Process development of a potent glucosylceramide synthase inhibitor

Article abstract of DOI:10.1021/op2001222

An economic, scalable process for the production of glucosylceramide synthase (GCS) inhibitor 7 has been developed. Herein we report a three-step synthesis to aldehyde 4 with high yield and purity that employs the selective cleavage of an endocyclic C-O b

Full text of DOI:10.1021/op2001222

ARTICLE
pubs.acs.org/OPRD
Process Development of a Potent Glucosylceramide
Synthase Inhibitor
Christopher G.F. Cooper,* Edward R. Lee, Richard A. Silva, Andrꢀe J. Bourque, Scott Clark, Sanjeev Katti, and
Vitaly Nivorozhkin
Genzyme Corporation, 153 Second Avenue, Waltham, Massachusetts 02451, United States
S Supporting Information
b
ABSTRACT: An economic, scalable process for the production of glucosylceramide synthase (GCS) inhibitor 7 has been
developed. Herein we report a three-step synthesis to aldehyde 4 with high yield and purity that employs the selective cleavage of an
endocyclic CÀO bond of a THP ether using borane/THF as the key step. This particular methodology has not been used previously
from a development standpoint and offers an attractive way towards introducing pentanol side chains. Aldehyde 4 is then coupled
with deoxynojirimycin via flow hydrogenation using an H-Cube to safely produce the free base of 7, which is isolated as an MSA salt
in 50% overall yield. Herein we discuss the evolution of this process from its original form and the thermodynamics of its associated
chemistry.
’ INTRODUCTION
hydrogenation.10 This change has allowed us to shorten our syn-
thetic route and substantially decrease processing times and
increase throughput. More importantly, it has provided us with
a safe, scalable route for future campaigns. Through these modifi-
cations, we are able to synthesize 7 through a 5-step process. The
synthesis is conveniently carried out in one vessel and affords high
purity material in a 52% overall yield.
Compounds that inhibit GCS have been of pharmacologic
interest for many years due to their potential utility in lowering
glycosphingolipid (GSL) levels. Elevated GSL levels have been
implicated in numerous disorders, including Type II Diabetes,
Gaucher, Sandhoff, TayÀSachs, NiemannÀPick C, and Fabry
diseases, and therefore GCS inhibitors have been investigated as
treatments for enzyme deficiencies.1À7 Iminosugars have been
under increased investigation due to their potential utility for
these conditions, and both Zavesca and Miglitol are currently
marketed therapies which make use of such a core.8 The potent
GCS inhibitor 7 was first reported by Aerts et al. in 1998, where
it was reported to be 100-fold better in inhibiting GCS than
n-butyldeoxynojirimycin.9 Since its discovery, 7 has been shown
to have potential applications in Gaucher’s Disease, Type II Dia-
betes, NiemannÀPick C, ulcerative colitis, and cystic fibrosis.8
While a great amount of literature is available on the potential
applications of 7, there are relatively few economically practical
ways to produce it.4À6 Of the methods that have been previously
published, none seemed to be immediately amenable to scale-up
by conventional economically viable methods. Our goals on this
project were to first establish an economically viable route to
aldehyde 4 and then to optimize the coupling conditions to form
alkylated aza-sugar 6. The aldehyde synthesis was approached
from many angles, first using literature procedures (Schemes 2
and 3), followed by a route employing a ketal intermediate
(Scheme 4). These routes failed due to purity, robustness, and
yield concerns but led us to our final route (Scheme 1) where the
aldehyde is constructed via a selective borane cleavage of a THP
ether followed by catalytic TEMPO oxidation.
’ RESULTS AND DISCUSSION
The synthetic strategy to constructing 7 is to assemble it via
reductive amination from its two building blocks, 5-adamantyl-
methoxy-1-pentanal (4) and 1-deoxynojirimycin (5). Groups
have previously carried out this transformation by coupling 4
with tetrabenzylated 5 through reductive amination, followed by
deprotection.4 While some elegant process work was performed
to carry this out at scale, the route is costly and not particularly
atom efficient. Aminosugars, such as 5, are now made via bio-
transformation11 which has recently allowed them to become
economically viable starting materials at scale. Our process pur-
chases 5 from Bayer AG and permitted us to concentrate on
optimizing the rest of the process.
The second component necessary for the production of 7 is
the aldehyde coupling partner, 4. The original route (Scheme 2)
employs 1,5-pentanediol as a starting material and produces
aldehyde 4 in five steps with an overall yield of 47%.4
There are some issues with running this process at scale.
Formation of tosylate 9 gives a statistical mixture of the desired
product and ditosylated product even when using a limiting
amount of tosyl chloride. This impurity is difficult to remove.
Additionally, tosylate displacement is carried out in DMF using
The reductive amination of aldehyde 4 had historically been
done using a tetrabenzylated aminosugar and catalytic hydro-
genation over Pd on C.4 We felt that this process could be
improved by direct coupling of the aldehyde to the unprotected
sugar and also by switching from batch hydrogenation to flow
Special Issue: Continuous Processing 2012
Received: May 4, 2011
Published: June 26, 2011
r
2011 American Chemical Society
1090
dx.doi.org/10.1021/op2001222 Org. Process Res. Dev. 2012, 16, 1090–1097
|

Organic Process Research & Development
ARTICLE
Scheme 1. Synthetic route to GCS inhibitor 1a
a Reagents and conditions: (a) 3,4-Dihydro-2H-pyran, cat. TsOH, THF, 50 °C, 1 h, used crude. (b) 1 M BH3 in THF, 40 °C, 24 h, used crude.
(c) TEMPO, NaOCl, DCM, aq. NaHCO3, 0À10 °C, 1 h, used crude. (d) THF/H2O, hydrogenated over Pd(OH)2 in a flow reactor. (e) ACN/MeOH,
MSA, then IPA recrystallization; 52% yield over 5 steps.
Scheme 2. Initial route to aldehyde 4a
a Reagents and conditions: (a) TsCl, DMAP, Et3N, CH2Cl2, 16 h, 73%. (b) TEMPO, NaOCl, KBr, CH2Cl2, 9% aq NaHCO3, 5 °C, 2.5 h 91%.
(c) Ethylene glycol, pTsOH, TBME, reflux, 3 h 96.5%. (d) NaH, DMF, 1-adamantanemethanol, 40 °C, 4 h. Product distilled, 80.5%. (e) 6 M HCl,
Acetone, 91%.
sodium hydride, a safety concern given the possibility of a self-
accelerating reaction.12,13 Purification of 12 was necessary;
however distillation was problematic as 1-adamantanemethanol
sublimed and caused blockages during the distillation.
In response to some of the issues associated with the process
outlined above for the manufacture of the aldehyde, an alter-
native route was developed in our laboratories (Scheme 3).6
This route afforded the desired aldehyde in 55% yield over
three steps. This route represents a more expedient way to get the
desired aldehyde but required optimization before large scale
work. Alkylation of 1-adamantanemethanol to afford 13 gave
inconsistent results. Under these reaction conditions, the pre-
dominant reaction taking place is elimination of the 5-bromo-1-
pentene, thus requiring large excesses of the bromide to react all
the alcohol. Phase transfer conditions produced similar results.
Several other protocols were attempted for this substitution;
both strong and weak bases were tried, as well as low and
high temperatures. The problem with the reaction is that 1-
adamantanemethanol is a neopentyl system and is therefore a
poor nucleophile for an SN2 reaction: the product is kinetically
1091
dx.doi.org/10.1021/op2001222 |Org. Process Res. Dev. 2012, 16, 1090–1097

Organic Process Research & Development
ARTICLE
unfavored. Furthermore, column purification was necessary for
13 to achieve acceptable purity of aldehyde 4.
the highest yields of the desired aldehyde. While aldol side
products were still possible, they could be minimized by running
the reaction at temperatures under 50 °C and by moni-
toring reaction progress via HPLC. This route was used to make
500 g of crude aldehyde. This material could be used crude to
form the desired API in acceptable purity or could be isolated as a
pure solid via formation of its bisulfite adduct in aqueous IPA.
While this route was capable of producing high purity material,
the overall yield was low (29%) and the product made via this
route contained aldol impurities that differed from those seen in
the toxological material. As this would have required a bridging
study to validate the new impurities, the route was abandoned. A
key feature of this route was that it introduced the alkyl chain
onto adamantly alcohol in an atom efficient manner. If possible,
we wanted to incorporate this feature into any new route.
One way we thought we might accomplish this was by the
reductive cleavage of a tetrahydropyranyl (THP) ether inter-
mediate. THP ethers are popular protecting groups for primary
alcohols as they are easily installed, stable to most nonacidic
conditions, and easily removed via mild acids. From the litera-
ture, these ethers are more or less inert towards catalytic
hydrogenation and are stable to treatment with both aluminum
hydride and borohydride reagents. Hence, we required prece-
dent for cleaving the endocyclic bond of a THP ether selectively
in order for this route to be successful (Scheme 5).
Few papers have been published on the regiospecific cleavage
of THP ethers, although Lewis acids have been shown to effect
cleavage with varying degrees of endo vs exo selectivity.16,17 The
first to report selective reductive ring cleavage was Cossy.18
Based on this literature, we decided to develop a route to 7 based
on a borane ring opening of a THP ether as shown in Scheme 1.
THP ether 2 was formed by reacting 1-adamantanemethanol
with a slight excess of dihydropyran in THF with catalytic tosic
acid. The reaction proceeds cleanly and can be used for the sub-
sequent step without purification and proceeds quantitatively
based on in-process NMR analysis.
Our initial attempts at a new route for aldehyde 4 were focused
on eliminating the ditosylated impurity in the original synthesis,
using cost-effective reagents, altering the tosylate displacement
conditions to remove DMF, and developing a procedure for
purifying the aldehyde without distillation, which resulted in a
new route (Scheme 4).
Formation of 14 went in modest yield (40À50%) when using
sulfonic acid catalysts and organic media, due to the equilibrium
between the desired ketal and THP ether in the initial step.14 By
running the reaction in aqueous acid, we favor the formation of
the desired ketal as a 2:1 mixture, vs a 1:1 mixture seen in organic
systems. While this has been previously observed for tetrahy-
drofuranyl reactions,15 it was previously unreported for these
ketalizations. Although the yield was low, 14 could be isolated in
>98% purity via crystallization from aqueous IPA. This crystal-
lization solvent offered the additional advantage of destroying
unreacted TsCl. The alkylation of this product went in high yield
in THF, giving 15 in yields of 98% or greater. Problems arose
when we attempted to deprotect the ketal to form 4. The use of
very strong acids (e.g., sulfuric acid, triflic acid) was successful in
removing the protecting group but also began to promote aldol
side reactions and discolored the reaction mixture immediately
upon addition. After an extensive screen, formic acid offered the
most economical solution to the problem, as well as producing
Scheme 3. First generation process for aldehyde 4a
Reduction of ether 2 to alcohol 3 via reaction with a borane/
THF complex was successful but required safety investigation
prior to scaleup. The dangers of working with borane/THF
complexes have been well documented.19 While the pro-
blems with this reagent typically occur with storage of the raw
material, the fact that it is a highly reactive substance that
can become unstable upon prolonged heating led to several pre-
cautions. The borane/THF complex used was stabilized with
a Reagents and conditions: (a) 5-bromo-1-pentene, NaOH, DMSO,
70 °C, 16 h, 65.4%. (b) (1) 9-BBN, diethyl ether, THF, rt, 16 h. (2) aq
NaOH, 30% H2O2, rt, 3 h, 89%. (c) (1) (COCl)2, DMSO, CH2Cl2,
À70 °C, 30 min. (2) Et3N, warm to rt, 95%.
Scheme 4. Synthesis of 4 via dihydropyrana
a Reagents and conditions: (a) (1) neopentyl glycol, 1 M HCl, 50 °C, 2 h. (2) CH2Cl2, Et3N, TsCl, 0 °C to rt, 2 h, 50% over 2 steps.
(b) 1-Adamantanemethanol, NaH, THF, reflux, 16 h, 98%. (c) Formic acid, 45 °C, 3 h, 60%.
1092
dx.doi.org/10.1021/op2001222 |Org. Process Res. Dev. 2012, 16, 1090–1097

Organic Process Research & Development
ARTICLE
N-isopropyl-N-methyl-tert-butylamine, which has been shown
to be much more stable than either unstabilized borane/THF
or borane/THF stabilized with sodium borohydride.20 Of parti-
cular importance to us was that the stabilized borane/THF has
a reported self-accelerated decomposition temperature (SADT)
threshold of >50 °C.20 This is a measure of the lowest tem-
perature at which a self-accelerating reaction can begin to occur
by itself. Unstabilized borane/THF solutions or solutions
stabilized with NaBH4 have SADT thresholds at or just above
40 °C.20
Our initial efforts focused on the kinetics of this reaction.
Based on the reaction mechanism, we reasoned that given
enough time, the reaction would proceed to completion if 0.33
equiv of borane/THF complex was charged. When the reaction
was run with this many equivalents at 40 °C, we observed a kinetic
trace that was reminiscent of second-order kinetics. We reason
that, after a relatively fast formation of an initial alkoxyborane
species, subsequent reactions are slower due to the decreased
Lewis acidity of the borane center—as we have previously seen,
coordination is essential to this ring-opening process. While the
reaction would run to completion with as few as 0.7 equiv of
borane/THF complex, we were interested in having the reaction
complete within a 24 h period, to minimize borane degradation.
Although the borane is stabilized, it undergoes ∼10% degradation
after 48 h at 48 °C.20 While complete conversion was observed for
all amounts of borane/THF g 0.7 equiv, 2 equiv of the borane/
THF complex at 40 °C for 16 h gave complete conversion of THP
ether 2 to alcohol 3. The use of 2 equiv of borane was chosen
because the reaction was consistently complete in 16 h, whereas
lower amounts had greater variation in reaction time.
A series of calorimetric experiments were run using a Mettler
Toledo RC-1 Reaction Calorimeter to quantify the process safety
parameters. The experiments were used to investigate three
operations within the reaction:
Scheme 5. THP ether reactivity
1 The heat rise and thermal behavior of borane being dosed
into the reaction
2 The heat flow during the reaction, with attention being paid
to any thermal accumulation
3 The heat rise, thermal behavior, and gas evolution during
the reaction quench.
While addition of borane to the reaction mixture was an
exothermic process, the heat generated was dosing controlled.
Figure 1. Calorimetric data for the borane quench. The heat flow is shown in pink, the jacket temperature in brown, and the mass of reaction in gray.
Based on these data, the adiabatic temperature rise for the operation was 32.24 °C.
1093
dx.doi.org/10.1021/op2001222 |Org. Process Res. Dev. 2012, 16, 1090–1097

Organic Process Research & Development
ARTICLE
Figure 2. H-Cube MIDI flow hydrogenator.
Because of this, the reaction was dosed at 0 °C, followed by
ramping the temperature over 2 h to 40 °C, where it was held for
16 h. During the hold point, there was no noticeable thermal
accumulation taking place during the reduction. Once the
reaction was deemed complete by either HPLC or NMR analysis,
it was cooled to 0 °C and quenched with 1 M HCl. While the
quench of the reaction was exothermic (adiabatic temperature
rise of 32.24 °C), the heat rise was dosing controlled (see
Figure 1). Furthermore, while off-gassing was observed, it was
also dosing controlled and did not lead to foaming or bumping.
With these data in hand, we felt confident that the reaction could
be safely scaled.
Once a safe process for producing 3 was developed, we could
easily convert it to aldehyde 4 via a TEMPO oxidation.21 The
resultant aldehyde was isolated crude after a solvent extraction
workup and could be used “as is” for the final reductive ami-
nation. Overall, the yield for aldehyde production has been al-
most doubled from previous attempts to an average yield of 85%
and now uses inexpensive materials throughout. The process to
make the aldehyde has been run on multiple occasions with
consistent yields and purity from batch to batch, signifying pro-
cess robustness.
Developing a route to couple 4 with DNJ (5) through
reductive amination proved to be surprisingly difficult. Due to
the water solubility of penultimate product 6, reductive amina-
tions using borohydrides or transfer hydrogenation conditions
were not suitable. Catalytic hydrogenation of the imine seemed
promising and worked well at small scale. However, the through-
put of this reaction was limited by the pressure vessels on hand.
Flow hydrogenation addressed the issue. From a safety stand-
point, the reaction volume in a flow system at any moment in
time is very small (∼ 8 mL in our system) and minimizes the
danger from any unexpected event. From a throughput
standpoint, we are no longer limited by batch size, as batches
can be continuously processed in a flow system. Additionally,
flow hydrogenation eliminates mass transfer limitations, which
can cause reactions to run slowly at increasing scale, and can also
lead to the formation of side products due to increased
reaction time.
While there were several options available for kilo-lab scale
flow systems, the H-Cube MIDI22 worked well for us and ad-
dressed our safety concerns. The apparatus is fully explosion
proof, could integrate into our kilo-lab facilities without issue,
and generates hydrogen in situ via the electrolysis of water elimi-
nating the need for high pressure hydrogen tanks. It is also
capable of flow rates between 3 and 25 mL/min, using tempera-
tures between ambient and 150 °C, and generating up to 100 bar
of hydrogen pressure (1450 psi) (Figure 2).
A particular challenge to running this reaction under flow
conditions was the selection of an appropriate solvent. Initially,
we were running these aminations in batch mode using IPA as a
solvent, as the final salt could be isolated directly once amination
was complete. Flow chemistry holds many advantages over batch
chemistry but cannot work unless all the materials in the reaction
are completely soluble. DNJ has only moderate solubility in IPA,
and while DNJ exhibited high solubility in water, aldehyde 4
was not miscible with the resultant solution. Acetic acid initially
offered itself as a replacement, as it is widely used in processes;
both DNJ and the aldehyde were readily soluble in it, and the
reaction ran very nicely in it (the reaction ran to full conversion
at lower temperatures and pressures than the solvent system
ultimately selected). However, the issue with acetic acid was
that the resultant acetate salt was difficult to free base and
isolate in order to convert to the MSA salt. Additionally,
attempting a salt exchange failed to give crystalline material.
Based on these concerns, we switched to a mixed solvent system
1094
dx.doi.org/10.1021/op2001222 |Org. Process Res. Dev. 2012, 16, 1090–1097

Organic Process Research & Development
ARTICLE
Figure 3. Percent conversion of starting material to product vs flow rate: The reductive amination was run at 150 °C, under 100 bar of H2 using a
600 mmol/L reaction solution.
of 2:2:1 MeOH/THF/H2O. This system was capable of keeping
all the reagents in solution at rt at a concentration of 600 mmol/L
DNJ. More concentrated solutions would result in crystallization
of the starting materials during the flow reaction. Adding more
water to the mixture to increase DNJ solubility was not an option,
as it caused the aldehyde to become immiscible.
Once the solubility issues of the starting materials was solved,
we focused on developing flow conditions that would maximize
our throughput for a single pass hydrogenation. For a given
catalyst, many reaction conditions can be screened quickly by
varying flow rate and temperature. We undertook a catalyst screen
to determine which catalysts were best suited for this chemistry,
with a particular focus on reaction throughput. A plot of reaction
conversion vs flow rate is shown in Figure 3.
From Figure 3, it is clear that the Pd based catalysts perform
the best with increasing flow rate leading to the decision to use
Pd(OH)2 on C as our catalyst. It is interesting to note that when
the above experiments were done in AcOH, conversion was
consistently higher at lower pressures and higher flow rates. The
stoichiometry of the reaction turned out to be critical for our final
impurity profile. While the reaction would run to >98% conver-
sion using equimolar amounts of aldehyde and iminosugar, it
turned out to be more or less impossible to remove unreacted
DNJ from our reaction mixture. Despite being more hydropho-
bic than DNJ, free base 6 was still very water-soluble, making
washing out residual DNJ through solvent extraction unachie-
vable. Additionally, forming MSA salt 7 did nothing to lower
residual amounts of DNJ. Initially, experiments were performed
using purified aldehyde and DNJ to determine the amount of
aldehyde needed to consume >99.8% of DNJ during a single pass.
These experiments showed that 1.6 equiv of aldehyde would
consistently reach this level. Based on an 85% yield of aldehyde 4
from the starting material, 0.53 equiv of DNJ could be used
relative to the amount of 1-adamantanemethanol started with.
The optimized conditions were to dissolve 4 and 5 in a mixture
of THF/MeOH/H2O and then to hydrogenate at 100 bar and
150 °C, at a flow rate of 15 mL/min. The reaction was complete
in a single pass at this rate. Based on the flow rate and the molarity
of the solution, 1 kg of this material per day could be processed.
To date, we have piloted this chemistry several times on 100 g
scale with consistent results but have no plans to scale it larger
due to a change in project focus. We routinely employ flow
methodologies on scale, not only for reductive aminations but
also for the reduction of other moieties.
The crude reaction mixture containing 6 was solvent swapped
into a mixture of MeOH and ACN, and the mesic acid salt
formed via addition of MSA. This salt was recrystallized from IPA
to give the final material, in 99.4% purity.
’ CONCLUSION
A scalable, economic, and novel route to GCS inhibitor 7 was
developed. The key step in the current route is a borane mediated
endocyclic CÀO bond cleavage of a THP ether. Because of this
step, we were able to increase the yield of building block 4
substantially and were able to use it to make suitable material for
clinical studies. Reaction calorimetry was used to confirm the
safety and scalability of all steps, but particularly of the borane
1095
dx.doi.org/10.1021/op2001222 |Org. Process Res. Dev. 2012, 16, 1090–1097

Organic Process Research & Development
ARTICLE
reduction. The use of flow hydrogenation was key for preparing 7
in a rapid manner, which increased both throughput and safety.
the reaction temperature did not rise above 10 °C at any point.
Once dosing was complete, reaction completion was confirmed
with NMR analysis. Excess oxidant was destroyed by addition of
10% aq sodium thiosulfate (600 mL), and the organic layer was
isolated. The organic layer was washed with 1 M HCl (500 mL),
sat. brine (500 mL), and water (500 mL). A 10 mL sample was
taken and concentrated to dryness to get an estimated yield and
an analysis. Based on the results, approximately 170 g (85% yield
over three steps) of aldehyde were formed over the three-step
synthesis. The aldehyde was left in solution in order to do the
final coupling. 1H NMR (400 MHz, CDCl3) 9.78 (s, 1H), 3.41
(t, 2 H), 2.98 (s, 2H), 2.44 (t, 2H), 1.98 (m, 3H), 1.45À1.80
(m, 16 H). 13C NMR (100 MHz, CDCl3) 202.88, 82.11, 71.09,
43.88, 39.89, 37.40, 34.18, 29.18, 28.43, 19.19.
’ EXPERIMENTAL SECTION
General. 1-adamantanemethanol was obtained from Carbo-
synth Ltd. (Berkshire, UK), and deoxynojirimycin was obtained
from Bayer AG (Leverkusen, Germany). Borane/THF complex
(1.0 M) stabilized with N-isopropyl-N-methyl-tert-butylamine
was purchased from Aldrich and used fresh. Due to the weak
absorbing nature of the compound, LC purity determination was
performed using an Agilent 1100 HPLC equipped with a Corona
CAD detector. All reactions were carried out in an inert atmo-
sphere using nitrogen as the blanket gas.
1-Adamantanemethanol Tetrahydropyranyl Ether (2). A
solution of 1-adamantane methanol (1, 132.8 g, 0.8 mol) and
tosic acid (3.04 g, 0.016 mol) in THF (400 mL) was heated to
45 °C under nitrogen. Once at temperature, 3,4-dihydro-2H-
pyran (73.4 mL, 0.8 mol) was added via addition funnel at a rate
such that the temperature was maintained within 0.5 °C of the set
point. Once addition was complete, the reaction was heated to
55 °C for 2 h. The reaction was cooled to rt and sampled for
analysis. The starting material was converted to product in excess
of 95% by NMR analysis and was used for subsequent steps
without further purification. A 2 mL sample was concentrated
under reduced pressure for analysis. 1H NMR (400 MHz,
CDCl3) 4.52 (t, 1H), 3.87 (m, 1H), 3.78 (m, 1H), 3.45 (m,
1H), 3.38 (d, 1H), 2.90 (d, 1H), 1.4À2.0 (m, 20 H). 13C (100
MHz, CDCl3) 98.99, 78.18, 61.75, 39.71, 37.26, 33.72, 30.60,
33.72, 28.31, 25.61, 19.45.
N-[5-(Adamantan-1-yl-methoxy)pentyl]-1-deoxynojirimycin
Methanesulfonic Acid Salt (7). The aldehyde solution from
the previous step was concentrated to a minimum stir volume
(120 mL), and 140 mL of THF and 280 mL of MeOH were
added. In a separate flask, 70 g of deoxynojirimycin (0.428 mol)
were dissolved in 280 mL of DI H2O and added to the aldehyde
solution. The solution was briefly stirred to ensure dissolution
and then polish filtered through a 0.22 μm filter. The reaction
solution was then passed through the H-Cube MIDI flow hydro-
genator at a flow rate of 15 mL/min, over a 20% Pd(OH)2 on a C
catalyst bed, at 145 °C under 95 bar of hydrogen pressure. The
reaction was complete after a single pass through the apparatus.
After hydrogenation, the reaction mixture was charged to a 5 L
three-neck round-bottom flask, equipped with an overhead
stirrer, nitrogen line, and thermocouple. The reaction mixture
was distilled to a minimum stir volume (200 mL), and toluene
(1 L) was charged. The reaction was again distilled to a minimum
stir volume (200 mL). The reaction mixture was dissolved in
1.5 L of 30% methanol in ACN and was brought to rt. Methane-
sulfonic acid (28 mL) was added to the reaction mixture over 1 h.
The reaction mixture was heated to 50 °C and allowed to cool to
rt over 1 h. The reaction was cooled to 0 °C, stirred for 1 h, and
filtered. The filtered solids were suspended in 700 mL of iso-
propanol and dissolved at 50 °C. The reaction was allowed to
cool to rt and stir for 2 h. Filtration followed by drying under
vacuum at 60 °C to a constant weight afforded 130.7 g of the
5-(Adamant-1-yl-methoxy)-1-pentanol (3). The solution of
2 isolated in the previous step was cooled to 5 °C, and 1.6 L of
1 M borane/THF complex in THF (1.6 mol) were slowly added
to the solution via addition funnel over 1 h. Addition was per-
formed slowly in order to prevent any thermal accumulation and
also control the amount of off-gassing observed. Once addi-
tion was complete, the reaction was warmed to rt over 1 h and
then heated at 45 °C for 18 h. The reaction progress was checked
via 1H NMR and was found to be complete. The reaction mixture
was cooled to 5 °C and was quenched by 300 mL of 1 M HCl via
addition funnel over 1 h. Once addition was complete, the re-
action was warmed to 55 °C for 1 h, to ensure destruction of any
borane complexes, and was cooled to rt. The reaction mixture
was diluted with 500 mL of brine, and the organic layer was
isolated. The organic layer was charged to a clean reaction flask,
and 1.8 L of solvent were distilled off. Methylene chloride was
charged to the flask (2 L), and 1 L of solvent was distilled off.
The reaction was filtered to remove a small amount of pre-
cipitated boric acid, and the filtrate was charged to a clean flask
for the subsequent step. A 2 mL sample was taken and
1
desired compound as a white solid (Yield 61.5%). H NMR
(400 MHz, DMSO-d6) 9.26 (s, 1H, from MSA), 5.51 (m, 4H
from OH groups), 3.92 (m, 1H), 3.78 (m, 1H), 3.58 (m, 1H),
3.32À2.92 (m, 11H), 2.39 (s, 3H from mesic acid), 1.91 (m, 3H),
1.68À1.30 (m, 18H). 13C NMR (100 MHz, DMSO-d6) δ 81.01,
76.38, 70.38, 67.19, 66.03, 65.36, 53.96, 53.33, 51.99, 39.63
(under DMSO peak, resolves w/APT), 39.23 (under DMSO
peak, resolves with APT), 36.71, 33.60, 38.54, 37.59, 22.92,
21.96. DSC, 182.58 °C, 95.86 J/g (melting point). TGA, 0%
LOD up to 191.3 °C, at which point decomposition occurs.
AUC Purity (HPLC, Corona CAD Detector) 99.4%. This
material contained 740 ppm of residual isopropanol, which
meets ICH guidelines. No other solvents were observed via
GC headspace analysis. Less than 1 ppm of residual Pd was
detected via ICP MS.
1
concentrated under reduced pressure for analysis. H NMR
(400 MHz, CDCl3) 3.63 (t, 2H), 3.40 (t, 2H), 2.95 (s, 2 H), 1.98
(m, 3 H), 1.4À1.8 (m, 19 H). 13C NMR (100 MHz, CDCl3)
81.89, 71.57, 62.77, 39.74, 37.25, 34.09, 32.48, 29.29,
28.30, 22.42.
5-(Adamant-1-yl-methoxy)-1-pentanal (4). The solution of
alcohol 3 from the previous step was cooled to 5 °C, and a solu-
tion of K2CO3 (14.4 g, 0.104 mol), NaHCOs (87.3 g, 1.04 mol),
and KBr (0.95 g, 0.08 mol) in 1.6 L of H2O was charged. TEMPO
(2.5 g, 0.16 mol) was charged to the stirred reaction mixture. The
reaction was then dosed with 10% sodium hypochlorite solution
(600 mL) over 2 h. The reaction was dosed at a rate such that
’ ASSOCIATED CONTENT
Supporting Information. 1H, 13C, APT NMR spectra,
S
b
HPLC data, DSC and TGA data. This material is available free of
charge via the Internet at http://pubs.acs.org.
1096
dx.doi.org/10.1021/op2001222 |Org. Process Res. Dev. 2012, 16, 1090–1097

Organic Process Research & Development
ARTICLE
’ AUTHOR INFORMATION
Corresponding Author
*E-mail: christopher.cooper@genzyme.com.
’ REFERENCES
(1) Wennekes, T.; Lang, B.; Leeman, M.; van der Marel, G. A.;
Weber, M.; van Wiltenburg, J.; Wolberg, M.; Aerts, J. M.; Overkleeft, H.
S. Org. Proc. Res. Dev. 2008, 12, 414.
(2) Aerts, J. M.; Ottenhoff, R.; Powlson, S.; Grefhorst, A.; van Eijk,
M.; Dubbelhuis, P. F.; Aten, J.; Kuipers, F.; Serlie, M. J.; Wennekes, T.;
Sethi, J. K.; O'Rahilly, S.; Overkleeft, H. S. Diabetes 2007, 55, 1341.
(3) Hirth, B.; Rennie, G. PCT Int. Appl. 2005, WO 2005063706.
(4) Jeyakumar, M.; Norflus, F.; Tifft, C. J.; Cortina-Borja, M.;
Butters, T. D. Blood 2001, 97, 327.
(5) Moran, M. T.; Schofield, J. P.; Hayman, A. R.; Gou-Ping, S.;
Young, E.; Cox, T. M. Blood 2000, 96, 1969.
(6) Zhao, H.; Przybylska, M.; Wu, I. H.; Zhang, J.; Siegel, C.;
Komarnitsky, S.; Yew, N. S.; Cheng, S. H. Diabetes 2007, 56, 1210.
(7) Shayman, J. A.; Radin, N. S. Amino Ceramide-Like Compounds
and Methods of Use. US Patent 6,255,336, July 9, 1999.
(8) Cox, T.; Platt, F.; Aerts, J. Iminosugars 2007, 295.
(9) Overkleeft, H.; Renkama, G.; Neele, J.; Vianello, P.; Hung, I.;
Strijland, A.; van der Burg, A.; Koomen, G.; Pandit, U.; Aerts, J. J. Biol.
Chem. 1998, 273, 26522.
(10) Ley, S. V.; Knudson, K. R.; Ladlow, M.; Saaby, S. Chem.
Commun. 2005, 2909.
(11) Kinast, G.; Schedel, M. (Bayer AG). 6-Amino-6-deoxy-L-
sorbose. Ger Offen. DE 2834122, February 14, 1980.
(12) Buckley, J. Chem. Eng. News 1982, 60 (28), 5.
(13) De Wall, G. Chem. Eng. News 1982, 60 (37), 43.
(14) Bartels, G. (Hoechst AG). Simplified preparation of 2-(3-
bromopropyl)- and 2-(4-bromobutyl)-5,5-dimethyl-1,3-dioxane Ger.
Offen. DE 3537289, February 23, 1987.
(15) Stowell, J.; Polito, M. J. Org. Chem. 1992, 57, 2195.
(16) Eliel, E.; Nowak, B.; Daignault, R. J. Org. Chem. 1965, 30, 2448.
(17) Guindon, Y.; Bernstein, M.; Anderson, P. Tetrahedron Lett.
1987, 28, 2225.
(18) Cossy, J.; Bellosta, V.; Muller, M. C. Tetrahedron Lett. 1992,
33, 5045.
(19) Urben, P. G., Ed. Brethericks’s Handbook of Reactive Chemical
Hazards, 6th ed.; Butterworth Heinemann: Boston, 1999.
(20) See Aldrich Technical Bulletin AL-218, available online at
http://www.sigmaaldrich.com.
(21) For a review, see: Ciriminna, R.; Pagliaro, M. Org. Process Res.
Dev. 2010, 14, 245.
(22) Information and schematics available through http://www.
thalenano.com/products/h-cube-midi.
’ NOTE ADDED AFTER ASAP PUBLICATION
This article was published on the Web on July 5, 2011. Omitted
references have been added and a misspelling of an author's name
has been updated. The corrected version was reposted on August
22, 2011. An additional reference was added; the corrected
version was reposted on December 16, 2011.
1097
dx.doi.org/10.1021/op2001222 |Org. Process Res. Dev. 2012, 16, 1090–1097