Process development of a novel azetidinyl ketolide antibiotic

Article abstract of DOI:10.1021/op300064b

Process development and the multikilogram synthesis of a novel azetidinyl ketolide antibiotic is described. Starting with clarithromycin, the eight-step synthesis features several telescoped operations and direct isolations, which results in a significant improvement in throughput and a major reduction in solvent usage and waste stream volume over the first scale-up campaign. Particular highlights of this effort include the development of an efficient synthesis of 3-hydroxy-1,5-naphthyridine-4-carbaldehyde via a Skraup process and engineering a robust final API synthesis. We also discovered a crystalline monotosylate salt that addressed significant formulation and degradation issues experienced when using the noncrystalline freebase.

Full text of DOI:10.1021/op300064b

Article
pubs.acs.org/OPRD
Process Development of a Novel Azetidinyl Ketolide Antibiotic
Bryan Li,*,† Thomas V. Magee,† Richard A. Buzon,† Daniel W. Widlicka,† Dave R. Bill,† Thomas Brandt,†
Xiaoping Cao,† Michael Coutant,† Haijian Dou,‡ Karl Granskog,† Mark E. Flanagan,†
Cheryl M. Hayward,† Bin Li,‡ Fengwei Liu,‡ Wei Liu,§ Thuy-Trinh Nguyen,† Jeffrey W. Raggon,†
Peter Rose,† Joseph Rainville,† Usa Datta Reilly,† Yue Shen,† Jianmin Sun,† and Glenn E. Wilcox†
†Groton Laboratories, Pfizer Global Research and Development, Groton, Connecticut 06340, United States
‡Shanghai ChemPartner Co. Ltd., 720 Cailun Road, Zhangjiang Hi-Tech Park, Shanghai, China
§Asymchem Life Science Co. Ltd., No. 71, 7th Street, TEDA, Tianjin, China
ABSTRACT: Process development and the multikilogram synthesis of a novel azetidinyl ketolide antibiotic is described. Starting
with clarithromycin, the eight-step synthesis features several telescoped operations and direct isolations, which results in a
significant improvement in throughput and a major reduction in solvent usage and waste stream volume over the first scale-up
campaign. Particular highlights of this effort include the development of an efficient synthesis of 3-hydroxy-1,5-naphthyridine-4-
carbaldehyde via a Skraup process and engineering a robust final API synthesis. We also discovered a crystalline monotosylate salt
that addressed significant formulation and degradation issues experienced when using the noncrystalline freebase.
acquired respiratory tract infections.2 The emergence of
resistant bacterial strains in public health settings has fueled
Macrolides are an important class of antibacterial agents.1
the need for novel antibacterial agents.3 Telithromycin (3a)4
INTRODUCTION
■
Clarithromycin (1b, Figure 1) and azithromycin (2) are among
represents a newer generation of ketolides that targets
macrolide-sensitive and -resistant strains. Cethromycin (3b) is
currently under late-stage development as an oral antibiotic for
the treatment of life threatening infections caused by
community-acquired bacterial pneumonia (CABP) and bio-
defense pathogens.5 Efforts in our antibacterial research
program led to the discovery of azetidinyl ketolide 4 (Figure
1) as a promising drug candidate for the treatment of CABP,
acute exacerbations of chronic bronchitis (AECB), sinusitis, and
pharyngitis.6
the most prescribed medicines for the treatment of community-
DISCUSSIONS AND RESULTS
■
1. Initial Route Optimization for the Synthesis of
Ketolide (4). In the original synthesis6 of macrolide 4 (Scheme
1), the azetidine (linker) and the naphthyridine (head piece)
are installed sequentially in an eight-step linear sequence from
clarithromycin (1b). In the first step, bis-acetylation of the
hydroxyl groups on the desosamine and cladinose sugars of 1b
gave diacetate 5.7 Treatment with carbonyl diimidazole (CDI)
in the presence of 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU)
gave the C-11/C-12 cyclic carbonate 6 which subsequently
eliminated to generate the intermediate C10−C11 enone,
which reacted with another equivalent of CDI to afford acyl
imidazole 7.7 Incorporation of the aminoazetidine side chain
(8) followed by cyclization was promoted by DBU in
acetonitrile and provided oxazolidinone 9. Subsequent removal
of the 4″-acetylcladinose using aqueous HCl and ethanol
provided the C-3 hydroxyl intermediate 10, which was oxidized
to ketone 11 using Corey−Kim oxidation conditions.8 The
penultimate intermediate 13 was generated by removal of the
Received: March 12, 2012
Figure 1. Structures of antibiotic macrolides.
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
a
was below 90% purity (by HPLC) despite repetitive chromato-
graphic purifications. (3) The final form of the active
pharmaceutical ingredient (API) was an amorphous solid. (4)
The original syntheses of both the azetidine (linker) and the
naphthyridine (head piece) were not deemed scalable (vide
infra).
Scheme 1. Synthesis of azetidinyl ketolide (4)
In an attempt to obtain higher throughput and process
efficiency, a more convergent synthesis was initially explored
(Scheme 2). In this approach, the fully assembled azetidyl-
Scheme 2. Attempted convergent synthesis of azetidinyl
a
ketolide 4
a
Reagents and conditions: a) Ac2O, TEA, DMAP, DCM; b) 2N HCl,
a
Reagents and conditions used in 10 kg campaign: a) Ac2O, TEA,
DMAP, THF; b) DBU, CDI, THF; 98% overall from 1b. c) DBU, 1-
benzhydryl-azetidine-3-ylamine, MeCN, 50 °C. d) pH 2 (aq HCl),
EtOH, 40 °C; 76% overall from 7. e) DMS, NCS, CH2Cl2. f) MeOH,
50 °C. g) Pearlman’s catalyst, H2 (50 psi), MeOH, conc. aq HCl, 35
°C; 84% overall from 10. h) aldehyde 14, pivalic acid, TEA,
Na(OAc)3BH (STAB); reversed phase chromatography; TsOH,
acetone/EtOAc, 44%.
EtOH, 40 °C; c) DBU, CDI, THF, isopropyl ether; d) DMSO,
pyridinium trifluoroacetate, EDC; e) DBU, CDI, MeCN; f) DIPEA,
MeCN, RT; g) DBU, 80 °C, MeCN or toluene, or KOtBu, THF, RT
to 65 °C.
napthyridine moiety was installed after the hydrolytic cleavage
of cladinose and oxidation of the resulting C-3 alcohol,
therefore shortening the linear sequence. Thus, intermediate
15 was obtained by acetylation of clarithromycin (1b) followed
by hydrolysis of the cladinose sugar. Treatment of 15 with CDI
gave the cyclic carbonate 16. Oxidation of the C-3 hydroxyl
group using Moffatt conditions followed by β-elimination of the
carbonate with 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) and
concurrent reaction with CDI afforded C-12 acylimidazoyl
ketolide 17.9 While reaction of 17 and the azetidinyl
naphthyridine 18 occurred readily to give the carbamate
(19), the subsequent intramolecular Michael addition to the
2′-acetyl group on the desosamine sugar by heating in methanol
followed by hydrogenolysis of the benzhydryl protecting group
using Pearlman’s catalyst. A final reductive amination was
utilized to couple 13 with the 1,5-naphthyridine aldehyde 14
and afforded the desired product 4 after a tedious workup and
purification. Several issues were present in this synthesis: (1)
The overall throughput was less than 5%, most notably the final
coupling step gave ∼20% yield. (2) The final product isolated
B
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
C10−C11 enone required forcing conditions and generated a
number of impurities. Consequently, the desired oxazolidinone
product 20 was isolated in only low yields (<20%) by
chromatography. Poor reactivity in this step was attributed to
steric hindrance due to the bulky 1,5-napthyridyl group. We
were unable to solve this issue and therefore had to abandon
this route.
Table 1. Comparison of first and second campaigns for steps
1 and 2
2. Optimization of the Process for Pilot-Plant
Manufacturing. In order to generate sufficient quantities of
4 to fund toxicology studies, the initial synthetic route was
quickly enabled for a first scale-up campaign. The primary focus
was to ensure the scalability of the chemistry to meet an initial
delivery target of approximately 2 kg. Subsequent process
development was carried out to facilitate a 10 kg API campaign
to support clinical trials. With the larger-scale campaign moving
into a pilot plant, it became necessary to develop a streamlined
and robust process. An additional challenge was that a stable
crystalline final API form had yet to be discovered and was
needed to support formulation development.
In the following sections, we describe how these objectives
were achieved, with concomitant reduction in solvent usage and
process time, and substantially greater throughput.
2.1. Preparation of Acyl Imidazole 7. In the initial scale-up,
the preparation of acyl imidazole 7 followed literature
conditions8 in two discrete steps (steps 1 and 2, Scheme 1).
The acetylation was accomplished with the use of Ac2O,
DMAP, and TEA in CH2Cl2. Formation of the acyl imidazole 7
was driven by employing excess CDI/DBU in THF. The
product isolation of both steps used typical extractive workup
procedures, followed by crystallization. A second crop recovery
was necessary for the isolation of 7.
base) offered a much cleaner reaction profile than using 2 N
HCl in EtOH. The product (10) was readily isolated by
adjusting the pH to 7.013 with triethylamine followed by
filtration.
Table 2. Comparison of first and second scale-up runs for
steps 3 and 4
By examining the unit operations, we realized that the
process time, solvent consumption, and waste streams were
mostly from the reaction workup/isolation. The process
efficiency could be greatly improved by telescoping the steps.
We found THF worked equally well as the reaction solvent for
step 1 and thus eliminated the need to exchange the solvent for
step 2. Although an aqueous workup for the acetylation step
was still needed, the organic phase could be effectively dried by
azeotropic distillation to remove a partial amount of the solvent
and then be carried to step 2 directly. For the product isolation,
a method was developed to precipitate 7 directly by the
addition of aqueous NH4Cl solution to the reaction mixture.
The stability of 7 in aqueous THF was studied and showed no
sign of decomposition at 60 °C after 4 h. This allowed the
isolation of 7 in 98% yield, over two steps, in one crop.
2.2. Azetidine Linker Incorporation and Cladinose
Hydrolysis. Reaction screening was performed for the
incorporation of the azetidine linker and concomitant intra-
molecular Michael addition (step 3, Scheme 1). The optimal
condition was to employ DBU in MeCN at 50 °C. In the first
scale-up, the product (9) was isolated in two crops, and the
second crop isolation required an aqueous/organic extractive
manipulation.11 For the hydrolysis (2 N HCl) of the cladinose
sugar, the reaction was worked up by inverse addition of the
reaction mixture to a vessel containing IPE, water, and TEA.
The product (10) isolation was accomplished with two crops,
with the second crop isolation involving extractive workup.
Since 10 is a crystalline intermediate, we recognized that the
isolation of 9 could be eliminated to improve the yield. Hence,
after step 3 completion, MeCN in the reaction was exchanged
with EtOH12 for the cladinose hydrolysis. We discovered that
hydrolysis using a pH 2 buffer solution (using DBU as the
2.3. Oxidation, Methanolysis, and Hydrogenolysis. The
Corey−Kim8 oxidation has been demonstrated to be the most
effective method for the oxidation of the C-3 alcohol for the
preparation of ketolides.14 Careful engineering control using a
bleach (5%) scrubber system and proper personnel protection
equipment (PPE) sufficiently addressed the concern of worker
and environmental exposure. Following the standard reaction
protocol, the reaction mixture after an extractive workup was
treated with methanol at reflux to remove the C-2′ acetate to
give 12. In the second scale-up campaign, we elected to carry
the methanol solution of 12 to the hydrogenolysis. In both
campaigns we found that the hydrogenolysis stalled regardless
whether 12 was isolated prior to step 7. Recrystallization and/
or active carbon treatment of 12 did not remedy this issue.15 In
C
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
the end, we resorted to pretreating the methanolic solution of
12 with Pearlman’s catalyst, followed by hydrogenolysis. This
protocol allowed the reaction to go to completion. Azetidine 13
was isolated in 84% yield overall (three steps) compared to
67% yield overall from the first campaign.
identification of a monotosylate salt as a white, crystalline solid.
The preparation of this salt proved straightforward on scale; a
solution of 4 in acetone was treated with toluenesulfonic acid
(1 equiv) at 40−45 °C to give the crystalline product. The final
API was isolated in greater than 98% purity. It is noteworthy
that preparation of the tosylate salt using ∼92% pure 4 (from
freebasing the fumarate salt of 4) was not successful.
Table 3. Comparison of first and second campaigns for steps
3. Synthesis of the Azetidine Linker and the
Napthyridine Headpiece. 3.1. Synthesis of 1-Benzhydryl-
azetidine-3-ylamine (8). Several preparations of 8 have been
reported in the literature from commercially available 1-
azetidinol 23.17 One method (Scheme 4, steps a/c/d)
employing the Gabriel synthesis is inefficient and suffers from
the use of hydrazine hydrate (high thermal potential) to
deprotect the phthalimide group. Another method (Scheme 4,
steps a/h/i) employs azide displacement of the sulfonate
followed by reduction18 This chemistry was not pursued due to
the process safety hazard associated with handling azides. The
most attractive chemistry in the literature is the direct
aminolysis (Scheme 4, steps a/e) of the mesylate 24, but the
procedure as described gave a poor yield (27%) of 8.19
5, 6, and 7
Our initial approach was a three-step sequence going through
an oxime intermediate (Scheme 4, steps g/j/k). Swern
oxidation of azetidinol 23 provided azetidinone 28 in good
yield. Condensation of 28 with hydroxylamine hydrochloride
provided oxime 29, which was reduced to the amine using
LiAlH4. This sequence routinely gave 8 in good yields with the
product conveniently isolated as the oxalic acid salt. To further
streamline the process, the oxime intermediate was taken
directly into the LAH reduction without isolation. Several
hundred grams of 8 were delivered by this route early in the
program, but as material needs increased, we recognized that
the low-temperature Swern oxidation and tedious workup of
the LAH reduction negatively impacted operational cost and
efficiency if they were to be run in the pilot plant.
We elected to revisit the direct aminolysis of mesylate 24 and
initiated an optimization of the reaction.20 For the first step
(synthesis of mesylate 24), we first addressed reaction solvent
and base, since the literature employed pyridine in both roles.
By simply changing the reaction conditions to dichloromethane
as the solvent and triethylamine as the base the yield was
improved to ∼80% after a simple extractive workup. Since
mesylate 24 is a crystalline solid with minimal solubility in
water, the use of acetonitrile (3 L/kg.) as reaction solvent
further streamlined the process by enabling product isolation
simply with the addition of water. The precipitated solid
product was filtered to give a quantitative yield. Subsequently,
we established that the wet cake could be directly used in the
subsequent aminolysis reaction without drying.
Attention was then turned to the aminolysis of the mesylate
(24). In our hands the literature conditions19 did indeed result
in poor yields of 8. Our initial modification of the procedure
used the more readily handled aqueous 28% ammonium
hydroxide (10 L/kg) and isopropanol (15 L/kg) at 75 °C in a
standard reaction flask open to the atmosphere. Under these
conditions the desired product 8 was observed as the major
product, but a dimeric byproduct (27, Scheme 4) was also
formed at high levels (>35% by HPLC). While we anticipated
that some 8 would react further with the mesylate (24) to give
the dimer (27), we had hoped that the large excess of ammonia
would suppress its formation. To prevent loss of NH3 during
the course of the reaction, the process was next conducted in a
Parr reactor using 7 N ammonia in methanol for the aminolysis.
2.4. Reductive Amination and Final Form. The original
reductive amination utilized to couple azetidine 13 with
naphthyridine aldehyde 14 employed acetic acid, 4 Å powdered
molecular sieves, and NaBH(OAc)3 in methylene chloride. The
reaction suffered from poor yields (∼20%), and even after
multiple silica gel chromatographies the highest purity achieved
was still well below 90% (HPLC area %). The isolated product
was an amorphous, bright yellow, foamy solid that presented
significant difficulty for scale-up and isolation.
We examined the reaction and found that the original
conditions gave a substantial amount of azetidinyl acetamide
byproduct (13a, Scheme 6). This was overlooked initially due
to the lack of chromophore by HPLC. In fact, when 13 was
heated with acetic acid (in the presence of 2 equiv TEA to
neutralize the HCl), formation of 13a was almost quantitative.
Simply changing the acid component to pivalic acid effectively
blocked azetidine amidation. An additional improvement was
realized by concentration of THF to azeotropically remove
water,16 which ensured that iminium intermediate 21 was
formed completely. The reaction mixture was then added to a
mixture of STAB in EtOAc and acetonitrile. We observed that
the presence of acetonitrile consistently gave a cleaner
reduction for reasons that are not well understood. The
reaction was quenched by the addition of 5% sodium
bicarbonate solution followed by a standard extractive workup.
During the scale-up, about 30% of the product was retained as
borate complex 22 as determined by LC−MS analysis. Thus,
the organic solution after extractive workup was concentrated
and heated at reflux in methanol for 24 h. This effectively
converted 22 to the desired product (4) cleanly.
Since 4 was an amorphous, foamy solid, its isolation on scale
was problematic. Salt screening led to the identification of the
fumarate salt of 4 as a noncrystalline solid form that, while not
ideal, was deemed a better alternative than the foamy freebase
as it allowed the product isolation by filtration. Thus, 4 was
treated with fumaric acid in ethyl acetate, and the resulting
precipitate was filtered to give the final API as fumarate salt in
54% yield in ∼92% HPLC purity. Further purification was only
possible by chromatography, and after preparative-scale reverse-
scale chromatography 4 was isolated in ≥97% purity. With
cleaner material, a reinvestigation of salts of 4 resulted in
D
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
a
Scheme 3. Endgame synthesis of 4
a
Reagents and conditions: a) pivalic acid, TEA, THF. b) Na(OAc)3BH in MeCN/EtOAc. c) MeOH, reflux. d) chromatography; tosylate salt
formation.
On heating this reaction to 70−75 °C, the pressure of the
vessel rose to 40−50 psi. After 3 h, the reaction was nearly
complete with a mixture of 8:27 in a 94:6 ratio by HPLC. The
product was isolated by evaporation of the volatiles and
recrystallized from MTBE to give a 70% yield of 8. The use of a
closed vessel for this process has a dual benefit of retaining the
NH3 and increasing the rate of the reaction, as the rate of SN2
reactions have been reported to be accelerated by applying
pressure to the reaction.21
Since 7 N NH3/MeOH is not readily available on scale, we
next employed 28% aqueous NH4OH. Again we opted to run
the same reaction (10 L/kg of 28% NH4OH and 15 L/kg of
isopropanol) in the Parr Reactor and observed the dimer was
reduced to approximately 4%. For workup, the reaction mixture
was concentrated under reduced pressure to remove the
isopropanol and the residue extracted into MTBE. As the free
base is difficult to isolate as a solid, the acetate salt was
precipitated directly from the organic extracts on addition of 1
equiv of acetic acid. This protocol gave a 72−84% yield of 8 on
multihundred gram scale in the lab. The methanolic ammonia
and the ammonium hydroxide aminolysis reactions were
comparable in yields, and azetidine ring-opening products
were not observed in either reactions. Using this protocol, more
than 20 kg of 3-amino-1-benzhydryl-azetidine monoacetate (8)
was prepared in high quality (>99% purity).
3.2. Synthesis of 3-Hydroxy-1,5-naphthyridine-4-carbalde-
hyde (14). The synthesis started with 3-bromo-5-methoxpyr-
idine.22 The initial approach followed a literature precedent23
by treatment with tert-butylcarbamate under palladium-
catalyzed conditions to give 30 (Scheme 5). The high cost of
the palladium catalyst prompted us to search for more
economical alternatives. Consequently, a more cost-effective
amination was found using aqueous ammonia under Cu2+-
catalyzed conditions at 130 °C.24 Installation of the methyl
group at the 4-position and the subsequent removal of the Boc
group proceeded uneventfully to afford 33.25
The naphthyridine was formed via a Skraup reaction
(Scheme 6). This process was highly exothermic26 and was
scaled up successfully by carefully controlling the addition rate
of glycerol to the heated reaction mixture. Following the
E
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
conversion was observed after 36 h. With the addition of one
equivalent of Bredereck’s reagent, however, complete con-
version to enamine 37 was observed. We reasoned that
HC(NMe2)3, a super base generated from disproportionation
of Bredereck’s reagent,32 acted as the catalyst in the reaction.
However, when HC(NMe2)3 was used directly as a base in the
reaction, the reaction could not be driven to completion (∼60%
conversion). More conveniently KOtBu or LiOH proved
competent catalysts and their use resulted in the enamine
formation being complete in 16 h at 120 °C.33 Enamine 37 was
easily isolated and purified after removing volatile materials and
triturating the residue in MTBE. Nevertheless, we found that
no benefit was gained through isolation of the intermediate for
scale-up. Thus, the reaction mixture was carried directly into
the oxidative cleavage. Ozonolysis of 37 was attempted but
failed to produce the desired aldehyde. Oxidation was effected
with sodium periodate uneventfully.34
Scheme 4. Synthesis of 1-Benzhydryl-azetidine-3-ylamine
(8)
a
Deprotection of the methyl ether was not straightforward.
The matter was further complicated by the poor solubility of
14.35 Typical conditions (BBr3, TMSI, and Me2BBr)36 to cleave
pyridine or quinoline methyl ethers did not offer satisfactory
results. Warm aqueous HCl gave the best results in the
preliminary screening. However, the product obtained was of
low potency (∼60%). Further investigation led to the use of
LiCl in DMF (110−120 °C).37,38 In this method, the lithium
salt of 14 was isolated from stirring in methanol after
evaporation of DMF. The lithium salt, dissolved in hot water,
was then purified by treatment with active carbon to remove
color. The hot aqueous solution was then acidified to pH 5−6.
Desired product 14 crystallized from the aqueous solution, and
was isolated in 56% yield with high purity (>99%) (see Scheme
6).
a
Reagents and conditions: a) MsCl, pyridine, −10 °C, 80%; b) MsCl,
NEt3, CH3CN, −10 °C, 100%; c) potassium phthalimide,
CH3(CH2)15P+Bun Br−, PhMe, reflux, 67%; d) H2NNH2·H2O,
3
MeOH, reflux, 96%; e) NH3/MeOH, 27%; f), 28% aq NH3/PriOH,
72−84%. g) Oxalyl chloride, DMSO, −78 °C, 93%. h) NaN3, 90%. i)
LAH, 74%. j) NH2OH·HCl, 96%. k) LAH, 71%.
Scheme 5. Synthesis of 5-amino-3-methyl-4-methylpyridine
(33)
a
a
Reagents and conditions: a) BocNH2, Pd2(dba)3−CHCl3, Xantphos,
Scheme 6. Final synthesis of 3-hydroxy-1,5-naphthyridine-4-
Cs2CO3, dioxane, 100 °C, 84%. b) NH4OH, CuSO4, 130 °C, 82%. c)
n-BuLi, THF; MeI, 80%. d) Boc2O, THF, 100%. e) 4 N HCl, MeOH;
96%.
a
carbaldehyde (14)
reported procedure of a quinoline synthesis,27 we found the
reaction gave the product (34) in 3−48% yield using
methanesulfonic acid. By switching to 75% aqueous H2SO4
the reaction yield was improved to 57−72%. Oxidation of 34 to
the corresponding aldehyde proved to be challenging. Direct
oxidation using selenium dioxide, or tBuI/DMSO (the
Vismara28 method) was not successful. A long sequence that
involved bromination with NBS followed by acetate displace-
ment, ester hydrolysis, and Dess−Martin oxidation was initially
used to meet preclinical needs. In addition to the process safety
concerns of using oxidation conditions (SeO2, tBuI/DMSO or
Dess−Martin methods), the approach suffered from the lack of
scalability of the free radical bromination, the instability of the
benzylic bromide 35, and the low throughput. Oxidation of the
benzylic bromide (35) with either modified Swern or Hass−
Bender conditions29 gave aldehyde 38 in poor yields.
a
Reagents and conditions: a) sodium 3-nitrobenzenesulfonate, 75%
H2SO4, FeSO4·7H2O, boric acid, glycerol, 125−130 °C; 57−72%. b)
NBS, dibenzoyl peroxide, CCl4, 72%. c) KOAc, DMF; then MeOH,
K2CO3; 63%. d) DMF−DMA, LiOH, DMF, 120 °C. e) NaIO4,
MeOH, 79% overall last two steps. f) IBX, EtOAc. g) LiCl, DMF, then
aq HCl; 56%.
In an alternative approach, we took inspiration from a two-
step process approach involving conversion of 4-methyl
pyridines to their corresponding aldehydes via an enamine
intermediate. In this approach, Bredereck’s reagent30 or N,N-
dimethylformamide dimethylacetal (DMF−DMA) was em-
ployed to form the N,N-dimethylenamine which,31 on
treatment with sodium periodate, could be readily transformed
to the corresponding aldehyde. When 34 was treated with
DMF−DMA (3.0 equiv) in DMF at 130 °C only 30%
CONCLUSION
■
We have described a scale-up process for the synthesis of
ketolide 4. For the synthesis leading to the penultimate
intermediate (13), the throughput was improved to 62% overall
in seven steps in the 10 kg API campaign, as compared to 23%
from the initial 2 kg scale-up. Solvent usage and waste streams
were reduced by nearly 80 L/kg. With telescoped operations
F
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
and direct isolations implemented, the overall process became
much more streamlined and efficient. The coupling of the
penultimate intermediate (13) with the naphthyridine aldehyde
(14) was enabled to provide a robust synthesis of the API. A
crystalline monotosylate salt was discovered, which cleared a
big hurdle in formulation development.
The preparations of both the amino azetidine (8) and the
naphthyridine aldehyde (14) in multikilogram scales were also
discussed. Aminolysis of mesylate 24 using either 7 N
methanolic ammonia or 28% aqueous ammonium hydroxide
under pressure was found to be more process efficient and
friendly than the traditional Gabriel method or the azide/
reduction approach. The process development and execution of
a seven-step synthetic sequence for the synthesis of
naphthyridine aldehyde (14) presented significant challenges.
We were pleased with the successful scale-up of the highly
energetic Skraup reaction and the identification of a convenient
and effective oxidation of 4-methylnaphthyridine using DMF−
DMA in the presence of catalytic amount of LiOH.
It is recognized that substantial development work remains
necessary should the demand arise for a much larger API
campaign. The convergent route (Scheme 2) is worthy of
reinvestigation by exploring in situ the 3-keto protection and/
or reordering oxidation step. Future considerations also include
the need to replace the Corey−Kim oxidation with a more
worker and environmentally friendly method, substitution of
the benzhydryl group with a more readily removed protecting
group, and elimination the chromatographic purification of the
API.
kg, 55.2 mol), and acetonitrile (240 L). After stirring for 15
min, DBU (24.0 kg, 155.5 mol) was added. The reaction was
heated at 50 °C for 18 h. HPLC analysis confirmed reaction
completion. The reaction was concentrated by vacuum
distillation to a lowest stirrable volume, the residual acetonitrile
was solvent-exchanged with ethanol under vacuum until
acetonitrile was below 0.5% by GC. Water (287 L) was
added, followed by slow addition of conc HCl (∼38 L) until
pH of 2.0 was reached. The reaction was heated at 38 °C for
16−18 h. The reaction completion was confirmed by HPLC
analysis. The reaction was cooled to 20−25 °C, and water (240
L) was added. The pH of the reaction was adjusted up by slow
addition of triethylamine with a target range of 7−7.5. The
resulting slurry was stirred for 2 h at 15−20 °C and filtered.
The filter cake was rinsed with water (240 L) and vacuum
oven-dried (45 °C, 50 mmHg). This gave the desired product
10 (35.1 kg, 76%) as a white solid: MS (API-ES) 878.5 (M +
1)+, 712.3, 679.3, 596. 1H NMR (CDCl3) δ 0.82 (t, 3H, J = 7.3
Hz), 0.91 (d, 3H, J = 7.4 Hz), 0.98 (d, 3H, J = 6.6 Hz), 1.05 (d,
3H, J = 7.5), 1.09 −1.88 (m, 10H), 2.05 (s, 3H), 2.13−2.15 (m,
2H), 2.17−2.30 (m, 6H), 2.38 −2.55 (m, 2H), 2.64−2.85 (m,
6H), 2.90 −3.01 (m, 2H), 3.24−3.65 (m, 6H), 3.68−3.79 (m,
3H), 4.25−4.78 (m, 6H), 5.02 (br s, 1H), 5.29 (s, 1H), 5.57
(dd, 1H, J = 6.0, 1.9 Hz) 7.05−7.58 9 (m, 10 H).
3-Descladinosyl-11,12-dideoxy-6-O-methyl-3-oxo-
12,11-(oxycarbonyl-(azetidin-3-yl)-imino)-erythromycin
A Dihydrochloride (13). To glass-lined reactor A was charged
NCS (4.57 kg, 34.2 mol) and anhydrous DCM (75 L). After
stirring for 15 min, the solution was cooled to −10 °C, and
dimethyl sulfide (2.23 kg, 35.9 mol) was added, while keeping
the temperature under −10 °C. In a separate glass-lined reactor
B was charged 10 (15 kg, 17.1 mol) and then anhydrous DCM
(75 L). The resulting solution was cooled to −10 °C and then
transferred to reactor A slowly, while maintaining the reaction
temperature below −10 °C. After the transfer was complete,
triethylamine (1.9 kg, 18.8 mol) was added, while keeping the
reaction below −10 °C. The reaction was stirred for 1 h under
−10 °C. HPLC analysis indicated reaction completion. The
reaction was allowed to warm to 10−15 °C, and saturated
sodium bicarbonate solution (90 L) was added. The organic
phase was separated and washed with saturated sodium
bicarbonate solution (90 L) and saturated brine solution (75
L) successively. The organic phase was concentrated to a small
stirrable volume, and MeOH (150 L) was added. The resulting
solution was heated at reflux (66 °C) for 12 h. The reaction was
cooled. Concentrated HCl (approximately 2 kg) was added to
adjust the pH to a target range of 3.8−4.2. The resulting
solution was then added to a nitrogen-purged vessel containing
3 kg of 20% Pd(OH)2 on carbon (50% wet). The resulting
mixture was stirred at 50 °C for 2 h and filtered. The filtrate
was then subjected to hydrogenolysis conditions under 50 psi
H2 in the presence of 1.5 kg 20% Pd(OH)2 on carbon (50%
wet) at 50 °C for 2 h. Upon reaction completion, the reaction
was cooled to RT and then purged with nitrogen three times
and filtered. Concentrated HCl (approximately 200 mL) was
added to readjust the pH to a range of 3.8−4.2. The resulting
solution was concentrated under partial vacuum. The residual
water was azeotropically removed by addition of THF (∼90 L
total). MTBE (75 L) was added, and the resulting mixture was
granulated for 2 h. The batch was filtered, and the solids were
dried under vacuum to afford the desired product 13 (10.6 kg,
84% overall three steps) as a white solid: MS (ESI+) for m/z
EXPERIMENTAL SECTION
■
General. Achiral HPLC analyses were carried out using
Agilent SB-CN columns (4.6 mm × 250 mm) with acetonitrile/
0.2% perchloric acid aqueous buffer (20/80 or 40/60) as
mobile phase (2 mL/min) and detection at 210 nm wavelength.
HPLC purity is reported by area %.
10,11-Didehydro-11-deoxy-6-O-methyl-2′,4″-diace-
tate-12-(1H-imidazole-1-carboxylate)-erythromycin (7).
To a glass-lined reactor was charged clarithromycin (50 kg,
66.89 mol), followed by THF (500 L). After stirring for 20 min,
TEA (20.3 kg, 200.7 mol) and DMAP (408 g, 3.35 mol) were
added. This was followed by addition of Ac2O (20.5 kg, 200.7
mol). The reaction was stirred at 20−25 °C for ∼30 h. Water
(150 L) was added, and the mixture was stirred for 30 min. The
aqueous layer was removed. The organic phase was
concentrated under atmospheric pressure (∼250 L of THF
was removed). A KF analysis confirmed water at ≤0.1%. The
reactor was replenished with the addition of 150 L of
anhydrous THF. CDI (22.8 kg, 140 mol) was added in one
portion, followed by the addition of DBU (10.2 kg, 66.89 mol)
at 20−25 °C. The reaction was heated to 50 °C for 2 h until
complete conversion to 7 was noted by HPLC. The reaction
was cooled, and 3.5 wt % aq NH4Cl solution (200 L) was
added. The resulting mixture was stirred for 1 h and filtered.
The filter cake was rinsed with water (25 L) and pulled dry
under nitrogen for 2 h. The product collected was further
vacuum oven-dried (45 °C, 60 mmHg) to give 7 as a white
solid (59.5 kg, 98%). Analytical data of the product were
identical to those reported.8
3-Descladinosyl-11,12-dideoxy-6-O-methyl-12,11-
(oxycarbonyl-(1-benzhydryl-azetidin-3-yl)-imino)-eryth-
romycin (10). To a glass-lined reactor was charged acyl
imidazole 7 (47.80 kg, 52.6 mol), 3-amino azetidine 8 (16.78
1
668.4 (M + H)+. H NMR (CD3OD) δ 0.78 (t, 3H, J = 7.46
G
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Hz), 0.91 (d, 3H, J = 6.64 Hz), 1.06 (d, 3H, J = 6.64 Hz),
1.08−1.30 (m, 11H), 1.32−1.79 (m, 2H), 1.95 (m, 1H), 2.38−
2.54 (m, 1H), 2.59 (s, 3H), 2.68 (s, 3H). 2.77 (s, 3H), 3.02−
3.39 (m, 6H), 3.52−3.65 (m, 3H), 3.94 (q, 1H, J = 7.46 Hz),
4.03−4.30 (m, 5H), 4.42 (t, 2H, J = 6.64 Hz), 4.51−4.63 (m,
2H), 4.78−4.93 (m, 2H).
(dd, J = 6.61 Hz, 1H), 3.83 (dd, J = 6.62 Hz, 1H), 3.96−4.00
(m, 1H), 4.00−4.05 (m, 2H), 4.06 (q, J = 6.66 Hz, 1H), 4.12
(d, J = 8.19 Hz, 1H), 4.27 (d, J = 7.17 Hz, 1H), 4.65 (s, 2H),
4.79 (dd, J = 10.24, 0.66 Hz, 1H), 5.97 (br. s, 1H), 7.07−7.14
(m, 2H), 7.44−7.52 (m, 2H), 7.56 (dd, J = 8.19, 4.35 Hz, 1H),
8.29 (dd, J = 8.19, 1.54 Hz, 1H), 8.64 (s, 1 H), 8.88 (dd, J =
4.35, 1.79 Hz, 1H). 13C NMR (151 MHz, DMSO-d6) δ 10.20,
13.45, 13.60, 13.93, 15.35, 17.62, 19.48, 20.74, 21.66, 29.80,
37.78, 38.11, 40.05, 44.63, 45.33, 46.43, 49.11, 50.07, 52.19,
57.36, 58.90, 61.74, 64.38, 67.74, 68.76, 75.76, 77.69, 77.78,
82.61, 102.60, 120.48, 121.26, 125.47, 127.99, 136.84, 136.85,
137.53, 142.83, 144.73, 145.76, 150.57, 153.66, 155.13, 169.61,
204.18, 216.04. HRMS: calcd for C43H64N5O11: 826.45969;
found: 826.45985.
1-Benzhydryl-azetidin-3-yl methanesulfonate (24). To
a reaction flask was charged 632 g (2.64 mol) of 1-benzhydryl-
azetidin-3-ol, acetonitrile (1.9 L), and triethylamine (601 g, 1.5
equiv). The mixture was cooled in an ice−acetone bath (−5
°C). Methanesulfonyl chloride (436 g, 1.20 equiv) was added
via a drop funnel while keeping the reaction temperature at <5
°C. HPLC showed reaction completion after 15 min. Water
(6.3 L) was added, and the reaction mixture was stirred for 2 h
at room temperature and filtered. The filter cake was rinsed
with water (2 × 1 L), pulled dry under vacuum, and directly
subjected to the amination reaction in the next step.
1-Benzhydryl-azetidine-3-ylamine (8). The mesylate wet
cake (838 g dry weight expected, 2.64 mol) was dissolved in
isopropanol at 50 °C. The solution was charged to a 2 gal Parr
reactor, followed by the addition of 28 wt % ammonium
hydroxide under vacuum. The Parr reactor was sealed and
heated to 71 °C for 3 h (38−40 psi pressure observed). The
reaction was assayed by HPLC and showed reaction
completion. The reaction mixture was cooled to room
temperature, discharged from the Parr reactor, and concen-
trated under vacuum. The product was extracted with isopropyl
ether (8.4 L). The organic extract was concentrated to ∼4 L
under atmospheric pressure, and 159 g (1 equiv) of acetic acid
was added; the mixture was stirred for 2 h, and the product
(monoacetate salt) was collected by filtration. The solids were
dried at 40 °C under vacuum to give the desired product as a
white solid (662 g, 84%). 1H NMR (CD3OD, 400 MHz) 7.42−
7.04 (m, 10 H), 4.44 (s, 1H), 3.78−3.62 (m, 1H), 3.43−2.36
(m, 2H), 3.03−2.99 (m, 2H), 1.93 (s, 3H). 13C NMR
(CD3OD, 100 MHz) 176.2, 141.4, 128.3, 127.3, 127.2, 77.5,
58.3, 41.2, 22.2.
5-Methoxy-4-methylpyridin-3-amine Hydrochloride
(33). A solution of tert-butyl 5-methoxypyridin-3-ylcarbamate
(10.1 kg; 45 mol) in THF (125 L) was cooled to −70 °C. n-
BuLi (2.5 M; 45.0 L; 112.5 mol; 2.5 equiv) was added
dropwise, keeping Tint < −60 °C. The resulting mixture was
warmed to −25 °C and stirred at that temperature for 1 h. The
resulting solution was cooled to −70 °C, and a 2.0 M solution
of MeI in MTBE (40 L, 67.5 mol, 1.5 equiv) was added
dropwise, keeping Tint < −65 °C. The resulting mixture was
stirred for 1 h at −65 °C and subsequently quenched by
addition of water (50 L), allowing the reaction to warm up to
10 °C. The layers were separated, and the organic layer was
washed with sat. NaCl, and concentrated to approximately 60 L
volume. MeOH (10 L) was added, followed by addition of 4 N
HCl (25 L, 100 mol). The resulting mixture was heated at 45
°C for 4 h and then concentrated to ∼35 L. The remaining
aqueous phase was basified to pH 10.5 using 4 N NaOH
solution. The precipitated product was collected by filtration.
3-Descladinosyl-11,12-dideoxy-6-O-methyl-3-oxo-
12,11-(oxycarbonyl-(1-((3-hydroxy-[1,5]-naphthyridin-4-
yl)-methyl)-azetidin-3-yl)-imino)-erythromycin A Tosy-
late Salt (4). The 3-hydroxy-1,5-naphthyridine-4-carbaldehyde
14 (2.0 kg, 11.6 mol) and 3-descladinosyl-11,12-dideoxy-6-O-
methyl-3-oxo-12,11-(oxycarbonyl-(azetidin-3-yl)-imino)-eryth-
romycin A dihydrochloride 13 (8.28 kg, 11.18 mol) were
combined in THF (50 L). TEA (1.98 kg, 19.6 mol) was added.
The mixture was stirred for 30 min, followed by the addition of
pivalic acid (2.67 kg, 26.1 mol). The mixture was heated to
reflux, and approximately 30 L of solvent was removed by
concentration under atmospheric pressure to ensure complete
removal of water (50 L additional amount of anhydrous THF
was added during the concentration). The mixture was then
cooled to 20−25 °C and transferred to another reactor that
contained sodium triacetoxyborohydride (5.56 kg, 27.6 mol),
acetonitrile (25 L), and ethyl acetate (50 L) at 20−25 °C under
agitation. After stirring for 30 min, 5% aq sodium bicarbonate
(50 L) was added. The layers were separated. The aqueous
layer was extracted with ethyl acetate (2 × 50 L). The
combined organic phase was treated with anhydrous MgSO4,
and then active carbon (2.0 kg) was added. The mixture was
filtered and concentrated to ∼10 L volume. The concentrate
was purified by preparative reverse-phase chromatography
using conditions as follows: Column: Kromasil C-18 100 Å;
Column size: 8 in. I.D.; Temperature: 20 °C; Mobile phase:
Mobile phase A = 90:10:0.06% acetonitrile/water/phosphoric
acid, Mobile phase B = acetonitrile; Gradient: 15% acetonitrile
0−2 min, to 90% acetonitrile over 30 min, hold 90%
acetonitrile 3 min, return to 10% acetonitrile and hold 5 min;
Flow rate: 1.95 L/min, UV Detection Wavelength: 254 nm;
Feed Concentration: 127 g/kg of 33% v/v acetonitrile in water;
Feed Amount: ∼125 mL each injection.
Desired fractions were combined, and the acetonitrile was
removed by concentration in vacuo under 20 °C. The
remaining aqueous mixture was pH adjusted to 8.4 using 4%
aq NaHCO3 solution and then was extracted with EtOAc (3 ×
90 L). The combined organic phase was washed with saturated
brine solution and then concentrated under partial vacuum
(100 mmHg, 45 °C) to ∼20 L. The reductive amination and
purification processes were repeated at the same scale.
The combined ethyl acetate concentrate obtained above was
further reduced in volume to ∼15 L by vacuum distillation.
Acetone (15 L) was added, followed by dropwise addition of a
solution of TsOH·H2O (2.39 kg, 12.4 mol) in acetone (40 L).
After the addition was complete, the mixture was stirred for 2 h
at 20−25 °C. The product crystallized out, was filtered, rinsed
with cold acetone (10 L), and dried under vacuum to give the
desired product (9.82 kg, 44%) as an off-white solid: mp 198.0
1
°C; H NMR (600 MHz, DMSO-d6) δ 0.80 (t, J = 7.42 Hz,
3H), 0.89 (d, J = 6.66 Hz, 3H), 1.10 (d, J = 7.17 Hz, 3H),
1.18−1.24 (m, 12H), 1.36 (dd, J = 12.29, 10.75 Hz, 1H), 1.48
(s, 3H), 1.55−1.60 (m, 1H), 1.60−1.64 (m, 1H), 1.65−1.71
(m, 1H), 1.71−1.77 (m, 1H), 1.90 (dd, J = 10.24, 3.07 Hz, 1H),
2.29 (s, 3H), 2.44 (dd, J = 7.17, 2.56 Hz, 1H), 2.51 (s, 3H),
2.61 (s, 6H), 3.05−3.12 (m, 2H), 3.16−3.26 (m, 1H), 3.29 (dd,
J = 10.75, 7.68 Hz, 1H), 3.46 (s, 1H), 3.60−3.66 (m, 1H), 3.69
H
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
The filter cake was sent back to the reactor, and dissolved in
EtOAc (50 L). Gaseous HCl (1.8 kg, 49.5 mol) was introduced
to the resulting solution slowly, while keeping the reaction 20−
25 °C. The resulting slurry was stirred for 2 h and filtered. The
filter cake was rinsed with EtOAc (5 L) and dried to give 33
hydrochloride salt as a light-yellow solid (5.8 kg, 74%). H
NMR (CDCl3, 300 MHz) δ 2.03 (s, 3H), 3.61 (bs, 2H), 3.87 g
(s, 3H), 7.71 (s, 1H), 7.76 (s, 1H).
second crop. Both crops from all four batches were combined
to give 38 as a yellow solid (10.2 kg, 79%): mp 170−172 °C
1
(MeCN). H NMR (CDCl3 400 MHz) δ 11.35 (1H, s), 9.09
(1H, s), 9.05 (1H, dd, J = 4.36, 1.86 Hz), 8.425 (1H, dd, J =
8.41, 1.55 Hz), 7.61 (1H, dd, J = 8.41, 4.05 Hz), 4.246 (3H, s).
13C NMR (CDCl3, 100 MHz) δ 16.0, 57.9, 122.6, 137.4, 141.4,
152.6, 191.9.
1
3-Hydroxy-1,5-naphthyridine-4-carbaldehyde (14).
LiCl (6.38 kg, 150 mol) was added to DMF (26.4 kg). The
resulting mixture was heated to 110 −120 °C. To a second
reactor, was charged DMF (44.2 kg) and 38 (9.4 kg, 50 mol),
and the mixture was heated to 110−120 °C. The LiCl in DMF
prepared above was transferred to the second reactor. The
mixture was stirred for 30 min at 110−120 °C and then cooled
to 30−40 °C. The mixture was concentrated at 60−70 °C
under vacuum until the residue was 20−30 L. The material was
transferred to a rotovap and continued to concentrate to
dryness at 60−70 °C under vacuum (5−10 mmHg). Methanol
(22.4 kg) was added, and the resulting mixture was transferred
to a reactor and cooled to 0−5 °C. After stirring for 5 h, the
mixture was filtered using a centrifuge filter, and the filter cake
was rinsed with MeOH (7.4 kg, precooled to 0−5 °C). The
filter cake was dried at 40−50 °C under vacuum to obtain the
lithium salt of 1. This was combined with water (140 kg) and
active carbon (0.93 kg). The mixture was heated at 90−100 °C
for 3 h. The mixture was filtered hot at 80−90 °C. To the
filtrate was added 6 N HCl solution (5.31 kg) until the pH was
5−6. The reaction mixture was then stirred for 10 h at 0−5 °C
and filtered. The filter cake was rinsed with water and dried at
40−50 °C under vacuum until KF ≤0.5%. This gave the
product as a yellow solid (4.87 kg, 56%): mp 238−240 °C
3-Methoxy-4-methyl-1,5-naphthyridine (34). To a
glass-lined reactor was charged conc. H2SO4 (49 kg) and
sodium 3-nitrobenzene sulfonate (20.9 kg, 95.5 mol) in
portions. FeSO4·7H2O (1.66 kg, 5.98 mol) and boric acid
(2.86 kg, 46.2 mol) were added in one portion, below 40 °C.
Water (13.3 kg) was then added slowly below 40 °C, followed
by addition of hydrochloride salt of 33 (10 kg, 57.3 mol). The
reaction mixture was heated to 135−140 °C. Glycerol (14.7 kg,
16.0 mol) was added at the rate of 50 10 mL/min, while
keeping the reaction at 135−145 °C. The reaction was heated
for 6 h and then cooled to 80−90 °C. The reaction mixture was
transferred into a vessel containing water (30 kg) and ice (100
kg); the pH was then adjusted to 8−9 with 20% aq NaOH.
Additional water (220 kg) was added to dissolve the inorganic
salts. The mixture was then extracted with EtOAc (3 × 200 kg).
The combined organic phase was stirred with Na2SO4 (5 kg)
and active carbon (1 kg) for 4 h. After filtration, the filtrate was
concentrated under reduced pressure until black solid appeared
as precipitate. DCM (54 kg) was added, the resulting mixture
was stirred into complete solution with heat (40 °C), cooled,
and washed with 10% aq NaOH (3 × 15 kg), followed by water
(15 kg). The organic phase was filtered to remove any insoluble
material and concentrated to dryness to give the desired
product as a yellow solid (6.79 kg, 68%): mp 100−101 °C
1
(MeOH). MS (ESI+) for m/z 175 (M + H)+. H NMR (D2O,
1
400 MHz) δ 10.04 (1H, s), 8.54−8.46 (1H, m), 8.46−8.40
(1H, m), 8.32 (1H, s), 7.56−7.46 (1H, m).
(MeCN). LCMS m/z (M + 1) 175.1. H NMR (CDCl3, 300
MHz) δ 2.71 (s, 3H), 4.11 (s, 3H), 7.50 (dd, J = 4.1/8.5 Hz,
1H), 8.32 (dd, J = 1.8/8.5 Hz, 1H), 8.81 (s, 1H), 8.95 (dd, J =
1.8/4.1 Hz, 1H).
3-Methoxy-1,5-naphthyridine-4-carbaldehyde (38).
To a glass-lined reactor was charged DMF (113 kg) and 34
(12 kg, 68.9 mol). After stirring into a complete solution, LiOH
(0.33 kg, 13.8 mol) and DMF−DMA (16.4 kg, 137.6 mol)
were added sequentially. The reaction was heated to 120−130
°C for 24 h and then cooled to RT. The reaction mixture was
filtered and the filtrate concentrated under reduced pressure
AUTHOR INFORMATION
■
Corresponding Author
*Telephone: (860) 715-1448. E-mail: bryan.li@pfizer.com.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
until 105
5 kg of distillate was collected. The resulting
We are indebted to Mr. Michael St. Pierre, Mr. John Draper,
and Mr. James Long for their manufacturing support in the
Pfizer Groton Kilo Lab and the Sandwich Pilot Plant in the UK.
We thank Dr. Martin A. Berliner and Mr. Michael Fichtner for
their valuable inputs during the preparation of the manuscript.
residue was cooled to RT, and methanol (115 kg) was added.
The resulting enamine solution in methanol was split into four
identical batches for the sodium periodate oxidation. Thus, to
one-fourth of the batch was added water (4.4 kg) followed by
sodium periodate (7.45 kg, 34.8 mol) in portions, while keeping
the reaction temperature at 30−40 °C. The reaction mixture
was stirred at 30−40 °C for 16 h. NaHCO3 (1.45 kg, 17.3 mol)
was added to adjust the pH to 7−8. The reaction mixture was
filtered, and the filter cake rinsed with methanol. The filtrate
was concentrated until the residue volume was 31−47 L. Then
47.2 kg water was added, and vacuum concentration was
continued until the residue was 31−47 L. To this was added 31
kg of water, and the resulting mixture was stirred for 1 h at RT.
The mixture was filtered and the filter cake rinsed with water.
The filter cake was dried at 50 °C under vacuum to give the
product as the first crop. The mother liquor was extracted with
DCM (3 × 50 L), and the combined organic phase was
concentrated to a minimum stir volume. The mixture was then
filtered. The filter cake was dried to give the product as the
REFERENCES
■
(1) Kaneko, T.; Dougherty, T. J.; Magee, T. V. Macrolide Antibiotics.
Therapeutic Areas II: Cancer, Infectious Diseases, Inflammation &
Immunology and Dermatology. Comprehensive Medicinal Chemistry II;
Taylor, J. B.; Triggle, D. J., Eds.; Elsevier Ltd.: Oxford, UK, 2007; Vol.
7, pp 519−566.
(2) (a) Kanatani, M. S.; Guglielmo, B. J. West J. Med. 1994, 160, 31.
(b) Pai, M. P.; Graci, D. M; Amsden, G. W. Ann. Pharmacol. Ther.
2000, 34, 495.
(3) (a) Metlay, J. P.; Fishman, N. O.; Joffe, M. M.; Kallan, M. J.;
Chittams, J. L.; Edelstein, P. H. Emerging Infect. Dis. 2006, 12, 1223.
(b) Lonks, J. R.; Garau, J.; Gomez, L.; Xercavins, M.; de Echaguen, A.
O.; Gareen, I. F.; Reiss, P. T.; Medeiros, A. A. Clin. Infect. Dis. 2002,
35, 556. (c) Lynch, J. P., III; Martinez, F. J. Clin. Infect. Dis. 2002, 34,
S27. (d) Hyde, T. B.; Gay, K.; Stephens, D. S.; Vugia, D. J.; Pass, M.;
I
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Johnson, S.; Barrett, N. L.; Schaffner, W.; Cieslak, P. R.; Maupin, P. S.;
Zell, E. R.; Jorgensen, J. H.; Facklam, R. R.; Whitney, C. G. J. Am. Med.
Assoc. 2001, 286, 1857. (e) Waterer, G. W.; Jones, C. B. Chest 2000,
118, 1839.
(4) Zuckerman, J. M. Infect. Dis. Clin. N. Am 2004, 18, 621.
(5) Hammerschlag, M. R.; Sharma, R. Expert Opin. Investig. Drugs
2008, 17, 387.
(26) It is commonly believed acrolein is generated by dehydration of
glycerol in the reaction. This is supported by the observation of tars in
the Skraup reaction from the polymerization of acrolein.
(27) Conlon, D. A.; Drahus-Paone, A.; Ho, G.-J.; Pipik, B.; Helmy,
R.; McNamara, J. M.; Shi, Y.-J.; Williams, J. M.; Macdonald, D.;
Deschenes, D.; Gallant, M.; Mastracchio, A.; Roy, B.; Scheigetz, J. Org.
Process Res. Dev. 2006, 10, 36.
(28) Vismara, E.; Fontana, F.; Minisci, F. Gazz. Chim. Ital. 1987, 117,
135.
(29) Hass, H. B.; Bender, M. L. Org. Synth. 1963, 4, 932.
(30) Kantlehner, W.; Wagner, F.; Bredereck, H. Liebigs Ann. Chem.
1980, 344.
(31) (a) Kantlehner, W.; Mergen, W. W.; Haug, E. Liebigs Ann. Chem.
́ ́ ́
1983, 2, 290. (b) Vaiu, L.; Senechal, K.; Maury, O.; Guegan, J.; Dupau,
P.; Toupet, L.; Le Bozec, H. Synthesis 2003, 4, 577. (c) Ragan, J. A.;
Jones, B. P.; Meltz, C. N.; Teixeira, J. J., Jr. Synthesis 2002, 4, 483.
(d) Coe, J. W.; Vetelino, M. G. Tetrahedron Lett. 1994, 35, 219.
(32) Csihony, S.; Beaudette, T. T.; Sentman, A. C.; Nyce, G. W.;
Waymouth, R. M.; Hedrick, J. L. Adv. Synth. Catal. 2004, 346, 1081.
(33) While there was no evidence a benzylic anion was involved in
the reaction, it was possible that transient deprotonation of the methyl
group occurred in the reaction. Thus, a strong base was needed in the
reaction.
(6) Magee, T. V.; Ripp, S. L.; Li, B.; Buzon, R. A.; Chupak, L.;
Dougherty, T. J.; Finegan, S. M.; Girard, D.; Hagen, A. E.; Falcone, M.
J.; Farley, K. A.; Granskog, K.; Hardink, J. R.; Huband, M. D.;
Kamicker, B. J.; Kaneko, T.; Knickerbocker, M. J.; Liras, J. L.; Marra,
A.; Medina, I.; Nguyen, T.-T.; Noe, M. C.; Obach, R. S.; O’Donnell, J.
P.; Penzien, J. B.; Reilly, U. D.; Schafer, J. R.; Shen, Y.; Stone, G. G.;
Strelevitz, T. J.; Sun, J.; Tait-Kamradt, A.; Vaz, A. D. N.; Whipple, D.
A.; Widlicka, D. W.; Wishka, D. G.; Wolkowski, J. P.; Flanagan, M. E. J.
Med. Chem. 2009, 52, 7446.
(7) (a) Asaka, T.; Kashimura, M.; Matsuura, A.; Sugimoto, T.;
Tanikawa, T.; Ishii, T. PCT Int. Appl. WO9921869, 1999; Chem.
Abstr. 1999, 130, 296955. (b) Esteban, J.; Costa, A. M.; Cruzado, M.
C.; Faja, M.; Garcia, P.; Vilarrasa, J. Tetrahedron Lett. 2006, 47, 1919.
(c) Kashimura, M.; Asaka, T.; Misawa, Y.; Matsumoto, K.; Morimoto,
S. J. Antibiot. 2001, 54, 664. (d) Baker, W. R.; Clark, J. D.; Stephens, R.
L.; Kim, K. H. J. Org. Chem. 1988, 53, 2340.
(8) Corey, E. J.; Kim, C. U. J. Am. Chem. Soc. 1972, 94, 7586.
(9) (a) Agouridas, C; Denis, A.; Auger, J.-M.; Benedetti, Y.;
Bonnefoy, A.; Bretin, F.; Chantot, J.-F.; Dussarat, A.; Fromentin, C.;
D’Ambrieres, S. G.; Lachaud, S.; Laurin, P.; Martret, O. L.; Loyau, V.;
Tessot, N. J. Med. Chem. 1998, 41, 4080. (b) Kaneko, T.; McMillen,
W.; Lynch, M. K. Bioorg. Med. Chem. Lett. 2007, 17, 5013.
(10) L/kg is defined as liters per kilogram of the limiting reactant in
the reaction.
(11) The second crop isolation was done by subjecting the filtrate to
an aqueous workup, which involved the removal of MeCN, followed
by addition of EtOAc (10 L/kg) and water (10 L/kg), layer separation,
solvent exchanged with EtOH (5 L/kg), and finally recrystallization
from EtOH.
(34) The use of aqueous methanolic solution effectively addressed
the process safety concerns in handling NaIO4.
(35) In all solvents tried that included DMSO, MeOH, H2O, DMF,
THF, DCM, MeCN, PhMe, and acetone, the solubility was <0.3 mg/
mL. The solubility was marginally better in hot DMSO (120 °C) at 1.2
mg/mL.
(36) Greene, T. W.; Wuts, P. G. M. Protective Groups in Organic
Synthesis, 3rd ed., John Wiley & Sons, 1999 and references cited
therein.
(37) Bernard, A. M.; Ghiani, M. R.; Piras, P. P.; Rivoldini, A. Synthesis
1989, 4, 287.
(38) One equivalent of chloromethane was generated from the
demethylation reaction. During the scale-up, the off-gas was scrubbed
with pyrrolidine.
(12) By vacuum distillation of acetonitrile to a minimum stirrable
tank volume, followed by addition of ethanol, and further azeotrope to
remove residual acetonitrile.
(13) We observed that at lower pH a substantial amount of product
was lost in the mother liquor, and at higher pH some impurities were
not excluded from the precipitated product.
(14) (a) Xu, G.; Tang, D.; Gai, Y.; Wang, G.; Kim, H.; Chen, Z.;
Phan, L. T.; Or, Y. S.; Wang, Z. Org. Process Res. Dev. 2010, 14, 504.
(b) Cink, R. D.; Chambournier, G.; Surjono, H.; Xiao, Z.; Richter, S.;
Naris, M.; Bhatia, A. V. Org. Process Res. Dev. 2007, 11, 270.
(15) Residual DMS or DMSO was suspected to be the scapegoat,
though they were detected at ppm levels.
(16) A reliable method was to add a reaction aliquot to a vial
containing STAB in EtOAc/MeCN. The sample was analyzed by
HPLC−MS.
(17) Nisato, D.; Frigerio, M. Fr. Demande, FR 2560194, 1985.
Chem. Abstr. 1986, 533737.
(18) Okada, T.; Ezumi, K.; Yamakawa, M.; Sato, H.; Tsuji, T. Chem.
Pharm. Bull. 1993, 41, 126.
(19) Arimoto, M.; Hayano, T.; Soga, T.; Yoshioka, T.; Tagawa, H.;
Furukawa, M. J. Antibiotics 1986, 39, 1243.
(20) Jung, M. E.; Choi, Y. M. J. Org. Chem. 1991, 56, 6729.
(21) (a) Brower, K. R.; Chen, J. S. J. Am. Chem. Soc. 1965, 87, 3396.
(b) Weale, K. E. Discuss. Faraday Soc. 1955, 20, 83. (c) Lee, K. I.;
Okamoto, Y. J. Org. Chem. 1976, 41, 1552. (d) Evans, M. G.; Polanyi,
M. Trans. Faraday Soc. 1935, 31, 875.
(22) Comins, D. L.; Killpack, M. O. J. Org. Chem. 1990, 55, 69.
(23) Dishington, A. P.; Johnson, P. D.; Kettle, J. G. Tetrahedron Lett.
2004, 45, 3733.
(24) Hauser, C. R.; Reynolds, G. A. J. Org. Chem. 1950, 15, 1224.
(25) (a) Tamura, Y.; Fujita, M.; Chen, L.-C.; Kiyokawa, H.; Ueno, K.;
Kita, Y. Heterocycles 1981, 15, 871. (b) Tamura, Y.; Fujita, M.; Chen,
L-C; Inoue, M.; Kita, Y. J. Org. Chem. 1981, 46, 3564.
J
dx.doi.org/10.1021/op300064b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX