steven J. Brenek,† Stephane Caron,† Nga M. Do,† Michele T. Drexler,† Shengquan Duan,†
́
Pascal Dube,†,§ Eric C. Hansen,† Brian P. Jones,† Kris N. Jones,† Tomislav A. Ljubicic,†
́
Teresa W. Makowski,† Jason Mustakis,† Jade D. Nelson,*,† Mark Olivier,‡ Zhihui Peng,†
Hahdi H. Perfect,† David W. Place,† John A. Ragan,† John J. Salisbury,‡ Corey L. Stanchina,†
Brian C. Vanderplas,† Mark E. Webster,† and R. Matt Weekly†
‡
†Chemical Research and Development, Analytical Research and Development, Pfizer Worldwide Research and Development,
Eastern Point Road, Groton, Connecticut 06340, United States
S
* Supporting Information
ABSTRACT: A practical synthesis of SGLT2 inhibitor candidate ertugliflozin (1) has been developed for potential commercial
application. The highly telescoped process involves only three intermediate isolations over a 12-step sequence. The dioxa-
bicyclo[3.2.1]octane motif is prepared from commercially available 2,3,4,6-tetra-O-benzyl-D-glucose, with nucleophilic
hydroxymethylation of a 5-ketogluconamide intermediate as a key step. The aglycone moiety is introduced via aryl anion
addition to a methylpiperazine amide. High chemical purity of the API is assured through isolation of the crystalline penultimate
intermediate, tetraacetate 39. A cocrystalline complex of the amorphous solid 1 with L-pyroglutamic acid has been prepared in
order to improve the physical properties for manufacture and to ensure robust API quality.
ment, an attractive alternative synthesis was reported by the
medicinal chemistry team.7 This stereoselective route provided
1 in a much-improved 26% overall yield from diacetone-α-D-
mannofuranose. This approach was also deemed unsuitable for
large scale application, however, primarily due to a lack of
crystalline isolable intermediates and the requirement for
cryogenic reaction temperatures in setting key stereochemistry.
A more recent report8 describes the second-generation
synthesis developed and implemented for manufacture of API
in support of early clinical studies. Although this process was
successfully scaled to produce tens of kilograms of ertugliflozin
in a pilot plant setting, the significantly larger quantities of API
required for phase 3 and beyond, prompted development of a
more efficient, scale-friendly process.
INTRODUCTION
■
The synthetic C-aryl glycoside ertugliflozin 1 is a sodium
glucose cotransporter 2 (SGLT2) inhibitor currently in clinical
development for the potential treatment of type 2 diabetes
mellitus (Figure 1).1−4 A medicinal chemistry synthesis of 1
Figure 1. Structure of SGLT2 inhibitor candidate ertugliflozin (1).
Evaluating Synthetic Approaches to the Carbohy-
drate Core. Among the published syntheses of SGLT2
inhibitor candidates under pharmaceutical evaluation, the direct
arylation of protected gluconolactones of type 5 is most
prevalent (Scheme 1, Approach B).8−13 This strategy provides
rapid access to the target compounds and is highly convergent.
Recent advances, such as protection of the gluconolactone
hydroxyls as labile TMS ethers,14,15 have made this general
approach particularly attractive for large scale, since a dedicated
protecting group removal step is unnecessary. On the basis of
this precedent, the initial strategy to 1 targeted utilization of the
analogous approach C (Scheme 1). Thus, fully protected D-
gluconolactone analogues, with an additional hydroxymethyl
substituent at C5 (i.e., 7), were required for further conversion
to compounds 8 via aryl anion addition to the lactone carbonyl.
was designed to enable the preparation of analogues on gram
scale during candidate selection, but was undesirable for large,
multikilogram scale manufacture.5,6 This first-generation syn-
thesis involved 13 linear steps from D-glucose, performed in an
overall yield of 0.3% and required HPLC purification to isolate
1 from a mixture of C4 epimers. The key step in the route
involved arylation of Weinreb amide 2 with aryllithium 3
(Scheme 1, Approach A).
This general strategy involving arylation of an open-chain
gluconamide was potentially applicable for large scale synthesis
of 1, but a more expeditious approach to the point of
convergency would be required. Furthermore, it was clear that
the C5 tertiary alcohol of 2 needed to be protected prior to
introduction of the aryl anion. Not only does the free hydroxyl
consume one equivalent of the anion, it contributes to
epimerization at C2, presumably via intramolecular deprotona-
tion of ketone 4 through a six-membered transition state. As
the ertugliflozin program transitioned into clinical develop-
Received: October 7, 2013
Published: December 20, 2013
© 2013 American Chemical Society
66
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
Scheme 1. General synthetic approaches considered
Although our initial strategies employed D-glucose as the
primary building block for the preparation of the advanced
carbohydrate backbone of 1, we considered a wide range of
chiral pool candidates during design and development of the
second-generation synthesis (Figure 2). After a period of rapid
Unfortunately, 5-ketogluconic acid proved very challenging to
manipulate. This material lies predominately in the furanose
form under acidic conditions and readily decomposes at
alkaline pH. Protection of the hydroxyl groups in advance of
reaction with aryl nucleophiles was not forthcoming.
A sensible approach to 7 involved persilyl protection of D-
gluconolactone, followed by selective deprotection of the
primary TBS ether to form alcohol 9 (Scheme 2). Oxidation of
Scheme 2. Attempted preparation of 7a from D-
gluconolactone
the primary hydroxyl of 9 to the corresponding aldehyde 1019
was accomplished with Dess−Martin periodinane. However,
this compound proved to be very unstable, and was
incompatible with the reductive alkylation conditions required
for hydroxymethylation to diol 7a.7 Lactone hydrolysis and
elimination byproducts predominated.
We also attempted to access diols of type 7 by utilizing
tribenzyl-methyl-D-glucopyranoside (11), to minimize by-
products arising from degradation of lactone 10 (Scheme 3).
While the key reductive hydroxymethylation step was
comparatively more successful, an efficient sequence for
conversion of methylpyranoside 13 to lactone 7b was not
forthcoming. This lengthy, unproductive approach was
ultimately abandoned.
Attempted Arylation of Lactones 7. Next, the open-
chain protected amide utilized in the original medicinal
chemistry synthesis (i.e., Weinreb amide 2) was converted to
lactone 7c via selective deprotection of the PMB ethers
followed by acid-catalyzed lactonization. While clearly not a
feasible long-term strategy, owing to the large number of steps
associated with preparation of amide 2, this allowed for quick
evaluation of lactones of type 7 in the key arylation reaction
(Scheme 4, reaction A). Surprisingly, reaction of 7c with aryl
anions failed to provide the desired C-aryl glycoside 8a;
unsaturated lactone 14a was the only product identified. In
silico models suggested the additional substitution at C5 may
be blocking approach of the bulky aryl nucleophile, although
Figure 2. Potential precursors to 1 from the chiral pool.
route scouting on small scale, the most promising synthetic
approaches were further assessed for viability on larger scale.
Several routes reached a proof of concept phase in the
laboratory, but were either relatively lengthy or exceedingly
costly, making them unsuitable for commercialization (e.g., D-
apiose, L-arabinose, and L-tartaric acid approaches).
Other carbohydrates highlighted in Figure 2 failed to provide
access to useful advanced intermediates due to a gap in
technology. For example, D-glucal has been showed to engage
in Heck-type C-arylation reactions.16,17 However, β-hydride
elimination of the intermediate alkyl palladium complex leads
to the C3−C4 olefin, destroying the C3 stereocenter. Likewise,
D-galactose has been utilized in a high-yielding C1 hydrox-
ymethylation with paraformaldehyde.18 However, known
pathways for inversion of the C2 stereocenter of galactose
were unsuccessful on the C1 homologated substrates.
67
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
Scheme 3. Attempted preparation of 7b from methyl-D-glucopyranose
Scheme 4. Arylation attempts with C6-substituted gluconolactone derivatives
Scheme 5. Arylation attempts with cyclic carbonate 7e
Scheme 6. Preparation of open-chain substrates
challenges in accomplishing nucleophilic aryl anion addition to
gluconolactones lacking a C5 substituent have also been
documented.20
To test the hypothesis regarding steric hindrance, the PMB
groups on 7c were removed to provide diol 7b, which was
subsequently converted to the corresponding dimethylaceto-
nide 7d (Scheme 4, reaction B). Models now suggested at least
one face of the lactone carbonyl was accessible for attack by aryl
nucleophiles, but multiple attempts yielded only elimination
product 14b.
Cyclic carbonate 7e was then prepared from diol 7b and
CDI, albeit in very low yield. This lactone was also evaluated in
a handful of arylation reactions (Scheme 5). Interestingly,
addition of aryllithium 3 to 7e provided desired C1 addition
product 8a; no elimination product was observed. Furthermore,
the intermediate alkoxide 8a partially converted to the bridging
ether 15 prior to quench. While this was clearly a breakthrough
for the “lactone approach” to API 1, carbonate 7e proved to be
very difficult to prepare from diol 7b or other reasonable
precursors.21 Furthermore, the long, linear sequences required
to access carbonate 7e from a suitable chiral pool source made
this strategy impractical for large scale.
The addition of a hydroxymethyl substituent at C5 of the
gluconolactone (e.g., 7) has a profound impact on the outcome
of this arylation reaction (Scheme 1, approach C). We turned
focus to acyclic analogues of the carbohydrate (i.e.,
gluconamides such as 2), according to precedented reactivity
from the first-generation synthesis (Scheme 1, approach A).
Evaluation of D-Gluconic Acid Derivatives. Various D-
gluconic acid analogues 16a−d were prepared via the two-step
sequence illustrated in Scheme 6. Benzyl protecting groups
were chosen for their ability to tolerate a broad range of
reaction conditions, the extensive precedent in carbohydrate
chemistry, and the relative ease of their removal via
hydrogenolysis.22 Thus, compound 17 was converted to lactone
5a under modified Swern conditions in nearly quantitative
68
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
Scheme 7. Complementary hydroxymethylation strategies
Scheme 8. Nucleophilic hydroxymethylation of ketoamide 21
yield.23 A range of oxidants was evaluated for this trans-
formation; TPAP-NMO24 and TEMPO-NaOCl9 are among
the conditions that have been reported, and these were also
effective on laboratory scale. Lactone 5a was then converted to
esters 16a and 16b, or amides 16c and 16d, by treatment with
the appropriate alcohol or amine, respectively. Compounds
were selected for their presumed reactivity in the downstream
aryl anion addition step.25−27 As anticipated, hydroxyesters 16a
and 16b converted back to lactone 5a at an appreciable rate
even during storage as isolated oils held under refrigeration.
The morpholine amide 16c was significantly more stable in its
open-chain form but was a thick oil that resisted crystallization
despite considerable effort. The methylpiperazine derivative
16d, containing a basic tertiary nitrogen, was prepared to
enable salt formation that could render the compound
crystalline and easier to handle. Fortuitously, 16d was found
to be a crystalline solid as its free base at ambient temperature,
requiring no salt formation for its isolation.
Introducing the Key Hydroxymethyl Substituent. With
a reliable supply of hydroxyamide 16d in hand, the key
imperative became introduction of the hydroxymethyl sub-
stituent. Two complementary approaches for this trans-
formation were considered (Scheme 7). The first (approach
A) required selective oxidation of the C6 primary alcohol of 18
to the corresponding aldehyde 19.28 The resulting compound
could then participate in an aldol condensation with form-
aldehyde. Utilization of a stoichiometric excess of formaldehyde
would promote Cannizzaro-type in situ reduction of the
aldehyde and lead to the desired diol 20.7 Challenges associated
with this strategy were quickly realized. Perhaps most
disconcerting was the relative incompatibility of the gluconic
acid derivatives to the alkaline conditions required for the
hydroxymethylation. Elimination, retro-aldol, and hydrolysis
byproducts were observed. Furthermore, the oxidation step was
not straightforward, as formation of the carboxylic acid (i.e.,
overoxidation product) was difficult to control. In parallel with
efforts on “electrophilic hydroxymethylation” (i.e., formalde-
hyde as a source of CH2−OH) was a study of alternative
“nucleophilic” strategies (i.e., ROCH2-M) for hydroxymethyla-
tion to convert ketoamide 21, obtained from secondary alcohol
16d via Parikh−Doering29 oxidation, to diol 22 (Scheme 7,
approach B).
Ketoamide 21 proved to be a versatile intermediate, as
several options were successfully demonstrated in the
laboratory for preparation of diol diastereomers 22a/b
(Scheme 8). Although not ideal for large scale, the two-step
sequence involving Wittig methylenation, followed by dihy-
droxylation with OsO4 provided 22a/b in excellent overall
yield. Treatment of ketoamide 21 with the Corey−Chaykovsky
reagent30 provided the corresponding epoxide 24a/b in
reasonable unoptimized yield; however, a brief study aimed at
further conversion to diol 22a/b was unsuccessful.31
Critical Evaluation of Hydroxymethyl Anion Equiv-
alents. The third, and ultimately most successful “nucleophilic
hydroxymethylation” approach studied, involved reaction of 21
with a one-carbon nucleophile to install the protected
hydroxymethyl group, providing tertiary alcohol 22a/b after
removal of the primary hydroxyl protecting group from 25a/b.
Numerous hydroxymethyl anion synthons are prece-
dented,32−36 but most have significant drawbacks, including:
(1) Cryogenic reaction temperatures are often required due
to poor solution stability of the hydroxymethyl anions.
(2) Precursors to the anions are often toxic and/or
mutagenic alkyl halides.
69
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
Scheme 9. “Nucleophilic” hydroxymethylation via Grignard reagents
Scheme 10. Preparation of crystalline intermediates 30a/b·oxalate and 30a free base
Table 1. Preparation of aryl bromide pre-nucleophiles
a
b
entry
ether 34
R
isolated yield
MP (°C)
1
2
3
4
5
6
34a
34b
34c
34d
34e
Et
98.8
74.3
79.0
40.0
54.2
i-Pr
n-Bu
Ph
35.7
c
0
N/A
62.3
Bn
88.6
80.5
34f
Ph(CH2)2
56.3
a
b
c
Yields for conversion of 33 to 34a−f. Melting point onset determined by DSC. The major product was Friedel−Crafts C4-alkylation of PhOH.
(3) Analytical characterization and in-process controls
(IPCs) are challenging to develop and implement in a
manufacturing environment.
(4) The additional step required to liberate the alcohol is
often incompatible with sensitive functionality in the
reaction product.
pivalate 28 (Scheme 9, option B). Developed largely by
Knochel and co-workers,35 this chemistry is applicable to a
broad range of electrophiles. The reagent is prepared via
Grignard exchange with iPrMgCl at low temperature. This
reagent also showed excellent selectivity for addition to the
ketone of 21, providing the intermediate pivalate ester 29a/b as
the exclusive addition product. Interestingly, this reagent
exhibits high facial selectivity at −78 °C, providing a 95:5
mixture of diastereoisomers at C5, favoring the 5R isomer, 29a.
The diols 22a/b could be liberated by reaction with solid
NaOMe in toluene or cyclopentyl methyl ether (CPME). The
two nucleophilic hydroxymethylation processes performed in
comparable yield during laboratory trials, providing 22a/b from
21 in ∼75% overall yield.
Preparation of Piperazine Amide 30a/b·Oxalate. Many
attempts to protect diols 22a/b in advance of the arylation step
failed, presumably due to steric hindrance surrounding the
tertiary alcohol. Success was eventually realized, however, by
forming the dimethylacetonide 30a/b with acetone or DMP in
the presence of MsOH or p-TsOH (Scheme 10). The
diastereomeric mixture of products 30a/b was found to be a
sticky oil, and thus a salt was formed between the basic tertiary
nitrogen of 30a/b and oxalic acid. The identification of a stable
crystalline intermediate at this point in the synthesis was a
critical breakthrough, as this allowed for effective purge of
Nevertheless, precedented reagents for nucleophilic hydrox-
ymethylation were evaluated, concentrating on those with the
greatest potential for success in large-scale applications. An
effective reagent would need to be highly chemoselective (for
ketone vs amide) and liberate the desired diol under mild
conditions in order to avoid epimerization or decomposition of
the labile product. While no single reagent satisfied all criteria,
it was determined that two deserved additional consideration
(Scheme 9).
The first Grignard reagent evaluated is derived from
chloromethylsiloxane 26 (Scheme 9, Option A).37 This
Grignard exhibits excellent solution-stability and exquisite
selectivity for addition to the ketone function of ketoamide
21. At a typical reaction temperature near −20 °C, intermediate
27a/b was prepared as a 3:2 mixture of diastereoisomers
(favoring the 5R isomer), but both compounds converge to a
single product later in the synthesis. Tamao−Fleming oxidation
liberates the masked diol isomers 22a/b.36,38−41 The second
method involves the Grignard reagent derived from iodomethyl
70
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
Scheme 11. Preparation of amorphous ertugliflozin (1)
Scheme 12. Options for deprotection/cyclization of 35
process-related impurities such as minor C2 and C4
diastereoisomers (resulting from epimerization during process-
ing), unreacted starting materials, reactants, and reagents, and
inorganic materials. The upgrade in purity immediately before
the key convergent (i.e., arylation) step also translated to
increased process performance in the end-game operations.
Interestingly, neutralization of the 30a/b oxalate salt with
aqueous NaOH in MTBE, followed by removal of water via
azeotropic distillation with cyclohexane, provided the crystalline
free base 30a (5R diastereomer) as the exclusive product. Both
Grignard options A and B (Scheme 9) were effective in
producing 30a via this protocol.
Identification of an Optimal Aryl Bromide. With a
crystalline amide intermediate in hand, we turned our attention
to identification of a suitable aryl pre-nucleophile. We desired a
compound that could be prepared via a short, scale-friendly
synthetic sequence and offered excellent crystallinity and
stability that allowed for bulk preparation and long-term
storage, since we envisioned this building block would serve as
one of two regulatory starting materials if the synthesis was
commercialized. After an evaluation of several related building
blocks, we decided to target benzhydryl ethers, as described in
Table 1.42 The two-step, one-pot sequence begins by halogen
metal exchange between 1-bromo-4-ethoxybenzene 31 and an
alkyllithium (e.g., n-BuLi or n-HexLi), or via Grignard
formation, to provide the corresponding aryl anion. This
intermediate is combined with 5-bromo-2-chlorobenzaldehyde
32 to afford benzhydryl alcohol 25 in excellent yield, as
determined by UHPLC/MS analysis. Intermediate 33 may be
isolated or treated directly with an alcohol in the presence of
H2SO4 to provide the corresponding benzhydryl ethers 34a−f.
Evaluation of the properties of each, and performance in the
subsequent reaction with amide 30, led us to selection of the
benzyl-protected ether 34e for further development. Both batch
and flow reactor options have been demonstrated for the
preparation of ether 34e on multigram scale.
Preparation of Aryl Ketone 35. In order to complete the
synthesis of ertugliflozin, the crystalline salt 30a/b·oxalate was
first converted to the corresponding free base.43 Although the
5R isomer 30a could be isolated as a crystalline solid from
cyclohexane (Scheme 10), both diastereomers were equally
effective in downstream chemistry, so isolation was typically
avoided. Instead, a solution of 30a/b was dried via azeotropic
distillation with toluene and then was reacted with the aryl
anion obtained by treatment of aryl bromide 34e with n-BuLi
or n-HexLi in THF (Scheme 11).44,45 Only monoaddition to
amide 30a/b was observed, even when multiple equivalents of
aryl anions were employed. Computational modeling suggests
strong stabilization of the tetrahedral intermediate, formed
upon addition of the aryl anion, via intramolecular chelation to
both C2- and C3-benzyloxy substituents.46 It should also be
noted that no detectable epimerization of the C2 stereocenter
was observed in this step. Following an acidic quench and
aqueous partition, a complex mixture of diastereomers 35 was
obtained due to the presence of a racemic stereocenter at the
benzhydryl ether and a variable mixture of isomers created in
the hydroxymethylation.47 The complex mixture of isomers
presumably contributes to the undesirable physical properties
71
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
Scheme 13. Conversion of 37a/b to 1
Scheme 14. Preparation of cocrystalline 1·L-PGA
of this compound, rendering it a gummy oil after silica gel
chromatography. As a result, this material was not isolated, but
was carried directly into subsequent steps as a solution.
providing the desired isomer of 1 as the only detectable
product. Analysis of the crude reaction mixture against a
purified standard of 1 confirmed >75% overall yield for the
three-step sequence.
Evaluation of Deprotection and Glycosylation Se-
quences. The aryl ketone 35 is an open-chain, fully protected
form of the final synthetic target 1. Several strategies were
evaluated for sequential removal of the acetonide, benzyl ethers,
and benzhydryl substituent, with a focus on balancing chemical
yields and product purity profile with overall process efficiency
(Scheme 12).48 The most attractive option involved global
deprotection and ether formation/thermodynamic equilibration
in a single step. Thus, 35 was subjected to catalytic
hydrogenolysis under acidic conditions. It was found that the
benzyl ethers and acetonide were cleaved readily, and near-
complete etherification resulted, but a mixture of benzhydryl
alcohols 36a and methyl ethers 36b were obtained. Attempts to
cleave these benzhydryl substituents under more forcing
hydrogenolysis conditions led to competitive reduction of the
aryl chloride, as well as C2 epimerization and traces of C1
reduction. Treatment of the mixture with various reducing
agents, with and without acid catalysis, produced desired
product 1 with modest purity. Therefore, we chose a more
conservative, stepwise approach for our end-game. Treatment
of crude solutions of 35 with a mild hydride source (e.g.,
Et3SiH) in the presence of an acid (e.g., TFA) led to
deprotection of both the acetonide and benzhydryl ethers,
affording a mixture of two diastereomers 37a/b. This mixture
could not be readily isolated and purified via crystallization, as it
was found to exist as a semisolid. As a result, this intermediate
was also used directly in subsequent steps as a crude solution.
Fortunately, both isomers 37a/b converged to the desired
target compound 1 during removal of the benzyl groups via
metal-catalyzed hydrogenolysis under acidic conditions
(Scheme 13). In the absence of acid, the debenzylation
proceeded smoothly, but a mixture of bridging ether
diastereoisomers of 1 was obtained. Treatment of this mixture
with HCl promoted equilibration of the bridging ethers,
Preparation of a Crystalline Analogue for Purity
Upgrade. Unfortunately, utilization of the crude stream of 1
produced via this sequence yielded cocrystal with unacceptable
chemical purity for pharmaceutical applications. Therefore,
opportunities for derivatization of 1 that would enable
purification via crystallization without adding excessive cost or
complexity were targeted.49 Both the primary monoacetate 38
and tetraacetate derivative 39 were prepared by reaction with
the appropriate stoichiometric quantity of Ac2O or AcCl and a
suitable base (e.g., NMI, pyridine, or iPr2NEt) (Scheme 14).50
The acetate products could be recrystallized to high purity from
a range of solvents (e.g., iPrOH, MeOH). Several related
compounds (e.g., pivalate, benzoate) were also prepared via
analogous protocols, but the low cost, high crystallinity, and
reasonably low molecular weight of the acetates made them the
most sensible choice for large-scale use. Furthermore, the
crystallization processes for isolation of 38 and 39 were found
to be particularly effective at purging process-related impurities
to very low levels. Since formation of monoacetate 38 required
strict control of parameters that could prove challenging on
large scale, tetraacetate 39 was pursued for further optimization.
Preparation of Ertugliflozin (1)·L-PGA. Following the
purity upgrade offered by isolation of 39, acetate groups were
removed with catalytic NaOH or NaOMe, providing a fairly
pure solution of 1. Despite extensive screening, a crystalline free
form of ertugliflozin (1) has not been identified. This
compound exists as a hygroscopic amorphous solid with a
low glass transition temperature. Since the compound is also
nonionizable near physiological pH, salt forms were not
pursued as a viable option. Instead, ertugliflozin (1) was
subjected to solvate and cocrystal form screening. Only two
crystalline forms have been identified: L-proline and L-
pyroglutamic acid (L-PGA) cocrystals. As the L-PGA cocrystal
72
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
proved to possess superior physical properties, it became the
focus of further development.
conditions of 100% acetonitrile. Approximate retention times
(min): 30a/b (13.3−13.5), 34e (20.1), 35 (25.2−25.6), 37a/b
(23.4−23.6), 1 (10.6), 38 (12.6), 39 (17.3).
Thus, the crude solution of ertugliflozin (1) was converted to
its L-PGA cocrystal by combination with L-PGA acid in a
mixture of iPrOH and water. Despite the miscibility of these
two solvents under normal circumstances, the presence of 1
and L-PGA leads to an interesting biphase. The more
hydrophobic 1 resides primarily in i-PrOH droplets, while the
polar L-PGA resides in the aqueous phase. Cocrystallization
occurs with high efficiency at the interface (Figure 3). The
The powder X-ray diffraction patterns of monoacetate 38
(Figure 2) and tetraacetate 39 (Figure 3) were carried out on a
Bruker AXS - D8 Advance diffractometer using Cu radiation
source. The tube voltage and amperage were set to 40 kV and
40 mA, respectively. The divergence slit was fixed at 1 mm,
whereas the scattering and the receiving slits were set at 0.6
mm. Diffracted radiation was detected by a scintillation counter
detector. Data were collected in the θ−θ goniometer at the Cu
wavelength Kα1 = 1.54056 Å from 3.0° to 40.0° 2θ using a step
size of 0.040° and a step time of 2.0 s. Samples were prepared
by placement in a Nickel Disk (Gasser & Sons, Inc. Commack,
NY) and rotated during data collection. Data were collected
and analyzed using Bruker Diffrac Plus software (Version 2.6).
Bruker AXS DIFFRACplus Basic EVA 12 software was used to
visualize and evaluate the PXRD diffractograms. PXRD data
files (.raw) were not processed prior to peak searching.
Generally, a threshold value of 2.0 and a width value of 0.3 were
used to make preliminary peak assignments. The results are
summarized in the Supporting Information.
(3R,4S,5R,6R)-3,4,5-Tris(benzyloxy)-6-((benzyloxy)-
methyl)tetrahydro-2H-pyran-2-one (5a). A solution of
2,3,4,6-tetra-O-benzyl-D-glucopyranose 17 (25.0 g) in dimethyl
sulfoxide (100 mL, 4 L/kg) was stirred at 20−25 °C for 30 min
under a nitrogen atmosphere. Acetic anhydride (75 mL, 3 L/
kg) was added at a rate of 5 mL/min (∼15 min) at 15−25 °C.
After complete addition, the reaction mixture was stirred for
18−24 h at 20−25 °C. After complete reaction was confirmed
by UHPLC/MS analysis, the reaction mixture was diluted with
toluene (225 mL, 9 L/kg) and a 3.3N aqueous solution of HCl
(225 mL, 9 L/kg) was slowly added to quench the excess acetic
anhydride. The reaction mixture was stirred for 20 min at 20−
25 °C to complete this quench. The phases were separated, the
aqueous layer was discarded, and the organic layer was washed
with a 2 M aqueous phosphate buffer (pH 7, 225 mL, 9 L/kg).
After stirring for 20 min at 20−25 °C, the phases were
separated, and the organic layer was dried over sodium sulfate
(as needed) and filtered. Analysis of the filtrate indicated 26.9 g
of lactone 5a was present (924 mg/g potency, 94.6% purity,
92% yield).
Figure 3. Biphasic cocrystallization of ertugliflozin (1)·L-PGA.
cocrystalline product is driven from this biphase in high yield
by use of a stoichiometric excess of L-PGA. A reslurry in either
toluene or acetonitrile provides cocrystalline 1·L-PGA complex
of high analytical purity and with an approximate stoichiometric
ratio of 1:1 (1:L-PGA).
CONCLUSION
■
In conclusion, an efficient, stereoselective synthesis of SGLT2
inhibitor candidate ertugliflozin (1) was identified and
developed (Scheme 15). The 12-step process starts from
tetra-O-benzyl-D-glucose 17 and includes only three isolated
intermediates (16d, 30a/b·oxalate, and 39) over the longest
linear sequence, due to a lack of crystallinity of most
intermediates. Highlights of this process include two
regioselective, scale-friendly options for nucleophilic hydrox-
ymethylation of ketogluconamide 21, a highly efficient arylation
of piperazine amide 30a/b as the key convergent step, a global
acid-promoted deprotection sequence that allows multiple
isomers to converge at a single desired product, and the use of
the highly crystalline late-stage tetraacetate 39 to purge process-
related impurities prior to isolation of the final cocrystalline
ertugliflozin (1)·L-PGA complex.
(2R,3S,4R,5R)-2,3,4,6-Tetrakis(benzyloxy)-5-hydroxy-
1-(4-methylpiperazin-1-yl)hexan-1-one (16d). A solution
of tetra-O-benzyl gluconolactone 5a (14.7 wt/wt%, 50 g) in
toluene (340 g total solution mass) was stirred at 20−25 °C
under nitrogen. N-Methylpiperazine (23.24 g, 2.5 equiv) was
then added at a rate of ∼0.5 mL/min (∼52 min) at ≤30 °C.
After complete addition, the reaction mixture was stirred at
20−25 °C for 18 h. Complete reaction was confirmed by
UHPLC/MS analysis, and then water (250 mL, 5 L/kg) was
added, the biphasic mixture was stirred for 20 min at 20−25 °C,
and then the aqueous layer was discarded. After a second
identical treatment with water (250 mL, 5 L/kg), heptane (500
mL, 10 L/kg) was added at a rate of 10 mL/min to produce a
slurry. After complete addition, the slurry was cooled to 10−15
°C and stirred for at least 30 min. Solids were collected on a
EXPERIMENTAL SECTION
■
General. All starting materials were prepared by Pfizer
Chemical Research and Development (Groton, CT). All
reactions were monitored by reverse phase UHPLC using a
Waters Acquity LC/PDA/SQD equipped with a CSH phenyl-
hexyl (100 mm × 2.1 mm, 1.7 μm) column with a column
temperature of 45 °C and a flow rate of 0.4 mL/min. A 30-min
linear gradient was employed, with initial conditions of 95%
aqueous trifluoroacetic acid and 5% acetonitrile and final
Buchner funnel, rinsed with heptane (100 mL, 2 L/kg), and
̈
dried under vacuum at 40 °C for 12−18 h to provide 16d (59.3
g) as a white solid (MP = 94−95 °C). UHPLC-MS: Potency
914 mg/g, purity 96.7%, 76.2% from 17. 1H NMR (DMSO-d6,
400 MHz, 25 °C): δ 7.44−7.13 (m, 20H), 5.10 (d, J = 5.6 Hz,
1H), 4.68 (s, 2H), 4.64−4.42 (m, 7H), 4.17 (dd, J = 6.6, 4.2
73
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
Scheme 15. Preparation of ertugliflozin (1)·L-PGA
Hz, 1H), 3.96−3.84 (m, 1H), 3.72−3.60 (m, 2H), 3.60−3.39
(m, 5H), 2.28 −2.00 (m, 4H), 2.08 (s, 3H). 13C NMR
(DMSO-d6, 101 MHz, 25 °C): δ 167.2, 138.7, 138.7, 138.4,
137.8, 128.2, 128.2, 128.1, 128.1, 127.8, 127.7, 127.6, 127.5,
127.4, 127.3, 127.2, 79.8, 79.8, 79.5, 74.6, 73.1, 72.4, 71.6, 71.5,
69.5, 54.6, 54.3, 45.5, 44.6, 41.6. HRMS: (ESI+) Calcd for
C39H47N2O6 (M + H)+: 639.34286, Found: 639.34228.
(2R,3S,4S)-2,3,4,6-Tetrakis(benzyloxy)-1-(4-methylpi-
perazin-1-yl)hexane-1,5-dione (21). To a pale-yellow
solution of 16d (154 g, 241.1 mmol) in toluene (462 mL, 3
L/kg,) and DMSO (308 mL, 2 L/kg) under nitrogen was
charged N,N-diisopropylethylamine (186.9 g, 252 mL, 1446
mmol,) at 20−25 °C. After this addition, the hazy solution was
cooled to 5 °C. A solution of sulfur trioxide pyridine complex
(76.74 g, 482.2 mmol) in DMSO (308 mL, 2 L/kg) was then
added at ≤15 °C (ca. 30 min addition time). The reaction
mixture was stirred at 5−10 °C until complete conversion was
confirmed by UHPLC-MS analysis. Toluene (462 mL, 3 L/kg)
was then added, followed by the slow addition of ammonium
hydroxide (5% w/w in water, 924 mL, 6 L/kg). Following
complete addition, the mixture was stirred for 10 min at 20−25
°C (pH 10−11), and then the phases were separated. The
organic phase was washed with saturated aqueous sodium
chloride (462 mL, 3 L/kg) and then was concentrated under
vacuum to ∼450 mL. Toluene (924 mL, 6 L/kg) was added,
and the solution was concentrated under vacuum to ∼770 mL
(water content ≤0.06% per KF analysis). A solution potency
measurement (UHPLC/MS) indicated 90% yield for ketone
21. The bulk crude product was utilized directly in the
subsequent step; however, a small portion of the mixture was
removed and concentrated for characterization of ketone 21
1
(semisolid). H NMR (DMSO-d6, 400 MHz, 25 °C): δ 7.07−
7.45 (m, 20H), 4.28−4.70 (m, 8H) 4.43 (d, J = 11.3 Hz, 1H),
4.50 (d, J = 11.3 Hz, 1H), 4.54 (d, J = 11.3 Hz, 1H), 4.66 (d, J
= 11.3 Hz, 1H), 4.23 (dd, J = 5.9, 4.0 Hz, 1H), 3.24−3.60 (m,
4H), 1.94−2.27 (m, 4H), 2.08 (s, 3H). 13C NMR (DMSO-d6,
101 MHz, 25 °C): δ 206.0, 166.8, 137.8, 137.7, 137.6, 137.4,
128.3, 128.2, 128.2, 128.2, 127.9, 127.9, 127.7, 127.7, 127.5,
127.5, 81.6, 79.9, 77.4, 73.8, 73.6, 72.8, 72.1, 71.4, 54.6, 54.2,
45.4, 44.6, 41.5. HRMS: (ESI+) Calcd for C39H44N2O6Na (M +
Na)+: 659.30916, Found: 659.30872.
74
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
(2R,3S,4S)-2,3,4-Tris(benzyloxy)-4-(4-((benzyloxy)-
methyl)-2,2-dimethyl-1,3-dioxolan-4-yl)-1-(4-methylpi-
perazin-1-yl)butan-1-one oxalate (30a/b). 30a:30b (1.4:1)
via Grignard Option A. Grignard. A suspension of magnesium
filings (16.72 g, 687.8 mmol) in dry THF (366 mL) was heated
to 65 °C with high agitation under an atmosphere of nitrogen.
1,2-Dibromoethane (3.88 g, 20.63 mmol) was then added over
5 min, followed by the slow addition of chloromethyl
dimethylisopropoxysilane 26 (86.01 g, 515.9 mmol) in THF
(86 mL).
CAUTION: An exotherm of 2 °C occurred during the first
20% of the addition. This exothermic event could be
considerably greater if the induction period is extended and
accumulation occurs; thus, an addition period of at least 30 min
is advisable, along with suitable headspace in the reaction vessel
to account for foaming and rapid boiling.
Following complete addition of silane 26, the reaction was
stirred for 1.5 h at 65 °C. GC analysis of an aliquot that had
been quenched into isopropyl alcohol and filtered indicated
<2% 26 remaining. The solution of Grignard reagent thus
prepared was then cooled to −25 °C for immediate use in the
next step. Alternately, this solution (1.0−1.1M) was stored
between 5 and 25 °C for up to 4 days with no significant loss of
potency.
To the solution of Grignard (515.9 mmol) was added a
solution of 21 (146 g, 229.3 mmol) in cyclopentyl methyl ether
(CPME) (292 mL) at ≤15 °C. Following complete addition,
the reaction was stirred for an additional 30 min at −20 °C, and
then analysis by UHPLC/MS confirmed complete conversion
to methylsilane adduct 27a/b (target specification ≤2% 21).
The reaction was quenched by the slow addition of saturated
aqueous ammonium chloride (365 mL) and then was diluted
with water (365 mL) and CPME (438 mL). The temperature
was allowed to warm to a maximum of 5 °C during the addition
and was further warmed to 20 °C after complete addition. The
mixture was filtered to remove any remaining Mg residues, and
then the phases were separated. The organic phase was washed
with water (365 mL) and then analyzed for potency of 27a/b.
The bulk crude product was utilized directly in the subsequent
step; however, a small portion of the mixture was removed and
concentrated for characterization of 27a/b (semisolid). 1H ): δ
(DMSO-d6, 600 MHz, 25 °C): δ 7.39−7.14 (m, 20H), 4.80 (d,
J = 11.5 Hz, 0.5H), 4.73 (d, J = 11.5 Hz, 1H), 4.63−4.36 (m,
7.5H), 4.24 (dd, J = 5.1, 4.4 Hz, 0.5H), 4.04 (b, 1H), 3.92 (sep,
J = 6.1 Hz, 0.5H), 3.86 (sep, J = 6.1 Hz, 0.5H), 3.70 (b, 1H),
3.66 (d, J = 4.4 Hz, 0.5H), 3.53−3.45 (b, 2H), 3.53 (d, J = 9.1
Hz, 0.5H), 3.49 (d, J = 9.3 Hz, 0.5H), 3.45 (d, J = 9.3 Hz,
0.5H), 3.42 (d, J = 9.1 Hz, 0.5H), 3.35 (b, 1H), 2.26−2.08 (b,
4H), 1.11 (d, J = 14.9 Hz, 0.5H), 1.04−0.99 (m, 1H), 1.04 (d, J
= 6.1 Hz, 1.5H), 1.03 (d, J = 6.1 Hz, 1.5H), 1.00 (d, J = 6.1 Hz,
1.5H), 0.99 (d, J = 6.1 Hz, 1.5H), 0.09 (s, 1.5H), 0.06 (s, 3H),
0.05 (s, 1.5H). 13C NMR (DMSO-d6, 150.8 MHz, 25 °C): δ
167.60, 167.50, 139.33, 139.02, 138.67, 138.56, 138.29, 138.23,
138.08, 137.66, 128.16, 128.12, 128.08, 128.07, 128.00, 127.97,
127.93, 127.80, 127.78, 127.75, 127.72, 127.64, 127.55, 127.41,
127.39, 127.28, 127.19, 127.09, 126.97, 82.02, 81.67, 81.49,
78.60, 77.94, 76.95, 76.02, 75.40, 74.78, 74.56, 73.67, 73.53,
73.38, 73.15, 72.52, 72.27, 71.99, 71.69, 64.07, 63.96, 54.77,
54.68, 54.33, 45.50, 44.53, 44.31, 41.65, 41.58, 25.71, 1.14.
HRMS: (ESI+) Calcd for C45H61N2O7Si (M + H)+: 769.42428,
Found: 769.42377.
concentration of 5 L/kg by either addition of CPME or
concentration at 35 °C (30−35 mbar). Methanol (517.5 mL)
was then added, followed by KF (14.25 g, 245.3 mmol) and
sodium bicarbonate (12.6 g, 149.5 mmol) at 20 °C. Hydrogen
peroxide (35 mass % in water, 52 mL, 598 mmol) was then
added dropwise over approximately 10 min.
CAUTION: The 35% hydrogen peroxide was diluted to 29%
with water prior to addition for safety; an exotherm of 3 °C was
observed during the addition of the diluted peroxide at this
scale.
Following complete addition, the reaction was stirred for 18
h at 25 °C, and then analysis of a sample quenched into
aqueous sodium bisulfite indicated complete conversion to diol
22a/b51 (target specification ≤2.5% 27a/b or analogues52).
The reaction was then slowly added to a solution of sodium
bisulfite (77.81 g, 747.7 mmol) in water (575 mL) and stirred
for 15 min.
CAUTION: This is a highly exothermic quench that must be
carried out with caution. Precooling the quench mixture to 5
°C is recommended. The complete consumption of peroxide
was confirmed with test strips before proceeding.
Following the quench operation, the reaction mixture was
warmed to 20 °C, CPME (345 mL) was added, and the mixture
was stirred for 15 min. The layers were then separated, the
organic phase was washed with saturated aqueous NaCl (345
mL), and the solution was carefully concentrated under vacuum
with a jacket temperature of 35 °C to 345 mL (3 L/kg). CPME
(805 mL) was then added, and the solution was concentrated
under vacuum with a jacket temperature of 35 °C to
approximately 690 mL (6 L/kg, target specification KF ≤
0.06%). In situ yield of 22a/b (determined by UHPLC via
comparison to a chromatographed standard) was typically
>75%. The bulk crude product was utilized directly in the
subsequent step; however, a small portion of the mixture was
removed and concentrated for characterization of diol 22a/b
1
(semisolid). H NMR (DMSO-d6, 600 MHz, 25 °C): δ 7.38−
7.16 (m, 20H), 4.74 (d, J = 11.5 Hz, 0.62H), 4.73 (d, J = 11.5
Hz, 0.38H), 4.70 (d, J = 11.3 Hz, 0.38H), 4.65 (d, J = 11.3 Hz,
0.62H), 4.59 (d, J = 11.3 Hz, 0.62H), 4.56 (d, J = 11.3 Hz,
0.76H), 4.55−4.45 (m, 4.62H), 4.44 (d, J = 11.5 Hz, 0.62H),
4.38 (d, J = 11.3 Hz, 0.38H), 4.30−4.27 (m, 1H), 3.97 (d, J =
5.3 Hz, 0.38H), 3.85 (d, J = 5.0 Hz, 0.62H), 3.72−3.38 (m,
8H), 2.32−2.11 (b, 4H), 2.13 (s, 1.14H), 2.11 (s, 1.86H). 13C
NMR (DMSO-d6, 150.8 MHz, 25 °C): δ 167.73, 167.72, 139.8,
139.0, 138.70, 138.67, 138.6, 138.53, 137.84, 137.83, 128.17,
128.13, 128.12, 128.0, 128.00, 127.97, 127.81, 127.80, 127.8,
127.7, 127.6, 127.5, 127.4, 127.3, 127.23, 127.20, 127.1, 127.0,
82.1, 81.8, 78.3, 77.9, 76.3, 76.1, 73.8, 73.74, 73.69, 73.6, 72.7,
72.5, 72.01, 71.95, 71.7, 62.9, 62.8, 54.4, 54.1, 45.1, 44.2, 41.4.
HRMS: (ESI+) Calcd for C40H49N2O7 (M + H)+: 669.35343,
Found: 669.35321.
Acetonide and Oxalate Formation. To a solution of 22a/b
(100 g) in CPME (6 L/kg; 600 mL), prepared as described
above, was added 2,2-dimethoxypropane (4 equiv; 62.29 g) at
20−25 °C. Methanesulfonic acid (2.5 equiv; 35.92 g) was
added, and then the reaction was stirred at 20−25 °C for 30
min. Analysis by UHPLC indicated complete reaction (target
specification ≤2% 22a/b). The reaction was transferred into a
vessel containing 20% aqueous dipotassium phosphate solution
(600 mL, 6 L/kg), and the biphasic mixture was stirred for 15
min. The phases were split, and the upper organic phase was
washed with saturated aqueous NaCl (5 L/kg; 500 mL). The
organic layer was then dried and concentrated via distillation to
Oxidation. The crude solution of 27 prepared above (115 g,
149.5 mmol) in THF/CPME (575 mL) was adjusted to a
75
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
250 mL total reaction volume (internal temperature ≤ 106 °C).
The reaction mixture was then cooled to 50 °C where a
solution of oxalic acid (14.81 g, 1.1 equiv) in MtBE (600 mL)
was added over 100 min (1 mL/min). The resulting mixture
was stirred for an additional 30 min at 50 °C following
complete addition and was then seeded with 30a/b·oxalate
(100 mg). The reaction mixture was stirred for 3 h at 50 °C,
was cooled to 20 °C over 3 h, and then was stirred for an
additional 3 h. Solids were collected by filtration, washed with
MTBE (100 mL), and then dried under full vacuum at 25−30
°C to provide 30a/b·oxalate (72 g) as a white solid (MP =
≤0.06%). The crude product was utilized directly in the
subsequent step; however, a small portion of the mixture was
removed and concentrated for characterization of pivalate 29a/
b (thick oil). 1H NMR (DMSO-d6, 600 MHz, 25 °C): δ 7.37−
7.13 (m, 20H), 5.24 (s, 1H), 4.76 (d, J = 11.5 Hz, 1H), 4.63 (d,
J = 11.3 Hz, 1H), 4.61 (d, J = 11.3 Hz, 1H), 4.59 (d, J = 11.3
Hz, 1H), 4.54 (d, J = 5.5 Hz, 1H), 4.49 (d, J = 11.5 Hz, 1H),
4.48 (s, 2H), 4.46 (d, J = 11.3 Hz, 1H), 4.25 (m, 2H), 4.10 (d, J
= 11.5 Hz, 1H), 3.85 (d, J = 4.7 Hz, 1H), 3.70 (b, 1H), 3.61 (d,
J = 10.2 Hz, 1H), 3.59 (d, J = 10.2 Hz, 1H), 3.50 (b, 2H), 3.38
(b, 1H), 2.28 (b, 1H), 2.17 (b, 2H), 2.10 (s, 3H), 2.08 (b, 1H),
1.11 (s, 9H). 13C NMR (DMSO-d6, 150.8 MHz, 25 °C): δ
177.0, 167.4, 138.7, 138.4, 137.6, 128.1, 128.0, 127.96, 127.7,
127.6, 127.5, 127.4, 127.3, 127.23, 127.20, 127.1, 82.0, 78.3,
77.4, 75.0, 73.7, 73.6, 72.7, 71.8, 71.6, 65.7, 54.6, 54.2, 45.4,
44.4, 41.6, 38.2, 26.8. HRMS: (ESI+) Calcd for C45H57N2O8 (M
+ H)+: 753.41094, Found: 753.41144.
1
109−112 °C). H NMR (DMSO-d6, 600 MHz, 25 °C): δ
7.37−7.19 (m, 20H), 4.77 (d, J = 11.4 Hz, 0.58H), 4.76 (d, J =
11.4 Hz, 0.84H), 4.73 (d, J = 11.2 Hz, 0.58H), 4.65 (d, J = 11.4
Hz, 0.58H), 4.64 (d, J = 11.0 Hz, 0.42H), 4.63 (d, J = 11.0 Hz,
0.58H), 4.62 (d, J = 6.8 Hz, 0.58H), 4.58 (d, J = 12.3 Hz,
0.58H), 4.54−4.46 (m, 3H), 4.44 (d, J = 12.1 Hz, 0.42H), 4.40
(d, J = 3.8 Hz, 0.42H), 4.39 (d, J = 11.1 Hz, 0.42H), 4.19 (dd, J
= 6.8, 3.1 Hz, 0.58H), 4.14 (dd, J = 5.9, 3.8 Hz, 0.42H), 4.13 (d,
J = 8.8 Hz, 0.42H), 4.03 (b, 0.42H), 3.91 (d, J = 9.3 Hz,
0.58H), 3.87 (b, 0.58H), 3.86 (d, J = 3.1 Hz, 0.58H), 3.82 (d, J
= 9.0 Hz, 0.42H), 3.68 (d, J = 9.3 Hz, 0.58H), 3.67 (b, 2H),
3.56 (d, J = 10.6 Hz, 0.58H), 3.55 (b, 1H), 3.54 (d, J = 10.2 Hz,
0.42H), 3.51 (d, J = 10.2 Hz, 0.42H), 3.48 (d, J = 10.6 Hz,
0.58H), 2.91−2.57 (b, 4H), 2.47 (s, 3H), 1.42 (s, 1.26H), 1.34
(s, 1.26H), 1.29 (s, 1.74H), 1.18 (s, 1.74H). 13C NMR
(DMSO-d6, 150.8 MHz, 25 °C): δ 168.10, 167.42, 163.45,
138.61, 138.41, 138.05, 137.98, 137.93, 137.38, 137.18, 128.21,
128.17, 128.15, 128.13, 128.12, 128.04, 127.98, 127.80, 127.77,
127.71, 127.66, 127.62, 127.50, 127.45, 127.43, 127.36, 127.33,
127.32, 127.21, 127.10, 127.06, 109.52, 108.39, 85.22, 83.99,
82.99, 81.08, 78.49, 78.08, 77.90, 76.38, 74.27, 73.96, 73.63,
72.63, 72.41, 71.81, 71.52, 67.33, 67.21, 52.92, 52.80, 52.69,
52.58, 43.12, 42.84, 42.67, 42.35, 42.29, 27.28, 26.62, 26.03,
26.00. HRMS: (ESI+) Calcd for C43H53N2O7 (M + H)+:
709.38473, Found: 709.38489.
Pivalate Cleavage. A solution of crude pivalate 29a/b (118
g, 157 mmol) in toluene (500 mL total solution) was diluted
with toluene (210 mL), and the water content was measured by
KF titration (target ≤0.03%). The solution was then cooled to
5 °C, and solid sodium methoxide (12 g, 218 mmol) was
added. The resulting mixture was stirred for 4 h at 0−5 °C, and
then UHPLC/MS analysis confirmed complete conversion to
intermediate diol 22a/b (target specification ≤5% 29a/b).
Acetonide and Oxalate Formation. Methanesulfonic acid
(29 mL, 436 mmol) was slowly added at 0−5 °C to the
solution of 22a/b, followed by 2,2-dimethoxypropane (73 mL,
593 mmol). Following complete addition, the temperature was
increased to 20−25 °C, and the reaction was stirred for 30 min
before UHPLC/MS analysis confirmed complete conversion to
acetonide 30a/b (target specification ≤1% 22a/b). The
mixture was then quenched with saturated aqueous sodium
bicarbonate (502 mL), the organic phase was washed with
water (502 mL), and the solution was dried via azeotropic
distillation with toluene at atmospheric pressure to a final
volume of 300 mL (final internal temperature 113 °C, target
specification KF ≤0.1% water). The reaction mixture was
cooled to 50 °C and then diluted with MTBE (502 mL). A
solution of oxalic acid (13.3 g, 148 mmol) in MTBE (118 mL)
was then slowly added, while maintaining the internal
temperature at 50 °C (15 min addition). After stirring for 0.5
h at 50 °C, the solution was seeded with crystalline 30a/b
oxalate (100 mg) and stirred slowly for 3 h at 50 °C. The
resulting slurry was cooled to 20 °C at a rate of 10 °C/h, and
then slowly stirred for an additional 5 h at 20 °C. The solids
were collected by filtration, the cake was washed with MTBE
(100 mL) and then dried to constant weight under vacuum at
25−30 °C to provide 30a/b oxalate (75 g, 60% from 21) as a
white crystalline solid (MP = 111−113 °C). 1H NMR (DMSO-
d6, 500 MHz, 25 °C): δ 7.37−7.19 (m, 20H), 4.77 (d, J = 11.4
Hz, 1H), 4.73 (d, J = 11.2 Hz, 1H), 4.65 (d, J = 11.4 Hz, 1H),
4.63 (d, J = 11.0 Hz, 1H), 4.62 (d, J = 6.7 Hz, 1H), 4.58 (d, J =
12.3 Hz, 1H), 4.53 (d, J = 12.3 Hz, 1H), 4.52 (d, J = 11.4 Hz,
1H), 4.51 (d, J = 11.4 Hz, 1H), 4.19 (dd, J = 6.7, 3.1 Hz, 1H),
3.91 (d, J = 9.2 Hz, 1H), 3.88 (b, 1H), 3.68 (d, J = 9.2 Hz, 1H),
3.67 (b, 2H), 3.56 (d, J = 10.6 Hz, 1H), 3.55 (b, 1H), 3.48 (d, J
= 10.6 Hz, 1H), 2.84−2.64 (b, 4H), 2.48 (s, 3H), 1.29 (s, 3H),
1.18 (s, 3H). 13C NMR (DMSO-d6, 125.8 MHz, 25 °C): δ
167.4, 163.5, 138.6, 138.1, 137.9, 137.4, 128.21, 128.17, 128.12,
128.05, 127.8, 127.7, 127.6, 127.5, 127.1, 127.06, 108.4, 85.2,
81.1, 78.5, 78.1, 74.3, 73.6, 72.6, 71.5, 67.3, 52.9, 52.7, 43.1,
42.6, 27.3, 26.0.
30a:30b (94:6) via Grignard Option B. Grignard. A
solution of iodomethyl pivalate 28 (95 g, 389 mmol) in
anhydrous THF (500 mL) was cooled to −75 °C under a
nitrogen atmosphere. A solution of isopropylmagnesium
chloride (236 mL, 471 mmol) in THF (commercial 2 M
solution) was then added slowly, while maintaining an internal
temperature of ≤−60 °C. After stirring for 1 h at ≤−60 °C,
GC/MS analysis of an aliquot that had been quenched into
MeOH indicated >95% conversion to the corresponding
Grignard reagent (as measured by formation of methyl
pivalate). A solution of 21 (100 g, 157 mmol) in toluene
(500 mL total solution) was then added slowly to the Grignard
solution, while maintaining an internal reaction temperature of
≤60 °C (∼30 min addition time). Following complete addition,
the solution was stirred for 15 min at ≤60 °C, and then
UHPLC analysis indicated <3% 21 remaining. The reaction was
warmed to 0 °C over 1−2 h and then quenched by the slow
addition of acetic acid (36 mL, 37.7 g, 628 mmol) in toluene
(100 mL), while maintaining an internal temperature of ≤15
°C. The mixture was further diluted with toluene (400 mL) and
then warmed to 25 °C where saturated aqueous sodium
bicarbonate (500 mL) was added. The mixture was stirred for
15 min after this addition to afford an aqueous layer with pH 8
and a clear organic phase. The organic layer was washed with
water (500 mL), and then concentrated via atmospheric
distillation to a volume of 500 mL (final internal temperature
observed during the distillation = 113 °C, final KF target
76
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
(2R,3S,4S)-2,3,4-Tris(benzyloxy)-4-((R)-4-((benzyloxy)-
methyl)-2,2-dimethyl-1,3-dioxolan-4-yl)-1-(4-methylpi-
perazin-1-yl)butan-1-one (30a free base). A suspension of
30a/b oxalate prepared via Grignard option B (150.0 g, 188.0
mmol) in MTBE (1.5 L) was charged with 1 N NaOH (405
mL, 405 mmol) and water (405 mL). The resulting mixture
was stirred under a nitrogen atmosphere at 40 °C for 30 min to
provide a brown solution. The phases were separated, the
organic layer was washed with water (500 mL) and then
concentrated at 45 °C/150 Torr to a volume of ∼450 mL.
Cyclohexane (500 mL) was then added, and the mixture was
concentrated at atmospheric pressure to a volume of ∼400 mL
(internal temperature 78 °C). Additional cyclohexane (500
mL) was added, and the distillation was repeated to an end
volume of ∼550 mL (internal temperature ∼83 °C). The
solution was then cooled to ∼40 °C, and seed crystals of 30a
free base (500 mg) were added. The resulting light suspension
was stirred at 40 °C for ∼2 h, and then was cooled to 12 °C
over 4 h and stirred with low agitation (100 rpm) for 18 h. The
resulting thick slurry was filtered, the solids were washed with
cyclohexane (3 × 80 mL) and dried under vacuum (30 °C/40
Torr) to provide 30a free base (115.7g, 87%) as off-white
crystals (MP = 87 °C). The minor diastereomer 30b was not
(d, J = 8.5 Hz, 1H), 6.89 (m, 2H), 5.77 (s, 1H), 4.54 (m, 2H),
4.04 (q, J = 7.0 Hz, 2H), 1.43 (t, J = 7.0 Hz, 3H). 13C NMR
(CDCl3, 100.61 MHz): δ 158.7, 142.1, 137.9, 131.8, 131.6,
131.5, 131.0, 130.9, 128.9, 128.5, 127.9, 127.8, 121.0, 114.4,
78.3, 70.8, 63.4, 14.9. HRMS Calcd for C22H24O2NBrCl (M +
+
NH4 ): 448.06735, Found 448.06796.
4Bromo-1-chloro-2-(ethoxy(4-ethoxyphenyl)methyl)-
benzene (34a). This compound was prepared from isolated
33 according to the general procedure described for the benzyl
ether analogue 34e to afford 34a in 98.8% yield as a white solid
(MP (DSC onset) = 40 °C). 1H NMR (CDCl3, 400.13 MHz):
δ 7.79 (d, J = 2.5 Hz, 1H), 7.31 (m, 2H), 7.29 (d, J = 8.5 Hz,
1H), 7.20 (d, J = 8.5 Hz, 1H), 6.87 (m, 2H), 5.67 (s, 1H), 4.03
(q, J = 6.9 Hz, 2H), 3.54 (m, 2H), 1.42 (t, J = 7 Hz, 3H), 1.28
(t, J = 6.9 Hz, 3H). 13C NMR (CDCl3, 100.61 MHz): δ 158.6,
142.4, 132.0, 131.4, 130.9, 128.7, 121.0, 114.4, 78.9, 64.7, 63.4,
15.3, 14.9. HRMS Calcd for C17H18O2BrCl: 368.01732, Found
368.01747.
4-Bromo-1-chloro-2-((4-ethoxyphenyl)(isopropoxy)-
methyl)benzene (34b). This compound was prepared from
isolated 25 according to the general procedure described for the
benzyl ether analogue 34e to afford 34b in 74.3% yield as a
1
white solid (MP (DSC onset) = 54.2 °C). H NMR (CDCl3,
1
400.13 MHz): δ 7.81 (d, J = 2.3 Hz, 1H), 7.29 (m, 2H), 7.27
(d, J = 8.3 Hz, 1H), 7.18 (d, J = 8.3 Hz, 1H), 6.85 (m, 2H),
5.81 (s, 1H), 4.02 (q, J = 6.8 Hz, 2H), 3.63 (m, 1H), 1.41 (t, J =
6.8 Hz, 3H), 1.24 (d, J = 6.0 Hz, 3H), 1.22 (d, J = 6.0 Hz, 3H).
13C NMR (CDCl3, 100.61 MHz): δ 158.5, 142.9, 132.6, 131.6,
131.3, 131.2, 130.8, 128.6, 128.4, 121.0, 114.5, 114.3, 75.9, 69.5,
63.4, 22.3, 14.8. HRMS Calcd for C18H20O2BrCl: 382.03297,
Found 382.03302.
detected by UHPLC. Purity >99.5% (UHPLC/MS). H NMR
(DMSO-d6, 500 MHz, 25 °C): δ 7.38−7.19 (m, 20H), 4.78 (d,
J = 11.4 Hz, 1H), 4.76 (d, J = 10.6 Hz, 1H), 4.63 (d, J = 11.4
Hz, 1H), 4.62 (d, J = 7.17 Hz, 1H), 4.61 (d, J = 11.5 Hz, 1H),
4.56 (d, J = 12.3 Hz, 1H), 4.53 (d, J = 12.3 Hz, 1H), 4.51 (d, J
= 11.5 Hz, 1H), 4.49 (d, J = 11.4 Hz, 1H), 4.18 (d, J = 7.17 Hz,
2.87 Hz, 1H), 3.92 (d, J = 9.1 Hz, 1H), 3.86 (d, J = 2.87 Hz,
1H), 3.66 (d, J = 9.1 Hz, 1H), 3.64 (b, 1H), 3.55 (d, J = 10.6
Hz, 1H), 3.53−3.44 (b, 2H), 3.47 (d, J = 10.6 Hz, 1H), 3.39 (b,
1H), 2.29−2.04 (b, 4H), 2.08 (s, 3H), 1.29 (s, 3H), 1.17 (s,
3H). 13C NMR (DMSO-d6, 125.8 MHz, 25 °C): δ 167.1, 138.7,
138.2, 137.9, 137.5, 128.17, 128.15, 128.1, 128.0, 127.8, 127.60,
127.57, 127.43, 127.36, 127.04, 127.01, 108.2, 85.4, 81.3, 78.6,
78.1, 74.3, 73.6, 72.7, 71.7, 71.3, 67.2, 54.5, 54.2, 45.4, 44.6,
41.5, 27.4, 26.0. HRMS: (ESI+) Calcd for C43H53N2O7 (M +
H)+: 709.38473, Found: 709.38407.
4-Bromo-2-(butoxy(4-ethoxyphenyl)methyl)-1-chlor-
obenzene (34c). This compound was prepared from isolated
25 according to the general procedure described for the benzyl
ether analogue 34e to afford 34c in 79.0% yield as a white solid
1
(MP (DSC onset) = 35.7 °C). H NMR (CDCl3, 400.13
MHz): δ 7.77 (d, J = 2.3 Hz, 1H), 7.30 (m, 2H), 7.28 (d, J =
8.5 Hz, 1H), 7.20 (d, J = 8.5 Hz, 1H), 6.87 (m, 2H), 5.64 (s,
1H), 4.03 (q, J = 6.9 Hz, 2H), 3.46 (t, J = 6.5 Hz, 2H), 1.64 (m,
2H), 1.45 (m, 2H), 1.42 (t, J = 6.5 Hz, 3H), 0.93 (t, J = 7.4 Hz,
3H). 13C NMR (CDCl3, 100.61 MHz): δ 158.6, 142.5, 132.1,
131.8, 131.4, 130.9, 128.7, 121.0, 114.3, 79.0, 69.0, 63.4, 31.9,
19.4, 14.9, 13.9. HRMS Calcd for C19H22O2BrCl: 396.04862,
Found 396.04880.
2-((Benzyloxy)(4-ethoxyphenyl)methyl)-4-bromo-1-
chlorobenzene (34e). To a solution of 4-ethoxyphenylmag-
nesium bromide (0.5 M in 2-MeTHF, 11.0 L, 5.47 mol) at 0 °C
was added a solution of 5-bromo-2-chlorobenzaldehyde 32 (1.0
kg, 4.56 mol) in 2-MeTHF (4.6 L) over approximately 1 h.
Following complete addition, the reaction was stirred for 1−2 h
at 0 °C, and then UHPLC-MS analysis indicated complete
conversion to intermediate alcohol 33. Sulfuric acid (98%, 170
mL, 3.19 mol) was then added slowly, followed by benzyl
alcohol (990 g, 9.11 mol). The resulting solution was heated to
reflux (76−85 °C) and stirred for 20 h. UHPLC/MS analysis
then indicated complete conversion of alcohol 33 to ether 34e
(target specification ≤3% 33). The reaction was quenched with
water (5 L), the layers were separated, and the organic layer
was washed with water (5 L). The organic layer was distilled to
a minimum volume (<1.5 L) under reduced pressure, and then
methanol (4 L) was added and the solution cooled to −10 °C
over 60 min. The batch was seeded with 1 mol % 34e, and the
resulting slurry was stirred slowly for 10 h, filtered, and washed
with cold methanol (4 L). The cake was dried under vacuum
(55 °C, 21 in Hg) to provide 34e (1.48 kg, 75%) as a white
solid (MP (DSC onset) = 62.3 °C). Potency >96% (UHPLC/
4-Bromo-1-chloro-2-((4-ethoxyphenyl)(phenethoxy)-
methyl)benzene (34f). This compound was prepared from
isolated 25 according to the general procedure described for the
benzyl ether analogue 34e to afford 34f in 80.5% yield as a
1
white solid (MP (DSC onset) = 56.3 °C). H NMR (CDCl3,
400.13 MHz): δ 7.69 (d, J = 2.5 Hz, 1H), 7.33 (m, 2H), 7.28
(m, 1H), 7.23 (m, 5H), 7.20 (m, 1H), 6.85 (m, 2H), 5.68 (s,
1H), 4.03 (q, J = 6.9 Hz, 2H), 3.69 (t, J = 7.0 Hz, 2H), 2.98 (t, J
= 7.2 Hz, 2H), 1.43 (t, J = 7.2 Hz, 3H). 13C NMR (CDCl3,
100.61 MHz): δ 158.6, 142.3, 138.8, 131.9, 131.7, 131.4, 130.9,
130.8, 129.0, 128.7, 128.4, 126.3, 121.0, 114.3, 79.1, 70.1, 63.4,
36.5, 14.9. HRMS Calcd for C23H22O2BrClNa (M + Na+):
467.03839, Found 467.03902.
(2R,3S,4S)-2,3,4-Tris(benzyloxy)-1-(3-((benzyloxy)(4-
ethoxyphenyl)methyl)-4-chlorophenyl)-4-(4-
((benzyloxy)methyl)-2,2-dimethyl-1,3-dioxolan-4-yl)-
butan-1-one (35). Salt Break. Amide 30a/b·oxalate53 (10.09
g) and toluene (100 mL) were combined in a suitable reaction
vessel to afford a white suspension at ambient temperature. A
1
MS). H NMR (CDCl3, 400.13 MHz): δ 7.87 (d, J = 2.5 Hz,
1H), 7.38 (m, 5H), 7.37 (m, 2H), 7.31 (d, J = 8.5 Hz, 1H), 7.21
77
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
solution of saturated aqueous sodium bicarbonate (55 mL) was
then added to the oxalate salt slurry, and the resulting mixture
was allowed to stir for 10 min. The reaction mixture was
transferred into a separatory funnel, and the aqueous phase
(pH 7) was removed. The organic phase was washed with
additional saturated aqueous sodium bicarbonate (20 mL, pH
9) and then was dried over sodium sulfate, filtered, and
concentrated to an orange-brown solution (∼15 mL).
Lithiation. Aryl bromide 34e (6.49 g), toluene (60 mL) and
2methyl tetrahydrofuran (6.5 mL) were charged to a second
reaction vessel to afford a clear, colorless solution. This was
cooled to 20 °C with stirring under a positive pressure of
nitrogen. A solution of n-butyllithium in hexane (2.5 M, 6.5
mL) was then added over 10 min at −15 °C. Following
complete addition, the reaction was allowed to stir for 10−20
min before use in the arylation step.
sulfate, filtered, and concentrated to an oil. The crude oil was
purified by silica gel column chromatography using an ethyl
acetate−hexanes gradient as eluent to afford the major
stereoisomer 37a for characterization purposes (5.56 g).53
Preferably, the crude, partially concentrated solution of 37a/b
1
is used directly without chromatography in the next step. H
NMR (DMSO-d6, 500 MHz, 25 °C): δ 7.43 (d, J = 2.5 Hz,
1H), 7.71−7.14 (m, 21H), 7.03 (d, J = 8.6 Hz, 2H), 6.86 (d, J =
7.3 Hz, 2H), 6.73 (d, J = 8.6 Hz, 2H), 4.57 (s, 2H), 4.56 (d, J =
12.2 Hz, 1H), 4.52 (d, J = 11.4 Hz, 1H), 4.48 (d, J = 12.2 Hz,
1H), 4.41 (d, J = 11.4 Hz, 1H), 4.21 (d, J = 11.8 Hz, 1H), 4.14
(d, J = 7.0 Hz, 1H), 4.02 (d, J = 11.8 Hz, 1H), 3.98 (s, 2H),
3.96 (d, J = 9.4 Hz, 1H), 3.91 (q, J = 7.0 Hz, 2H), 3.86 (s, 1H),
3.71 (d, J = 7.0 Hz, 1H), 3.64 (d, J = 9.4 Hz, 1H), 3.61 (s, 1H),
3.57 (s, 1H), 1.27 (t, J = 7.0 Hz, 3H). 13C NMR (DMSO-d6,
125 MHz, 25 °C): δ 156.8, 138.1, 138.0, 137.92, 137.88,
137.61, 132.8, 131.1, 129.4, 128.8, 128.6, 128.33, 128.28,
128.15, 128.1, 127.9, 127.8, 127.68, 127.66, 127.63, 127.58,
127.51, 127.4, 125.6, 114.2, 106.5, 82.7, 76.8, 75.1, 74.5, 72.9,
71.3, 71.1, 70.9, 68.9, 68.3, 62.8, 37.5, 14.6. HRMS (28a):
(ESI+) Calcd for C50H49Cl1O7Na (M + Na)+: 819.30590,
Found: 819.30676.
Arylation. The solution of free base 30a/b in toluene (∼15
mL) was then added to the aryllithium prepared above over 10
min at −15 °C. After complete addition, UHPLC analysis
confirmed reaction completion. The reaction was therefore
quenched via addition of 1 N hydrochloric acid (50 mL), and
was then warmed to 20 °C. The phases were separated, and the
organic phase was washed with saturated aqueous sodium
chloride (30 mL), dried with sodium sulfate, filtered, and
concentrated to an orange-brown oil. The crude product was
purified by silica gel column chromatography using an ethyl
acetate/hexanes gradient as eluent to afford a mixture of
stereoisomers of 3553 (10.83 g). Alternately, the crude, partially
concentrated solution of 35 may be used directly without
(1S,2S,3S,4R,5S)-5-(4-Chloro-3-(4-ethoxybenzyl)-
phenyl)-1-(hydroxymethyl)-6,8-dioxabicyclo[3.2.1]-
octane-2,3,4-triol (1). A pressure reactor was charged with
5% Pd/C (Johnson Matthey-type A5R87L, 20 wt %, 2.8 g). A
solution of tetra-O-benzyl ether isomers 37a/b (14 g) in
methanol (28 mL) and toluene (560 mL) was then added,
followed by 36.5% aqueous hydrochloric acid (2.39 mL). The
vessel was purged successively with nitrogen (4×) and
hydrogen (4×), and then the slurry was warmed to 25 °C.
The vessel was then pressurized to 50 psi with hydrogen and
stirred for 18 h. The vessel was then purged with nitrogen
(4×). The catalyst was removed via filtration, and the cake was
washed with methanol. UHPLC analysis indicated complete
conversion to 1. The in situ yield, obtained by UHPLC via
comparison to a chromatographed standard was typically 55−
60% from 30a/b. The crude product was typically utilized
directly in the subsequent step; however, a portion of the
mixture was purified by silica gel column chromatography to
afford an analytically pure sample of 1 as a white solid after
1
chromatography in the next step. H NMR (DMSO-d6, 600
MHz, 25 °C): δ 8.47 (d, J = 2.0 Hz, 1H), 8.41 (d, J = 2.0 Hz,
1H), 7.89 (dd, J = 8.3, 2.0 Hz, 1H), 7.88 (dd, J = 8.3, 2.0 Hz,
1H), 7.47 (d, J = 8.3 Hz, 1H), 7.46 (d, J = 8.3 Hz, 1H), 7.31−
7.15 (m, 25H), 7.03 (d, J = 7.6 Hz, 2H), 7.01 (d, J = 7.6 Hz,
2H), 6.80 (d, J = 8.6 Hz, 2H), 6.79 (d, J = 8.6 Hz, 2H), 5.77 (s,
1H), 5.76 (s, 1H), 5.00 (m, 2H), 4.69 (d, J = 11.7 Hz, 1H),
4.68 (d, J = 11.7 Hz, 1H), 4.64−4.53 (m, 4H), 4.49−4.30 (m,
14H), 3.94−3.89 (m, 4H), 3.88−3.85 (m, 4H), 3.71 (d, J =
11.2 Hz, 1H), 3.69 (d, J = 11.2 Hz, 1H), 3.51−3.43 (m, 4H),
1.29 (s, 3H), 1.28 (s, 3H), 1.26 (t, J = 7.0 Hz, 3H), 1.25 (t, J =
7.0 Hz, 3H), 1.11 (s, 3H), 1.10 (s, 3H). 13C NMR (DMSO-d6,
150 MHz, 25 °C): δ 198.18, 197.96, 158.05, 139.91, 139.88,
138.33, 138.30, 137.83, 137.80, 137.76, 137.65, 137.58, 137.46,
137.41, 136.89, 136.76, 134.86, 134.74, 131.23, 131.18, 129.74,
129.69, 128.95, 128.89, 128.83, 128.62, 128.18, 128.16, 128.14,
128.01, 127.95, 127.79, 127.65, 127.60, 127.56, 127.50, 127.47,
127.46, 127.41, 127.28, 127.26, 127.20, 127.17, 127.11, 127.09,
127.06, 114.11, 108.84, 108.74, 84.73, 84.66, 84.37, 84.18,
78.94, 78.76, 78.36, 78.33, 78.10, 77.81, 74.17, 74.02, 73.78,
73.74, 72.58, 72.50, 72.11, 72.03, 71.58, 71.49, 70.01, 69.96,
68.09, 67.94, 62.85, 48.64, 27.10, 27.05, 26.74, 26.03, 14.49,
14.48. HRMS: (ESI+) Calcd for C60H61Cl1O9Na (M + Na)+:
983.38963, Found: 983.39026.
(2S,3S,4R)-2,3,4-Tris(benzyloxy)-1-((benzyloxy)-
methyl)-5-(4-chloro-3-(4-ethoxybenzyl)phenyl)-6,8-
dioxabicyclo[3.2.1]octane (37a/b). A solution of aryl
ketones 3553 (8.06 g) and toluene (40 mL) was treated with
trifluoroacetic acid (3.2 mL) and triethylsilane (10.4 mL) at
ambient temperature. After complete reaction, according to
UHPLC/MS analysis (target <2% 35 isomers remaining),
saturated aqueous sodium bicarbonate was added. The layers
were separated, the organic phase was washed with water and
saturated aqueous sodium chloride and then dried over sodium
1
concentration. H NMR (DMSO-d6, 600 MHz, 25 °C): δ 7.40
(d, J = 2.1 Hz, 1H), 7.39 (d, J = 8.3 Hz, 1H), 7.30 (dd, J = 8.3,
2.1 Hz, 1H), 7.09 (d, J = 8.6 Hz, 2H), 6.83 (d, J = 8.6 Hz, 2H),
5.19 (d, J = 5.5, 1H), 4.97 (d, J = 5.6, 1H), 4.89 (d, J = 6.7, 1H),
4.75 (t, J = 6.0, 1H), 3.99 (s, 2H), 3.98 (d, J = 7.0 Hz, 1H), 3.97
(q, J = 6.9 Hz, 2H), 3.63 (dd, J = 12.4, 6.0 Hz, 1H), 3.54 (dd, J
= 7.9, 5.5 Hz, 1H), 3.49 (dd, J = 12.4, 6.0 Hz, 1H), 3.46 (d, J =
7.0 Hz, 1H), 3.43 (td, J = 7.9, 5.6 Hz, 1H), 3.40 (dd, J = 7.8, 6.8
Hz, 1H), 1.30 (t, J = 6.9, 3H). 13C NMR (DMSO-d6, 150 MHz,
25 °C): δ 156.8, 138.0, 137.6, 132.5, 131.0, 129.5, 129.2, 128.3,
126.1, 114.2, 107.6, 84.9, 77.3, 76.0, 71.4, 66.1, 62.8, 59.8, 37.5,
14.6. HRMS: (ESI−) Calcd for C22H24Cl1O7 (M − H)−:
435.12160, Found: 435.11993.
((1R,2S,3S,4R,5S)-5-(4-Chloro-3-(4-ethoxybenzyl)-
phenyl)-2,3,4-trihydroxy-6,8-dioxabicyclo[3.2.1]octan-1-
yl)methyl Acetate (38). A solution of 1 (26.4 g) in toluene
(270 mL) and pyridine (6.6 mL) was charged to a suitable
reaction vessel. The solution was cooled to −10 °C, and then
acetic anhydride (5.8 mL) was added over 5 min. The reaction
was stirred for 60 min at −10 °C, and then was warmed slowly
to 20 °C and stirred for 18 h to provide a slurry. Toluene (100
mL) was then added, followed by water (200 mL). After
78
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
stirring for 15 min, the water layer was removed and the solids
were collected by filtration. The cake was washed with
additional water and toluene, then solids were dried under
vacuum at room temperature to provide monoacetate 38 (19.2
g, 79.7%) as a white solid (MP = 160 °C). UHPLC/MS: 97.6%
stirring solution of NaOMe (0.5 M in MeOH, 0.65 mL) in
MeOH (40 mL) at 20 °C was added tetra-O-acetate 39 (20.0
g) in two solid portions separated by 15 min. The additions
provided a slurry that gave way to a solution after a few min.
After stirring for ∼3 h, UHPLC/MS analysis of an aliquot
indicated <2% combined monoacetate intermediates, and so
the crude solution was carried directly into the subsequent
cocrystallization step. For the purpose of characterization, a
small portion of the mixture was concentrated to dryness on a
rotovap and then further dried under high vacuum, to provide
1
product, 2.0% C-4 acetate isomer, 0.4% 1. H NMR (DMSO-
d6, 500 MHz, 25 °C): δ 7.400 (d, J = 8.3 Hz, 1H), 7.398 (d, J =
2.1 Hz, 1H), 7.30 (dd, J = 8.3, 2.1 Hz, 1H), 7.09 (d, J = 8.7 Hz,
2H), 6.83 (d, J = 8.7 Hz, 2H), 5.50 (d, J = 5.7 Hz, 1H), 5.10 (d,
J = 5.6 Hz, 1H), 5.01 (d, J = 6.6 Hz, 1H), 4.27 (d, J = 12.4 Hz,
1H), 4.07 (d, J = 12.4 Hz, 1H), 4.06 (d, J = 7.4 Hz, 1H), 3.99
(s, 2H), 3.97 (q, J = 7.0 Hz, 2H), 3.58 (t, J = 6.5 Hz, 1H), 3.50
(dd, J = 7.4, 0.9 Hz, 1H), 3.44 (m, 1H), 3.42 (m, 1H), 2.02 (s,
3H), 1.29 (t, J = 7 Hz, 3H). 13C NMR (DMSO-d6, 125 MHz,
25 °C): δ 170.1, 156.9, 137.8, 137.6, 132.7, 131.1, 129.6, 129.2,
128.5, 126.1, 114.3, 108.2, 82.8, 77.2, 75.7, 71.2, 66.4, 62.8,
62.0, 37.6, 20.6, 14.7. HRMS: (ESI+) Calcd for C24H28Cl1O8
(M + H)+: 479.14672, Found: 479.14630.
1
amorphous ertugliflozin (1) as a white solid mass. H NMR
(DMSO-d6, 600 MHz, 25 °C): δ 7.40 (d, J = 2.1 Hz, 1H), 7.39
(d, J = 8.3 Hz, 1H), 7.30 (dd, J = 8.3, 2.1 Hz, 1H), 7.09 (d, J =
8.6 Hz, 2H), 6.83 (d, J = 8.6 Hz, 2H), 5.19 (d, J = 5.5, 1H),
4.97 (d, J = 5.6, 1H), 4.89 (d, J = 6.7, 1H), 4.75 (t, J = 6.0, 1H),
3.99 (s, 2H), 3.98 (d, J = 7.0 Hz, 1H), 3.97 (q, J = 6.9 Hz, 2H),
3.63 (dd, J = 12.4, 6.0 Hz, 1H), 3.54 (dd, J = 7.9, 5.5 Hz, 1H),
3.49 (dd, J = 12.4, 6.0 Hz, 1H), 3.46 (d, J = 7.0 Hz, 1H), 3.43
(td, J = 7.9, 5.6 Hz, 1H), 3.40 (dd, J = 7.8, 6.8 Hz, 1H), 1.30 (t,
J = 6.9, 3H). 13C NMR (DMSO-d6, 150 MHz, 25 °C): δ 156.8,
138.0, 137.6, 132.5, 131.0, 129.5, 129.2, 128.3, 126.1, 114.2,
107.6, 84.9, 77.3, 76.0, 71.4, 66.1, 62.8, 59.8, 37.5, 14.6. HRMS:
(ESI−) Calcd for C22H24Cl1O7 (M − H)−: 435.12160, Found:
435.11993.
Cocrystal Formation. A solution of 1 (4.80 g) in MeOH
(17.3 mL) was concentrated under vacuum (30 Torr) at 40 °C
to minimal volume. The residue was taken up in i-PrOH (18.7
mL) and the resulting solution was passed through a speck-free
filter into a suitable reaction vessel. The solution was heated to
60 °C where water (18.7 mL) was added. In a second reaction
vessel, L-pyroglutamic acid (3.91 g) was dissolved in water
(56.2 mL). This solution was passed through a polishing filter
into the reactor containing 1. The resulting mixture was heated
to 80 °C, and then was cooled to 40 °C at 3 °C/minute and
seeded. The mixture was granulated for 10 h at 40 °C, and then
cooled to 20 °C at 0.1 °C/min. The resulting white slurry was
isolated on a Coors filter, washed with toluene (2 × 9.8 mL),
and was dried under vacuum at 55 °C for 4 h to provide 1·L-
PGA (5.26 g, 85.3%) as a white crystalline solid (MP (DSC
onset) = 142.5 °C). 1H NMR (DMSO-d6, 600 MHz, 25 °C): δ
7.90 (bs, 1H), 7.40 (d, J = 2.1 Hz, 1H), 7.39 (d, J = 8.3 Hz,
1H), 7.30 (dd, J = 8.3, 2.1 Hz, 1H), 7.09 (d, J = 8.6 Hz, 2H),
6.83 (d, J = 8.6 Hz, 2H), 5.19 (bs, 1H), 4.97 (bs, 1H), 4.90 (bs,
1H), 4.75 (bs, 1H), 4.06 (ddd, J = 9.0, 4.3, 0.7 Hz, 1H), 3.99 (s,
2H), 3.98 (d, J = 7.2 Hz, 1H), 3.97 (q, J = 7.0 Hz, 2H), 3.63
(dd, J = 12.5, 4.5 Hz, 1H), 3.54 (d, J = 7.8 Hz, 1H), 3.49 (dd, J
= 12.5, 4.5 Hz, 1H), 3.46 (d, J = 7.2 Hz, 1H), 3.43 (t, J = 7.9
Hz, 1H), 3.40 (dd, J = 7.9, 5.5 Hz, 1H), 2.32 (m, 1H), 2.12 (m,
2H), 1.96 (m, 1H), 1.30 (t, J = 7.0, 3H). 13C NMR (DMSO-d6,
150 MHz, 25 °C): δ 176.9, 174.3, 156.8, 138.0, 137.6, 132.5,
131.0, 129.5, 129.2, 128.3, 126.1, 114.2, 107.6, 84.9, 77.3, 76.0,
71.4, 66.1, 62.8, 59.8, 54.6, 37.5, 28.9, 24.5, 14.6. HRMS:
(ESI−) Calcd for C27H31Cl1N1O10 (M − H)−: 564.16420,
Found: 564.16315.
(1R,2S,3S,4R,5S)-1-(Acetoxymethyl)-5-(4-chloro-3-(4-
ethoxybenzyl)phenyl)-6,8-dioxabicyclo[3.2.1]octane-
2,3,4-triyl Triacetate (39). A solution of 1 (83 g), toluene
(830 mL) and pyridine (122 mL) was cooled to 5 °C with
stirring. Acetic anhydride (108 mL) was then added over 15
min. Following complete addition, the reaction was stirred at 5
°C for 1 h then at 20 °C for 23 h. The reaction mixture was
then cooled to 5 °C where water (800 mL) was added over 10
min. The resulting mixture was stirred at 5 °C for 15 min then
was warmed to 20 °C before the phases were separated. The
organic phase was then washed with water (500 mL), dried
over magnesium sulfate, filtered, and concentrated at 40 °C/50
Torr to an oil. Isopropanol (1.2 L) was added, and the resulting
solid−liquid mixture was heated to 60 °C to provide a solution.
This solution was cooled to 45 °C over 1 h and then seeded
with tetraacetate 39 (50 mg). This resulted in formation of a
thick slurry that was thinned by the addition of isopropanol
(100 mL). The solids were collected after 1 h at 20 °C, the cake
was washed with isopropanol (2 × 100 mL) and dried under
vacuum to afford tetraacetate 39 (102.0 g. 88.7%) as a white
crystalline solid (MP (DSC onset) = 135 °C). UHPLC/MS
purity: 99.6%, 0.4% triacetate. The crystallization liquors were
concentrated to 300 mL at 40 °C/40 Torr. The resulting slurry
was stirred at 20 °C for 1 h, and then solids were collected,
washed with isopropanol, and dried under vacuum at 20 °C.
This provided a second crop of tetraacetate 39 (6.2 g, 5.4%) of
slightly lower purity (UHPLC/MS purity: 93.7%, 6.3%
1
triacetate). H NMR (DMSO-d6, 500 MHz, 25 °C): δ 7.46
(d, J = 8.3 Hz, 1H), 7.39 (d, J = 2.1 Hz, 1H), 7.33 (dd, J = 8.3,
2.1 Hz, 1H), 7.06 (d, J = 8.6 Hz, 2H), 6.83 (d, J = 8.7 Hz, 2H),
5.35 (dd, J = 8.4, 0.8 Hz, 1H), 5.25 (t, J = 8.4 Hz, 1H), 5.19 (d,
J = 8.4 Hz, 1H), 4.47 (d, J = 12.8 Hz, 1H), 4.34 (d, J = 8.4 Hz,
1H), 4.04 (d, J = 12.8 Hz, 1H), 4.03 (d, J = 15.2 Hz, 1H), 3.97
(d, J = 15.2 Hz, 1H), 3.97 (q, J = 7.0 Hz, 2H), 3.79 (dd, J = 8.4,
0.8 Hz, 1H), 2.02 (s, 3H), 2.00 (s, 3H), 1.92 (s, 3H), 1.66 (s,
3H), 1.29 (t, J = 7 Hz, 3H). 13C NMR (DMSO-d6, 125 MHz,
25 °C): δ 169.71, 169.70, 169.1, 168.6, 157.0, 138.6, 134.4,
133.9, 130.9, 129.5, 129.2, 128.7, 125.4, 114.3, 106.9, 82.6, 74.8,
72.2, 68.7, 67.9, 62.9, 60.4, 37.4, 20.39, 20.36, 20.32, 19.9, 14.6.
HRMS: (ESI+) Calcd for C30H33Cl1O11Na (M + Na)+:
627.16036, Found: 627.16211.
ASSOCIATED CONTENT
■
S
* Supporting Information
1
Copies of H and 13C NMR spectra, powder X-ray diffraction
(1S,2S,3S,4R,5S)-5-(4-Chloro-3-(4-ethoxybenzyl)-
phenyl)-1-(hydroxymethyl)-6,8-dioxabicyclo[3.2.1]-
octane-2,3,4-triol, (1:1) Compound with (S)-5-Oxopyrro-
lidine-2-carboxylic Acid (1·L-PGA). Deacetylation. To a
(PXRD) patterns for compounds 38 and 39, and single-crystal
X-ray structure determination of compound 1·L-PGA. This
material is available free of charge via the Internet at http://
pubs.acs.org.
79
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
Hagita, H.; Ozawa, K.; Yamaguchi, K.; Kato, M.; Ikeda, S. J. Med.
Chem. 2012, 55, 7828.
(14) Murphy, P. V.; McDonnell, C.; Hamig, L.; Paterson, D. E.;
Taylor, R. J. K. Tetrahedron Asymmetry 2003, 14, 79.
(15) Singh, J.; DiMarco, J.; Kissick, T. P.; Deshpande, P.; Gougoutas,
J. Z. Carbohydr. Res. 2002, 337, 565.
AUTHOR INFORMATION
Corresponding Author
*E-mail: jade.nelson@pfizer.com
■
Present Address
§Nalas Engineering Services, Inc., 85 Westbrook Road,
Centerbrook, CT 06409.
(16) Farr, R. N.; Outten, R. A.; Cheng, J. C. Y.; Daves, G. D., Jr.
Organometallics 1990, 9, 3151.
Notes
(17) Friesen, R. W.; Loo, R. W. J. Org. Chem. 1991, 56, 4821.
(18) Mazur, A. W.; Hiler, G. D., II. J. Org. Chem. 1997, 62, 4471.
(19) Yang, Y.-Y.; Yang, W.-B.; Teo, C.-F.; Lin, C.-H. Synlett 2000,
1634.
(20) Schmidt, R. R.; Frick, W. Tetrahedron 1988, 44, 7163.
(21) Modeling studies carried out on diol 7b illustrated that the two
primary alcohols were not in close proximity in the minimized
structure.
(22) Allyl protecting groups, as well as a range of substituted benzyl
derivatives, were also demonstrated to be feasible.
(23) Rajanikanth, B.; Seshadri, R. Tetrahedron Lett. 1989, 30, 755.
(24) Kang, S. Y.; Song, K.-S.; Lee, J.; Lee, S.-H.; Lee, J. Bioorg. Med.
Chem. 2010, 18, 6069.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors gratefully acknowledge Bao Nguyen and David
Foley for NMR analysis, Andy Palm for process analytical
technology (PAT) support, Kapildev Arora for single-crystal X-
ray structure determination of anhydrous 1·L-PGA, and Yuriy
Abramov for computational modeling. We also thank current
and past members of the ertugliflozin API Technical Team, the
Pfizer Structural Elucidation Group, the Pfizer Chemical
Research & Development and Analytical Research & Develop-
ment Management Teams, and the global Pfizer brainstorming
session participants for valuable contributions to the program.
(25) Martin, R.; Romea, P.; Tey, C.; Urpi, F.; Vilarrasa, J. Synlett
1997, 1414.
(26) The utility of morpholine amides as electrophiles has been
documented; however, the use of methylpiperazine amides as
precursors to aryl ketones appears to be unprecedented.
(27) Kaku, T.; Tsujimoto, S.; Matsunaga, N.; Tanaka, T.; Hara, T.;
Yamaoka, M.; Kusaka, M.; Tasaka, A. Bioorg. Med. Chem. 2011, 19,
2428.
(28) Oxidation of primary alcohol 18 to aldehyde 19 was only briefly
studied. Dess−Martin periodinane was found to be moderately
effective, but was not optimized.
REFERENCES
■
(1) Kalgutkar, A. S.; Tugnait, M.; Zhu, T.; Kimoto, E.; Miao, Z.;
Mascitti, V.; Yang, X.; Tan, B.; Walsky, R. L.; Chupka, J.; Feng, B.;
Robinson, R. P. Drug Metab. Dispos. 2011, 39, 1609.
(2) Maurer, T. S.; Ghosh, A.; Haddish-Berhane, N.; Sawant-Basak,
A.; Boustany-Kari, C. M.; She, L.; Leininger, M. T.; Zhu, T.; Tugnait,
M.; Yang, X.; Kimoto, E.; Mascitti, V.; Robinson, R. P. AAPS J. 2011,
13, 576.
(3) Miao, Z.; Nucci, G.; Amin, N.; Sharma, R.; Mascitti, V.; Tugnait,
M.; Vaz, A. D.; Callegari, E.; Kalgutkar, A. S. Drug Metab. Dispos. 2013,
41, 445.
(4) Mascitti, V.; Collman, B. M., Preparation of dioxa-bicyclo[3.2.1.]-
octane-2,3,4-triol derivatives as antidiabetic agents. Application: WO/
2009/IB53626, 2010/023594, 2010.
(5) Mascitti, V.; Maurer, T. S.; Robinson, R. P.; Bian, J.; Boustany-
Kari, C. M.; Brandt, T.; Collman, B. M.; Kalgutkar, A. S.; Klenotic, M.
K.; Leininger, M. T.; Lowe, A.; Maguire, R. J.; Masterson, V. M.; Miao,
Z.; Mukaiyama, E.; Patel, J. D.; Pettersen, J. C.; Preville, C.; Samas, B.;
She, L.; Sobol, Z.; Steppan, C. M.; Stevens, B. D.; Thuma, B. A.;
Tugnait, M.; Zeng, D.; Zhu, T. J. Med. Chem. 2011, 54, 2952.
(6) Mascitti, V.; Thuma, B. A.; Smith, A. C.; Robinson, R. P.; Brandt,
T.; Kalgutkar, A. S.; Maurer, T. S.; Samas, B.; Sharma, R.
MedChemComm 2013, 4, 101.
(29) Parikh, J. R.; Doering, W. v. E. J. Am. Chem. Soc. 1967, 89, 5505.
(30) Corey, E. J.; Chaykovsky, M. J. Am. Chem. Soc. 1965, 87, 1353.
(31) A range of Lewis and Brønsted acids were screened as catalysts
in attempts to open epoxide 24 with water or benzyl alcohol.
Debenzylation, elimination, and other destructive reactivity was
observed, but desired products were not detected.
(32) Corey, E. J.; Eckrich, T. M. Tetrahedron Lett. 1983, 24, 3163.
(33) Corey, E. J.; Eckrich, T. M. Tetrahedron Lett. 1983, 24, 3165.
(34) Hara, R.; Furukawa, T.; Horiguchi, Y.; Kuwajima, I. J. Am. Chem.
Soc. 1996, 118, 9186.
(35) Avolio, S.; Malan, C.; Marek, I.; Knochel, P. Synlett 1999, 1820.
(36) Tamao, K.; Kakui, T.; Kumada, M. J. Am. Chem. Soc. 1978, 100,
2268.
(37) Tamao, K.; Ishida, N.; Ito, Y.; Kumada, M. Org. Synth. 1990, 69,
96.
(38) Donohoe, T. J.; Winship, P. C. M.; Pilgrim, B. S.; Walter, D. S.;
Callens, C. K. A. Chem. Commun. (Cambridge, U. K.) 2010, 46, 7310.
(39) Tamao, K.; Ishida, N.; Ito, Y.; Kumada, M. Org. Synth. 1990, 69,
96.
(40) Fleming, I.; Barbero, A.; Walter, D. Chem. Rev. 1997, 97, 2063.
(41) Oxidative conversion of 27 to diols 22a/b can also be achieved
by reaction with sodium perborate. TBAF has been found to accelerate
the rate of the reaction but is not required. The practical utility of this
alternative for large-scale application is being investigated, and details
will be reported separately.
(42) Benzhydryl amines and thioethers were also prepared, but ethers
were selected for additional study due to their relative simplicity, low
cost, and excellent compatibility with downstream processing.
(43) Several procedures were identified for conversion to 30a/b (free
base) in the laboratory, including treatment of 30a/b·oxalate with
solid Na3PO4 in toluene, or neutralization with aqueous solutions of
NaOH or NaHCO3.
(7) Mascitti, V.; Preville, C. Org. Lett. 2010, 12, 2940.
(8) See the preceding article in this journal (Brandt, et al.) for a
detailed report on the second-generation synthesis to ertugliflozin.
(9) Deshpande, P. P.; Singh, J.; Pullockaran, A.; Kissick, T.;
Ellsworth, B. A.; Gougoutas, J. Z.; Dimarco, J.; Fakes, M.; Reyes,
M.; Lai, C.; Lobinger, H.; Denzel, T.; Ermann, P.; Crispino, G.;
Randazzo, M.; Gao, Z.; Randazzo, R.; Lindrud, M.; Rosso, V.; Buono,
F.; Doubleday, W. W.; Leung, S.; Richberg, P.; Hughes, D.; Washburn,
W. N.; Meng, W.; Volk, K. J.; Mueller, R. H. Org. Process Res. Dev.
2012, 16, 577.
(10) Kim, M. J.; Lee, J.; Kang, S. Y.; Lee, S.-H.; Son, E.-J.; Jung, M.
E.; Lee, S. H.; Song, K.-S.; Lee, M.-W.; Han, H.-K.; Kim, J.; Lee, J.
Bioorg. Med. Chem. Lett. 2010, 20, 3420.
(11) Lee, J.; Lee, S.-H.; Seo, H. J.; Son, E.-J.; Lee, S. H.; Jung, M. E.;
Lee, M.; Han, H.-K.; Kim, J.; Kang, J.; Lee, J. Bioorg. Med. Chem. 2010,
18, 2178.
(12) Nomura, S.; Sakamaki, S.; Hongu, M.; Kawanishi, E.; Koga, Y.;
Sakamoto, T.; Yamamoto, Y.; Ueta, K.; Kimata, H.; Nakayama, K.;
Tsuda-Tsukimoto, M. J. Med. Chem. 2010, 53, 6355.
(44) The corresponding Grignard and turbo-Grignard reagents were
also prepared and evaluated.
(45) Krasovskiy, A.; Knochel, P. Angew. Chem., Int. Ed. 2004, 43,
3333.
(13) Ohtake, Y.; Sato, T.; Kobayashi, T.; Nishimoto, M.; Taka, N.;
Takano, K.; Yamamoto, K.; Ohmori, M.; Yamaguchi, M.; Takami, K.;
Yeu, S.-Y.; Ahn, K.-H.; Matsuoka, H.; Morikawa, K.; Suzuki, M.;
80
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81

Organic Process Research & Development
Article
(46) Resubmission of isolated aryl ketone 35 to the reaction
conditions resulted in rapid formation of the expected tertiary alcohol,
providing further evidence of a stabilized tetrahedral intermediate in
the reaction or aryl anions derived from 34e with amide 30a/b.
(47) A diastereomeric mixture of ∼3:2 was obtained from the
Grignard option A, while Grignard option B typically resulted in a 95:5
mixture of diastereomers at C5.
(48) Process efficiency, in this context, refers to number of unit
operations, total reactor volume, total processing time, and isolated
yield.
(49) Purification of this intermediate by column chromatography is
an option for laboratory scale, but this is far less desirable for large-
scale manufacture.
(50) Mascitti, V. Dioxa-bicyclo[3.2.1]octane-2,3,4-triol derivatives.
Application: WO/2010/IB54775, 2011/051864, 2011.
(51) A significant quantity of the methylpiperazine-N-oxide derivative
of 22a/b is observed via UHPLC/MS analysis at this stage of the
reaction in the absence of a sodium bisulfite treatment. This byproduct
coverts quantitatively to 22a/b during the workup.
(52) The isopropoxy group of 27 readily exchanges with water and
MeOH to form silanol and methyl ether analogues, respectively.
(53) Grignard approach B (Scheme 4) was utilized for the
preparation of 22, so the diastereomeric ratio at C5 was approximately
95:5 favoring the (5R) isomer. The (5S) isomer performs in an
equivalent manner throughout the process. Both isomers converge to a
single product 1 during treatment with acid during/after the
hydrogenolysis step.
81
dx.doi.org/10.1021/op4002802 | Org. Process Res. Dev. 2014, 18, 66−81