An efficient scalable route for the synthesis of enantiomerically pure tert -butyl-(1 R,4 S,6 R)-4-(hydroxymethyl)-3-azabicyclo[4.1.0]heptane-3-carboxylate

Article abstract of DOI:10.1021/op100164v

An efficient scalable route to synthesize the enantiomerically pure tert-butyl-(1R,4S,6R)-4-(hydroxymethyl)-3-azabicyclo[4.1.0]heptane-3- carboxylate is described. Compared to the original routes, significant improvements were made by using an innovative approach starting from commercially available chiral lactone. In this approach, one of the key steps described is an elegant epimerization/hydrolysis of the undesired diastereoisomer avoiding tedious purification. The chemistry has been scaled up to produce kilogram amounts of tert-butyl-(1R,4S,6R)-4-(hydroxymethyl)-3- azabicyclo[4.1.0]heptane-3-carboxylate in 43% yield over nine chemical transformations.

Full text of DOI:10.1021/op100164v

Organic Process Research & Development 2010, 14, 1239–1247
An Efficient Scalable Route for the Synthesis of Enantiomerically Pure
tert-Butyl-(1R,4S,6R)-4-(hydroxymethyl)-3-azabicyclo[4.1.0]heptane-3-carboxylate
William M. Maton,†,* Federica Stazi,† Angelo Maria Manzo,† Roberta Pachera,† Arianna Ribecai,† Paolo Stabile,†
Alcide Perboni,† Nicola Giubellina,† Fernando Bravo,† Damiano Castoldi,† Stefano Provera,‡ Lucilla Turco,‡ Simon Bryant,†
Pieter Westerduin,† Roberto Profeta,§ Arnaldo Nalin,§ Emanuele Miserazzi,§ Simone Spada,§ Anna Mingardi,§ Mario Mattioli,§
and Daniele Andreotti§
Chemical DeVelopment, Analytical Chemistry, and Neurosciences Synthetic Chemistry CEDD, GlaxoSmithKline Medicine
Research Centre, Via A. Fleming 4, 37135 Verona, Italy
Abstract:
antagonists in rats3 and more recently dogs and humans further
support this. Therefore a substantial amount of work has been
reported by our Medicinal Chemistry colleagues,4 as well as
by others, aimed at identifying new chemical entities 1
possessing orexin antagonist activity. Due to the promising
activity of compounds 1 as an orexin antagonist and the
complexity of the molecules bearing the same structural motif,
relatively large amounts for several of these compounds were
requested for toxicological testing and clinical evaluation.
An efficient scalable route to synthesize the enantiomerically pure
tert-butyl-(1R,4S,6R)-4-(hydroxymethyl)-3-azabicyclo[4.1.0]heptane-
3-carboxylate is described. Compared to the original routes,
significant improvements were made by using an innovative
approach starting from commercially available chiral lactone. In
this approach, one of the key steps described is an elegant
epimerization/hydrolysis of the undesired diastereoisomer avoiding
tedious purification. The chemistry has been scaled up to produce
kilogram amounts of tert-butyl-(1R,4S,6R)-4-(hydroxymethyl)-3-
azabicyclo[4.1.0]heptane-3-carboxylate in 43% yield over nine
chemical transformations.
Introduction
The orexin ligand and receptor system have been well
characterized since their discovery.1 From these studies it has
become clear that orexin receptors play a number of important
physiological roles in mammals and open up new avenues for
therapeutic treatments for a variety of diseases and CNS
disorders. Also, the role of the orexin system in sleep and
wakefulness is now well established.2 The orexin receptors
antagonists may therefore be useful in the treatment of sleep
disorders including insomnia. Studies with orexin receptor
Structurally, the complex target molecules 1 consist of
constraint piperidine moiety bearing a methanamino
heterocyclic subunit and a carboxylic acid linked via an
amide bond (Scheme 1).
This paper describes the development of an efficient scalable
synthesis of 3-azabicyclo[4.1.0]heptane ring system, 4, key
intermediate to reach the targeted molecules, 1.
Background
The Medicinal Chemistry approach to synthesize the can-
didate molecules, 1, is briefly described in Scheme 2. Com-
mercially available (2S)-2-amino-4-pentenoic acid, 5, was
reduced with lithium aluminum hydride, and the amino group
was then selectively protected as tosyl amide by reaction with
tosyl chloride to provide 6. Subsequent TBDPS protection of
the primary alcohol and allylation of tosylamide using allyl
bromide and cesium carbonate in DMF afforded 7. Ring-closing
metathesis (RCM)5 of 7 in the presence of Grubbs I catalyst6
in dichloromethane gave the dehydropiperidine 8 in good yield.
The next step involved Simmons-Smith cyclopropanation as
the key reaction to reach the desired [4.1.0] moiety using the
modified conditions described by Yang et al.7 In fact, reaction
of olefin 8 in the presence of diethylzinc, trifluoroacetic acid,
and diiodomethane led to the formation of the desired compound
* To whom correspondence should be addressed. Email: william.maton@
gsk.com.
† Chemical Development.
‡ Analytical Chemistry.
§ Neurosciences Synthetic Chemistry CEDD.
(1) (a) Sakurai, T.; Amemiya, A.; Ishii, M.; Matsuzaki, I.; Chemelli, R. M.;
Tanaka, H.; Williams, S. C.; Richardson, J. A.; Kozlowski, G. P.;
Wilson, S.; Arch, J. R. S.; Buckingham, R. E.; Haynes, A. C.; Carr,
S. A.; Annan, R. S.; McNulty, D. E.; Liu, W. S.; Terrett, J. A.;
Elshourbagy, N. A.; Bergsma, D. J.; Yanagisawa, M. Cell 1998, 92,
573–585. (b) Smart, D.; Jerman, J. C.; Brough, S. J.; Rushton, S. L.;
Murdock, P. R.; Jewitt, F.; Elshourbagy, N. A.; Ellis, C. E.;
Middlemiss, D. N.; Brown, F. Br. J. Pharmacol. 1999, 128, 1–3. (c)
Willie, J. T.; Chemelli, R. M.; Sinton, C. M.; Yanagisawa, M. Annu.
ReV. Neurosci. 2001, 24, 429–458. (d) Sakurai, T. Nat. ReV. Neurosci.
2007, 8, 171–181. (e) Ohno, K.; Sakurai, T. Front. Neuroendocrinol.
2008, 29, 70–87.
(2) (a) Chemelli, R. M.; Willie, J. T.; Sinton, C. M.; Elmquist, J. K.;
Scammell, T.; Lee, C.; Richardson, J. A.; Williams, S. C.; Xiong, Y.;
Kisanuki, Y.; Fitch, T. E.; Nakazato, M.; Hammer, R. E.; Saper, C. B.;
Yanagisawa, M. Cell 1999, 98, 437–451. (b) Lee, M. G.; Hassani,
O. K.; Jones, B. E. J. Neurosci. 2005, 25, 6716–6720. (c) Piper, D. C.;
Upton, N.; Smith, M. I.; Hunter, A. J. Eur. J. Neurosci. 2000, 12,
726–730. (d) Smart, D.; Jerman, J. C. Pharmacol. Ther. 2002, 94,
51–61.
(3) Smith, M. I.; Piper, D. C.; Duxon, M. S.; Upton, N. Neurosci. Lett.
2003, 341, 256–258.
(4) Amantini, D.; Alvaro, G.; Di Fabio, R.; Castiglioni, E.; Pavone, F.
PCT/EP2009/066017, 2009.
10.1021/op100164v  2010 American Chemical Society
Published on Web 08/25/2010
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1239

Scheme 1. Retrosynthetic approach to molecule 1
Scheme 2. Medicinal chemistry routea
a Reagents and conditions: a) LiAlH4, THF; b) TsCl, Na2CO3, H2O/EtOAc/THF; c) TBDPSCl, imidazole, DMF; d) allyl bromide, Cs2CO3, DMF; e) Grubbs I,
DCM; f) Et2Zn/TFA/CH2I2, DCM; g) TBAF, THF; h) Dess-Martin, DCM; i) Amino-Het, NaBH(OAc)3, AcOH, THF; j) Na, naphthalene, THF, -78 °C; k) TBTU,
DIPEA, DCM (24% overall yield).
9 as a single diastereoisomer. Despite an excess of reagents
Scheme 3. Rhodium-catalyzed cyclopropanation
(8-20 equiv) that was necessary to ensure full conversion of
the olefin, only a modest yield (∼70%) was obtained. After
TBDPS deprotection, alcohol 10 was oxidized to the corre-
sponding aldehyde 11 using Dess-Martin reagent. Reductive
amination of aldehyde 11 with sodium triacetoxyborohydride
in the presence of a substituted amino heterocyclic compound
gave 12 in moderate yield. The tosyl group cleavage of 12 by
sodium naphthalenide in THF followed by amide coupling with
the carboxylic acids 2, using TBTU as coupling reagent,
afforded 1.
retrosynthetic route
fied several key issues. In particular, Simmons-Smith cyclo-
propanation required an excess of reagents (up to 20 equiv),
and the yield was not reproducible. Furthermore, the removal
of the tosyl group and the use of Dess-Martin were not
practically suitable for a scale-up.
Despite the excellent diastereoselectivity during the cyclo-
propanation, the Medicinal Chemistry Support chemists identi-
(5) For reviews on RCM, see: (a) Samojłowicz, C.; Bieniek, M.; Grela,
K. Chem. ReV. 2009, 109, 3708–3742. (b) Vougioukalakis, G. C.;
Grubbs, R. H. Chem. ReV. 2010, 110, 1746–1787. (c) Fu¨rstner, A.
Angew. Chem., Int. Ed. 2000, 39, 3012–3043. (d) Maier, M. E. Angew.
Chem., Int. Ed. 2000, 39, 2073–2077. (e) Armstrong, S. K. J. Chem.
Soc., Perkin Trans. 1 1998, 371–388. (f) Grubbs, R. H.; Chang, S.
Tetrahedron 1998, 54, 4413–4450.
Medicinal Chemistry Support Synthesis. The Medicinal
Chemistry Support chemists decided to investigate alternative
routes to access to the [4.1.0] system. The new diastereoselective
synthesis was based on intramolecular rhodium-catalyzed cy-
clopropanation of intermediate 14 (Scheme 3).
As shown in Scheme 4, the allyl diazoamide 14 is con-
structed again from (2S)-2-amino-4-pentenoic acid 5. Reduction
of 5 with lithium aluminum hydride and subsequent amidation
(6) Schwab, P.; France, M. B.; Ziller, J. W.; Grubbs, R. H. Angew. Chem.,
Int. Ed. Engl. 1995, 34, 2039–2041.
(7) Yang, Z.; Lorenz, J. C.; Shi, Y. Tetrahedron Lett. 1998, 39, 8621–
8624.
1240
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development

Scheme 4. Medicinal chemistry support routea
a Reagents and conditions: a) LiAlH4, THF; b) 16, THF/MeOH; c) 2,2-dimethoxypropane, pTSA, toluene; d) TMSOTf, DCM, 2,6-lutidine; e) NaNO2, DCM/buffer
pH ) 5; f) Rh2(OAc)4, DCM; g) HCl; h) BH3 in THF, reflux, then Boc2O, NaOH (6% overall yield).
Scheme 5. Chemical development retrosynthetic approach
of the resulting primary amine 15 with the commercially
available Boc-Gly-OSu 16 gave 17. To have more constraint
on the carboskeleton in the key step and a relative stable
protective group, the amido alcohol 17 was protected as
acetonide by reaction with 2,2-methoxypropane in the presence
of a catalytic amount of p-toluenesulfonic acid, affording 18 in
good yield after column chromatography. After Boc deprotec-
tion, the amine 19 was converted into the corresponding
diazoamide 20. Diazoamide 20 was treated with Rh2(OAc)4 in
dichloromethane at room temperature to afford a mixture of
21 and 22 in moderate yield in 1:1 ratio.
Interestingly, switching the catalyst in the above condition
to Rh2(5S-MEPY)4 or Rh2(5R-MEPY)4 did not give improve-
ment on the diastereoselectivity of the reaction. In fact, both
catalysts afforded mainly the undesired diastereoisomer 21 (ratio
21/22 7:3).
After acetonide removal, reduction of 23 with borane at
reflux followed by protection of the resulting secondary amine
with tert-butoxycarbonyl group and purification by column
chromatography provided the key scaffold 24.
As evidenced in both routes, there were several key issues
to be considered before envisaging a scale-up process:
(i) Safety issues: use of excess of reagents in Simmons-Smith
cyclopropanation (possible problem during the work-up), use
of sodium naphthalenide at low temperature (note: other amino
protective group have been tried such as Boc, Cbz, but no
reaction took place under cyclopropanation conditions), and
hazards associated with diazochemistry.
(ii) Several chromatographic purifications were required.
(iii) Low overall yield (Medicinal Chemistry route 24%, and
Medicinal Chemistry Support 6%).
Therefore, a more efficient scalable synthesis for 4 was
necessary.
Chemical Development Synthesis. Considering the dif-
ficulty to form the cyclopropyl moiety, we sought to focus our
retrosynthetic approach on a chiral cyclopropyl starting material.
Therefore, a new scalable route was identified, which involved
an alkylation of glycine derivative 26 with alkyl iodide 27
obtained in two steps from commercially available chiral
lactone8 28 as depicted in Scheme 5.
The synthesis started from the opening of chiral lactone 28
(>99% ee) with TMSI,9 generated in situ using trimethylsilyl-
chloride (TMSCl) and sodium iodide in acetonitrile at 50 °C
(8) Chiral lactone 28 can be purchased from Minakem or Dr. Reddys.
For the syntheses of 28 see: (a) Sabbioni, G.; Jones, J. B. J. Org.
Chem. 1987, 52, 4565–4570. (b) Bolm, C.; Schiffers, I.; Dinter, C. L.;
Gerlach, A. J. Org. Chem. 2000, 65, 6984–6991. (c) Peschiulli, A.;
Gun’ko, Y.; Connon, S. J. J. Org. Chem. 2008, 73, 2454–2457. (d)
Doyle, M. P.; Austin, R. E.; Bailey, A. S.; Dwyer, M. P.; Dyatkin,
A. B.; Kalinin, A. V.; Kwan, M. M. Y.; Liras, S.; Oalmann, C. J.;
Pieters, R. J.; Protopopova, M. N.; Raab, C. E.; Roos, G. H. P.; Zhou,
Q.-L.; Martin, S. F. J. Am. Chem. Soc. 1995, 117, 5763–5775.
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1241

Scheme 6. Chemical development synthesisa
a Reagents and conditions: a) TMSCl, NaI, CH3CN; b) TMSCl, MeOH; c) KOtBu, 2-MeTHF, 0 °C; d) 30% w/w citric acid; e) i. HCl cat., toluene, reflux; ii.
n-heptane; f) DBU, toluene, reflux (60% overall yield).
(Scheme 6). Under these conditions the reaction is fast and
clean. Some attempts to precipitate 29 were conducted involving
solubility screening. In the best precipitation conditions (addition
of water to the reaction mixture), the isolated yield was only
70% (25% of the product was still present in the mother liquor).
Instead of isolating 29, a solvent swap with methanol was done.
The next step, methyl ester formation, was carried out under
classical conditions using trimethylsilylchloride in methanol at
room temperature. The reaction performed well and provided
product in high yield. Unfortunately, the methyl ester 27 is an
oil, and therefore, direct isolation was not ideal. A procedure
was developed whereby 27 was isolated as a solution suitable
for the next reaction. This work-up included methanol displace-
ment Via distillation with 2-methyltetrahydrofuran (2-MeTHF),
aqueous reductive work-up to destroy any trace of iodine, and
extractive work-up and distillation to azeotropically dry the
2-MeTHF solution. The solution of 27 in 2-MeTHF was then
used in the next transformation.
Conversion of 27 to the lactam 25 was achieved using a
three-step procedure. Addition of 27 to the preformed anion of
N-(diphenylmethylene)glycine tert-butyl ester 26 using potas-
sium tert-butoxide at 0 °C provided a diastereoisomeric mixture
of 30 and 31 in a good yield.10 The diastereoisomeric ratio at
this point was 67:33 in favor of the desired anti isomer 30
(precursor of the desired lactam 25). As 30 and 31 were oils
and therefore column chromatography was necessary to separate
them, we sought to postpone the isolation and purification until
later on in the synthesis, as it is known that compounds 25 and
33 are solids.
Knowing that the benzophenone imine deprotection condi-
tions used aqueous citric acid11 and the cyclization step was
accomplished at reflux in toluene, a telescoping procedure was
investigated. After nucleophilic substitution of the iodide 27,
the mixture was treated with aqueous citric acid solution to
afford a solution of the diastereoisomeric aminoacids 32 and
benzophenone. The resulting acidic aqueous solution was then
washed with cyclohexane to remove the benzophenone and then
the aqueous layer was basified. The free amine 32 could be
extracted with ethyl acetate which was then displaced Via
distillation by toluene, in order to have a higher boiling point
solvent. The solution of 32 in toluene was refluxed to induce
thermal cyclization to furnish the lactames 25 and 33. Then
the mixture was cooled to room temperature which induced
partial precipitation of 33. Nevertheless, the process was not
robust enough since the reaction time during the last step (lactam
formation) was not reproducible (from 12 h to 4 days during
the first kilo laboratory experiments). To solve this issue, a
catalytic amount of 37% aqueous hydrochloric acid was added
and provided a consistent reaction time of 23 h. Unfortunately,
the same diastereoisomeric ratio present in the mixture 30/31
is observed also in the mixture 25/33.
As the diastereoselectivity of the coupling reaction between
27 and 26 is moderate, a decrease of the reaction temperature
to improve the selectivity was investigated. Nevertheless, only
the reaction rate (slower reaction) was affected and not the
diastereoisomeric ratio. Although asymmetric additions to
N-(diphenylmethylene)glycine tert-butyl ester using Maruoka’s
catalyst12 were planned, time restrictions limited exploration of
(9) Kolb, M.; Barth, J. Synth. Commun. 1981, 11, 763–767.
(10) Kozhushkov, S. I.; Zlatopolskiy, B. D.; Brandl, M.; Alvermann, P.;
Radzom, M.; Geers, B.; de Meijere, A.; Zeeck, A. Eur. J. Org. Chem.
2005, 854–863.
(11) (a) O’Donnell, M. J.; Boniece, J. M.; Earp, S. E. Tetrahedron Lett.
1978, 19, 2641–2644. (b) Yamada, S. I.; Oguri, T.; Shioiri, T. J. Chem.
Soc., Chem. Commun. 1976, 136–137.
1242
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development

Figure 1. Phase diagram for compounds 25 and 33.
such alternatives to future research. As a result, different
approaches were applied in order to isolate the pure anti isomer
25:
Upon suspending a mixture of acids 34/35 in a ratio 72:28 in
dichloromethane, it was found that the composition of the
polysaturated solution was anti/syn 59:41, so that in principle
it was possible to partially precipitate the desired diastereoisomer
34 (Figure 2). The main issues were that the crystallization had
to be run in a large volume of dichloromethane (clearly this
could have been solved by searching for a solvent where 35
was more soluble) and that, because we started from a 67:33
mixture, the maximum possible recovery of pure anti isomer
was calculated to be 20%. On the basis of these scarce results,
this approach was discarded.
1 chromatographic separation
2 selective crystallization
3 epimerization of syn diastereoisomer 33
As the first approach was the least desirable, most of the
efforts were concentrated on the second and third solution in
view of future scale-ups, where chromatography is not practical,
time-consuming, and environmentally unfriendly.
The direct crystallization is one of the most popular methods
to enrich the diastereoisomeric purity of a compound, and with
one simple experiment it is possible to understand if this
approach is feasible: building the ternary phase diagram and
determining the polysaturated solution as depicted in Figure 1.13
In the phase diagram, four zones are identified: zone A,
where both diastereoisomers 25/33 are in solution; zone B,
where 33 crystallizes out leaving mainly 25 in solution; zone
C, where 25 crystallizes out leaving mainly 33 in solution; and
finally zone D, where both diastereoisomers 25/33 crystallize.
In our case, the polysaturated point was determined by suspend-
ing a mixture of 25/33 in toluene and leaving the mixture to
reach the equilibrium. Analysis of the mother liquors showed
a 25/33 ratio of 67:33. As shown in Figure 1 (right-hand side
diagram), in principle it is possible to dissolve a mixture 25/33
in a ratio 50:50 and precipitate the pure isomer 33, leaving the
mother liquors enriched in compound 25 by using an amount
of solvents that allows crystallization and isolation of the product
within zone B.
Unfortunately, as we wanted to isolate the isomer 25, this
approach was not useful for our objective. In addition, because
the reaction between 27 and 26 afforded a mixture of 30/31
(and subsequently a mixture 25/33) in the same ratio of the
polysaturated solution, a crystallization would deliver a solid
once again with a 67:33 ratio, and therefore none of the two
compounds could be selectively precipitated. The same exercise
was applied to a mixture of the corresponding acids 34/35.14
In parallel, an important breakthrough was obtained follow-
ing the third approach, i.e. the epimerization of the undesired
diastereoisomer 33. Indeed, the undesired diastereoisomer 33
in the presence of DBU in toluene afforded a mixture in a 83:
17 ratio in favor of 25.
Having shown the possibility to epimerize the undesired
diastereoisomer under basic conditions in favor of the desired
one, it was decided to isolate both esters 25/33 from the reaction
mixture after lactam formation, and then isomerize the undesired
diastereoisomer. Indeed, the addition of n-heptane as antisolvent
led to the precipitation of 25/33 in an excellent quality. Thus,
the telescoped procedure for the chiral lactone opening, esteri-
fication, alkylation reaction of 26 with 27, hydrolysis, cycliza-
tion, and isolation produced 1.6 kg of 25/3315 in high purity
for an overall yield of 60% over five steps.
With a robust procedure to synthesize the lactam as an anti/
syn mixture in 67:33 ratio, the next transformation entailed an
appropriate epimerization of this intermediate to then reach
efficiently the key scaffold, 4.
Since DBU in toluene was able to epimerize the undesired
diastereoisomer 33 and provided an 83:17 ratio in favor of 25,
a screening of 14 additional solvents was performed to improve
this ratio. Unfortunately, no improvement was obtained.
Nevertheless, from an extended base screening, it emerged
that, treating a mixture 25/33 in 67:33 ratio with powdered
potassium hydroxide in 2-propanol/toluene at reflux, epimer-
(12) (a) Hashimoto, T.; Maruoka, K. Chem. ReV. 2007, 107, 5656–5682.
(b) Maruoka, K.; Ooi, T. Chem. ReV. 2003, 103, 3013–3028.
(13) Marchand, P.; Lefe`bvre, L.; Querniard, F.; Cardinae¨l, P.; Perez, G.;
Counioux, J.-J.; Coquerel, G. Tetrahedron: Asymmetry 2004, 15, 2455–
2465.
(14) A 72:28 mixture of tert-butyl ester 25/33 resulting from column
chromatography and treated with trifluoroacetic acid afforded quan-
titatively the corresponding carboxylic acid 34/35 in a 72:28 ratio.
Figure 2. Phase diagram for compounds 34 and 35.
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1243

Scheme 7. Hypothetical epimerization mechanism
Scheme 8
ization/ester hydrolysis occurred, giving a ratio of 98:216 in favor
of the anti diastereoisomer of the corresponding potassium salt
(36, Scheme 7). This result could be explained because the
hydrolysis of the syn ester is slower than its epimerization to
the anti ester, as depicted in Scheme 7. Although the reaction
was very clean and the diastereoisomeric ratio was high (96%
de), the isolation of the salt was not feasible because of its high
hygroscopicity.
We decided to convert the carboxylic acid salt 36 into the
corresponding methyl ester, speculating that the methyl ester
might be isolated as a solid. Therefore, a solvent screening was
performed in order to develop a two-step, one-pot procedure.
Thus, addition of powdered potassium hydroxide to a solution
of 25/33 in toluene/methanol, then followed by addition of an
excess of TMSCl led to the complete conversion to 39 (Scheme
8). The remarkable success of these reactions provided us with
a simple and efficient procedure to recycle the undesired
diastereoisomer 33 into the desired one. Unfortunately, the
methyl ester 39 is very soluble in water, prohibiting any possible
aqueous work-up to remove any inorganic impurity. Thus, an
adequate work-up was optimized to obtain 39 as solution
suitable for use in the next reaction. After complete conversion
to the methyl ester, methanol and toluene were partially
displaced Via distillation with THF, which was the desired
solvent for the subsequent transformation.
performed by the medicinal chemists on a similar substrate.
This reduction involved the use of BH3 at reflux to reduce the
lactam moiety, followed by protecting the secondary amine with
tert-butoxycarbonyl group.
In process development for safety reasons, BH3 is commonly
replaced by the procedure generating BH3 in situ by addition
of BF3 ·THF to sodium borohydride. However, addition of
BF3 ·THF to the mixture of 39 and sodium borohydride in THF
did not give complete conversion and surprisingly led to a
mixture of 23 and 40 in favor of 23 (Scheme 9). Investigation
on lab scale showed that the premixing of 39 with BF3 ·THF
was required to have complete reduction. Consequently, the
solution of 39 precomplexed with BF3 ·THF was added to a
suspension of sodium borohydride at 30 °C affording 40.
However, the conversion was not reproducible when we
scaled up. We reasoned that the difficulty to reach complete
conversion stemmed from the heterogeneous aspect of the
reaction mixture. To overcome this, we replaced powdered
sodium borohydride by a solution of lithium borohydride in
THF as reducing agent. Thus, the reaction was cleaner, faster,
and was not scale dependent. Because of the high water
solubility of 40, the work-up was optimized, isolating 40 as a
suitable solution for the use in the next reaction. The procedure
included careful quench of the excess of borane with methanol
at 50 °C, distillation to remove methanol and the boronic ester,
6 M aqueous hydrochloric acid treatment to destroy amino
boronate complex and finally careful basification of aqueous
layer. The basic aqueous solution of 40 in THF in the presence
of a slight excess of Boc-anhydride afforded 24, which can be
precipitated by addition of n-heptane in moderate yield (Scheme
9). In fact, the presence of a large amount of two impurities
(4-chlorobutanol and n-butanol) resulting from the THF deg-
Indeed, the next step required the complete reduction of 39
to produce the desired aminoalcohol 40, followed by the
protection of the resulting secondary amine to afford the desired
scaffold 4. Originally, this two-step procedure was already
(15) The analytical standards of 25 and 33 were obtained by column
chromatography separation on silica gel (eluent: cyclohexane/ethyl
acetate 7:3).
(16) Ratio obtained in solution.
1244
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development

Scheme 9
Scheme 10a
a Reagents and conditions: a) KOH, toluene/MeOH; b) i. TMSCl; ii. NaHCO3, THF; c) LiBH4 (4 M in THF), BF3 · THF; d) i. Boc2O, THF; ii. isooctane (73%
overall yield).
radation were able to solubilize 24 and therefore decrease the
overall yield. Additionally, 24 was contaminated by 41, the
result of a double protection during the last step (Scheme 9).
To increase the recovery of 24, several modifications of the
reaction conditions were introduced. First, the decrease of the
amount of Boc-anhydride to 0.98 equiv and a pH control (pH
) 9) of the reaction mixture were able to prevent the formation
of 41 and to obtain a very clean reaction profile. Second,
4-chlorobutanol was found to derive during the solvent swap
from methanol to THF after the formation of the methyl ester
due to the presence of hydrogen chloride (arising from TMSCl
and MeOH) and also from the 6 M HCl treatment of the amino
boronate complex initially done at reflux. Consequently, to
prevent its formation, after the formation of the methyl ester
39, the reaction mixture was neutralized with powdered
NaHCO3 prior to solvent swap, while the cleavage of the amino
boronate complex was performed at room temperature. Regard-
ing the removal of n-butanol (due to the THF opening by lithium
borohydride), it was achieved by washing the acidic aqueous
layer with toluene prior to basification and Boc-protection.
Finally, an extensive solubility screening was performed to
identify the best precipitation conditions. Among the solvents
screened, isooctane proved to be the most suitable solvent for
the isolation.
Thus, these modifications were implemented, and the five
telescoped-step procedure involving epimerization/hydrolysis,
methyl esterification, concomitant reduction of both lactams to
amine and ester to alcohol, protection and isolation gave over
four batches, 3 kg of 24 in high purity (98% by HPLC, >99%
ee and >99% de) and in an overall yield of 73% (Scheme 10).
It is worth noting that most of the intermediates were not
UV sensitive; therefore, the reaction monitoring was a real
challenge. Analytical procedures were developed to monitor the
reaction Via proton NMR.
In conclusion, an efficient and scalable synthetic route for
the synthesis of the key related scaffold 24, a common
intermediate required for the synthesis of the targeted molecules
1, has been developed from the commercially available chiral
lactone 28. One of the key steps is the epimerization of the
undesired diastereoisomer 33. Compared to the discovery route
and Medicinal Chemistry Support synthesis, this route offers
significant advantages in terms of yield (44% overall versus
24% for Medicinal Chemistry and 6% for Medicinal Chemistry
Support) with only two isolations Via precipitation and avoiding
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1245

all chromatography purifications. This process has been utilized
on kilo lab scale to produce over 3 kg of 24.
solution was added 37% aqueous HCl (6 mL, catalytic amount).
The solution was heated to 105 °C for 23 h. The solution was
cooled at 40 °C and reduced to 4 volumes (4.8 L) under reduced
pressure, and n-heptane (8.4 L) was added over 1 h. The mixture
was stirred at 40 °C for 30 min and then cooled at 15 °C over
1 h; a solid was precipitated. The slurry was stirred at 15 °C
for approximately 16 h and then filtered. The solid was washed
with n-heptane (2 × 3 L) and dried in a vacuum oven at 40
°C. 1,1-Dimethylethyl-(1R,4S,6R)-2-oxo-3-azabicyclo[4.1.0]-
heptane-4-carboxylate (25) and 1,1-dimethylethyl-(1R,4R,6R)-
2-oxo-3-azabicyclo[4.1.0]heptane-4-carboxylate (33) (1.58 kg,
60%) were obtained as white solids in 67:33 ratio.
Experimental Section
All the materials were purchased from commercial suppliers
and used without further purification. All reactions were carried
out under atmosphere of nitrogen.
1,1-Dimethylethyl-(1R,4S,6R)-2-oxo-3-azabicyclo[4.1.0]hep-
tane-4-carboxylate (25) and 1,1-Dimethylethyl-(1R,4R,6R)-
2-oxo-3-azabicyclo[4.1.0]heptane-4-carboxylate (33). Sodium
iodide (2.76 kg, 18.4 mol, 1.5 equiv) was partially dissolved in
acetonitrile (12 L) after stirring at 20 °C for 10 min under
nitrogen atmosphere. TMSCl (2.3 L, 18.4 mol, 1.5 equiv) was
added over 15 min, and the resulting yellow slurry was stirred
at 20 °C for 1 h. A solution of (1R,5S)-3-oxabicyclo[3.1.0]hexan-
2-one (28, 1.2 kg, 12.2 mol, 1 equiv, > 99% ee, [R]D ) +61.5°
(c ) 1, CHCl3)) in acetonitrile (2.4 L) was added over 5 min
at 20 °C. The suspension was heated to 50 °C (internal
temperature) and then kept for 2 h at 50 °C. The mixture was
diluted with methanol (12 L) at 20 °C and concentrated to 5
volumes (6 L) under reduced pressure. Methanol (12 L) was
then added followed by TMSCl (0.72 L, 5.7 mol, 0.5 equiv).
The resulting mixture was stirred at 20 °C overnight. The
mixture was concentrated under vacuum to 5 volumes (6 L),
then 2-MeTHF was added (12 L), and the solution was
concentrated to 5 volumes (6 L). 2-MeTHF was added (12 L).
The dark-red solution was washed with aqueous 20% w/w
Na2SO3 (4.8 L) at 20 °C (solution became colourless-light
yellow). The biphasic system was separated, and the organic
layer was washed with water (4.8 L), then concentrated under
vacuum to 4 volumes (4.8 L). 2-MeTHF (12 L) was added,
and the solution was concentrated to 5 volumes (6 L), to obtain
finally the solution of methyl (1R,2S)-2-(iodomethyl)cyclopro-
panecarboxylate 27.
N-(Diphenylmethylene)glycine tert-butylester (26, 3.6 kg,
12.2 mol, 1 equiv) was suspended in dry 2-MeTHF (12 L) at
20 °C. The mixture was cooled to 0 °C, and KOtBu (1.38 kg,
12.3 mol, 1 equiv) was added in three portions. The slurry
became a yellow-orange solution and was stirred at 0 °C for
30 min. The previous solution of methyl-(1R,2S)-2-(iodometh-
yl)cyclopropanecarboxylate 27 in 2-MeTHF was slowly added
over 25 min, keeping the temperature lower than 5 °C during
the addition. The mixture was stirred at 0 °C for 2.5 h. The
mixture was quenched with buffer pH ) 7 (KH2PO4/Na2HPO4,
2.4 L) at 0 °C. The biphasic system was warmed at 20 °C. The
water phase was discharged. The organic phase was cooled to
0 °C and treated with 30% w/w citric acid (9.6 L). The biphasic
system was warmed to 20 °C and stirred overnight at 20 °C.
Cyclohexane (24 L) was added, and the phases were separated.
The water phase was washed with cyclohexane (24 L). Ethyl
acetate (24 L) was added to the aqueous phase, and then the
system was basified to pH ) 8.5 with aqueous saturated K2CO3
(6 L) and then diluted with water (3 L). The biphasic system
was separated. The aqueous layer was back extracted with ethyl
acetate (24 L). The combined organic phases were washed with
water (3.6 L) and concentrated to 10 volumes (12 L). Toluene
(24 L) was added, and the solution was concentrated to 10
volumes (12 L) and diluted again with toluene (4.8 L). To this
1
(25) H NMR (400 MHz, DMSO-d6): δ (ppm) 6.86 (br s,
1H), 3.75 (dd, J ) 11.9, 4.7 Hz, 1H), 2.20 (ddd, J ) 13.2, 4.5,
2.3 Hz, 1H), 1.74 (td, J ) 12.6, 3.4 Hz, 1H), 1.42 (s, 9H),
1.40-1.60 (m, 2H), 1.12 (q, J ) 5.3 Hz, 1H), 0.90 (ddd, J )
9.3, 7.9, 5.3 Hz, 1H). 13C NMR (151 MHz, DMSO-d6): δ (ppm)
7.0, 12.5, 15.6, 24.6, 27.5, 49.7, 81.4, 170.1, 171.0. HRMS
(ESI): m/z [M + H]+ calcd for C11H17NO3: 212.1287; found
212.1288.
(33) 1H NMR (400 MHz, DMSO-d6): δ (ppm) 7.39 (d, J )
4.9 Hz, 1H), 3.87 (dddd, J ) 8.2, 5.2, 1.3, 0.7 Hz, 1H), 2.33
(d, J ) 13.9 Hz, 1H), 2.11 (ddd, J ) 14.2, 8.1, 3.2 Hz, 1H),
1.42 (s, 9H), 1.35-1.60 (m, 2H), 0.87 (m, 1H), 0.69 (td, J )
5.4, 4.4 Hz, 1H). 13C NMR (151 MHz, DMSO-d6): δ (ppm)
9.0, 13.9, 15.6, 21.8, 27.5, 51.2, 81.1, 170.8, 172.7. HRMS
(ESI): m/z [M + H]+ calcd for C11H17NO3: 212.1287; found
212.1288.
1,1-Dimethylethyl-(1R,4S,6R)-4-(hydroxymethyl)-3-
azabicyclo[4.1.0]heptane-3-carboxylate (24). The mixture of
1,1-dimethylethyl (1R,4S,6R)-2-oxo-3-azabicyclo[4.1.0]heptane-
4-carboxylate (25) and 1,1-dimethylethyl-(1R,4R,6R)-2-oxo-3-
azabicyclo[4.1.0]heptane-4-carboxylate (33) (1 kg, 4.73 mol,
1 equiv) as 67:33 ratio was dissolved in toluene (3 L) and
MeOH (7 L) and stirred for 5 min at 20 °C. The mixture was
cooled to 15 °C, and powdered KOH (0.4 kg, 7.13 mol, 1.5
equiv) was added in eight portions over 1 h. The solution was
stirred at 20 °C for 3 h. The solution was cooled to 10 °C, and
TMSCl (2.4 L, 18.9 mol, 4 equiv) was added, keeping the
temperature around 10-15 °C over 45 min. White solid was
precipitated (KCl). The slurry was stirred at room temperature
overnight. NaHCO3 solid (1.4 kg) was added in three portions
to reach pH ) 5. The mixture was concentrated to 4 volumes
(4 L). THF (8 L) was added, and the volume was reduced to 4
volumes (4 L) by distillation under reduced pressure. The solid
was filtered and washed with THF (3 × 2 L). The filtrate
appeared cloudy. The solution was reduced to 2.5 volumes (2.5
L) by distillation under reduced pressure, and BF3 ·THF (3.13
L, 28.4 mol, 6 equiv) was added under stirring whilst maintain-
ing an internal temperature of 25 °C. The resulting solution
was added slowly to a solution of LiBH4 (4 M in THF, 4.8 L,
19.2 mol, 4 equiv), diluted with THF (3 L), keeping the
temperature at 25-30 °C [the line was washed with THF (0.5
L)]. The mixture was stirred at 30 °C overnight (17 h). The
mixture was slowly quenched with MeOH (4 L) at 25-30 °C
(addition time in about 1 h). The solution was stirred at 50 °C
for approximately 1 h. After this time, the solution was reduced
to 5.5 volumes (5.5 L) by distillation under reduced pressure.
1246
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development

HCl [3 M (4 L)] was then added at 10-15 °C. The mixture
was stirred at 20 °C for 1 h and toluene (4 L) was added. The
phases were separated. The aqueous phase was washed with
toluene (3 × 4 L). The aqueous layer was basified with 6 M
NaOH (3.5 L) until pH ) 9. To the basic aqueous solution at
25 °C were successively added THF (0.5 L) and a solution of
di-tert-butyl dicarbonate (1 kg, 4.58 mol, 0.98 equiv) in THF
(1 L). The pH was adjusted to 9 by addition of 6 M NaOH
(1.2 L). The resulting slurry was stirred for 30 min at 25 °C,
and the pH was adjusted to pH ) 9 by addition of 6 M NaOH
(1 L). The slurry was then stirred for 3 h and then filtered. The
inorganic salts were washed with TBME (2 × 2 L). The filtrate
was diluted with TBME (2 L). The biphasic system was
separated. The aqueous layer was back extracted with TBME
(6 L), and the combined organic phases were washed with NaCl
20% w/w (4 L) and then concentrated under reduced pressure
to 5 volumes (5 L). Isooctane (10 L) was added and the solution
was reduced to 6 volumes (6 L) by distillation under reduced
pressure. Seed (1 g) of the title compound was added at 40 °C,
and the slurry was cooled at 20 °C in 30 min to allow
precipitation. The slurry was stirred for at least 4 h and filtered.
The solid was washed with cold isooctane (2 × 2 L) and dried
in a vacuum oven at 40 °C. 1,1-Dimethylethyl-(1R,4S,6R)-4-
(hydroxymethyl)-3-azabicyclo[4.1.0]heptane-3-carboxylate (24,
0.8 kg, 73%) was obtained as white solid (98% purity by
HPLC). Chiral HPLC analysis of 24 showed product with >99%
ee and no trace of undesired syn diastereoisomer. Mp ) 71
°C.
1H NMR (600 MHz, DMSO-d6): δ (ppm) 4.66 (br m, 1H),
3.65-3.85 (br m, 2H), 3.41 (br m, 1H), 3.37 (m, 1H),
3.20-3.35 (br m, 1H), 1.89 (m, 1H), 1.53 (ddd, J ) 14.3, 6.8,
3.3 Hz, 1H), 1.37 (s, 9H), 0.91 (m, 2H), 0.58 (br m, 1H), -0.09
(q, J ) 4.8 Hz, 1H). 13C NMR (151 MHz, DMSO-d6): δ (ppm)
5.2, 8.4/8.5, 8.8/8.9, 22.3/22.8, 28.1, 37.2/38.2, 49.2/50.4, 60.8/
61.2, 78.3/78.4, 154.7/154.9. HRMS (ESI): m/z [M + H]+ calcd
for C12H21NO3: 228.1600; found 228.1608.
Acknowledgment
We thank Simone Guelfi, Stefano Belletato, Giuliana
Scardoni, and Simone Battaggia from the scale-up facili-
ties for their contributions to the development and the
scale-up of the chemistry to prepare 24. We thank also
Claudio Bismara from Business Operation for having
found suppliers for the chiral lactone 28. We thank Luca
Massari and Franck Mallet for their help in the phase
diagram. And finally, we thank our colleagues from Me-
dicinal Chemistry for their original synthesis.
Received for review June 11, 2010.
OP100164V
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1247