Development of a supply route for the synthesis of an iNOS inhibitor: Complications of the key SN2 reaction

Article abstract of DOI:10.1021/op900108b

The original medicinal chemistry synthesis of an iNOS inhibitor presented several challenges that had to be overcome in order to constitute a supply route suitable for operation on a multikilo scale. The key step in the synthesis is an S_N2 reac

Full text of DOI:10.1021/op900108b

Organic Process Research & Development 2009, 13, 774–780
Development of a Supply Route for the Synthesis of an iNOS Inhibitor: Complications
of the Key SN2 Reaction
Geracimos Rassias,* Stephen A. Hermitage,* Mahesh J. Sanganee, Peter M. Kincey, Neil M. Smith, Ian P. Andrews,
Gary T. Borrett, and Graham R. Slater
GlaxoSmithKline R&D, Chemical DeVelopment, Synthetic Chemistry, Gunnels Wood Rd, SteVenage,
Hertfordshire SG1 2NY, U.K.
Scheme 1. Synthesis of protected cysteine componenta
Abstract:
The original medicinal chemistry synthesis of an iNOS inhibitor
presented several challenges that had to be overcome in order to
constitute a supply route suitable for operation on a multikilo scale.
The key step in the synthesis is an SN2 reaction that assembles
the chiral carbon framework, but this reaction proved far
more complex than we anticipated. Significant improve-
ments were made to all stages. The modified route per-
formed well over two pilot-plant campaigns and delivered
over 250 kg of the active pharmaceutical ingredient (API).
a Reagents and conditions: (a) tert-butyl trichloroacetimidate, DCM/heptane;
(b) DCM, NEt3, dithiothreitol, 80% from 2.
Scheme 2. Protection and activation of
(S)-(+)-1-amino-2-propanola
a Reagents and conditions: (a) Aq K2CO3, CbzCl, DCM; (b)TosCl, NEt3,
NMe3.HCl, 82% from 5.
1. Introduction
Nitric oxide (NO) is an important physiological mediator
that plays a role in the regulation of blood pressure and blood
flow, as a neurotransmitter in the central and peripheral systems
and in the immune system.1 However, the overproduction of
NO by the inducible isoform of the NO synthetases (iNOS) is
implicated in the pathophysiology of several disease states such
as septic and endotoxic shock, neurodegenerative disorders, and
inflammatory diseases such as asthma, rheumatoid arthritis, and
multiple sclerosis. It is desirable, therefore, to develop selective
inhibitors of iNOS for therapeutic use with a high degree of
selectivity over other isoforms.1 This paper describes our efforts
to synthesise kilogram quantities of 1 (Figure 1), an iNOS
inhibitor candidate, to support toxicology and clinical studies.
N-Boc cystine bis-tert-butyl ester 3 into the monomer 4
employing standard literature conditions.2-4
In a two-step process (Scheme 2), (S)-(+)-1-amino-2-
propanol (5) was converted to tosylate 7 by protection with
CbzCl followed by tosylation using a combination of p-
toluenesulfonyl chloride, triethylamine, and trimethylamine
hydrochloride according to a known method.5-7 The trimethy-
lamine generated in this process reacts with tosyl chloride to
form the trimethylammonium adduct. This reactive intermediate
acts as the tosylating agent and not only accelerates the reaction
but also aids the hydrolytic decomposition of the excess tosyl
chloride during the workup. A side reaction is the chlorodem-
ethylation of the activated species, giving rise to N,N-dimeth-
yltolylsulfonamide, but this is removed in the crystallisation of
the product.
In total, more than 500 kg of each of 7 and 4 were prepared
using the chemistry described above.
Development of the SN2 Process. The reaction of 4 and 7
in an apparently simple SN2 reaction formed the basis of the
quality critical step in the preparation of 1.
Figure 1
2. Results and Discussion
Synthesis of the SN2 Reaction Partners. In our first
campaign the synthesis began with the preparation of bis-N-
Boc cystine bis-tert-butyl ester 3 from bis-N-Boc-cystine 2 and
tert-butyl trichloroacetimidate (Scheme 1). For the campaigns
that followed we purchased this material from a commercial
source, and our synthesis started with the reduction of the bis-
The original medicinal chemistry procedure for the prepara-
tion of 8 involved combination of equimolar amounts of 4 and
7 and anhydrous cesium carbonate, in 43 volumes of degassed
(2) Tsuda, Y.; Okada, Y.; Tanaka, M.; Shigematsu, N.; Hori, Y.; Goto,
T.; Hashimoto, M. Chem. Pharm. Bull. 1991, 39 (3), 607.
(3) Kadereit, D.; Waldmann, H. ChemBioChem 2000, 1 (3), 200.
(4) Roth, A.; Espuelas, S.; Thumann, C.; Frisch, B.; Schuber, F.
Bioconjugate Chem. 2004, 15 (3), 541–553.
* Authors for correspondence. E-mail: geracimos.9.rassias@gsk.com;
stephen.a.hermitage@gsk.com.
(1) Alderton, W. K.; Angell, A. D. R.; Craig, C.; Dawson, J.; Garvey, E.;
Moncada, S.; Monkhouse, J.; Rees, D.; Russell, L. J.; Russell, R. J.;
Schwartz, S.; Waslidge, N.; Knowles, R. G. Br. J. Pharmacol. 2005,
145, 301.
(5) Sugiyama, H.; Yokokawa, F.; Shioiri, T. Org. Lett. 2000, 2 (14), 2149.
(6) Hayashi, T.; Yamasaki, K.; Mimura, M.; Uozumi, Y. J. Am. Chem.
Soc. 2004, 126 (10), 3036.
(7) Galvez-Ruiz, J. C.; Jaen-Gaspar, J. C.; Castellanos-Arzola, I. G.;
Contreras, R.; Flores-Parra, A. Heterocycles 2004, 63 (10), 2269.
774
•
Vol. 13, No. 4, 2009 / Organic Process Research & Development
10.1021/op900108b CCC: $40.75  2009 American Chemical Society
Published on Web 07/02/2009

Scheme 3. Reaction of 7 and 4 to afford 8 is the quality
critical step
Scheme 4. Degradation of 4 and 8 and mechanism of
impurity formation
acetonitrile at 10 °C for 13 h (Scheme 3). Thioether 8 was
obtained in 82% yield after column chromatography. Problems
associated with this method included the large volumes of
acetonitrile needed to be degassed in order to prevent dimer-
ization of thiol 4 to disulfide 3, and chromatography to remove
unreacted 7 and other side products.
During the course of our investigation a crystallisation
procedure was developed that allowed, for the first time,
isolation of 8 without the need for chromatography. This process
tolerated significant amounts of 4 (more than 1 equivalent in
excess) but not of tosylate 7; hence, our focus was turned to
the stoichiometry of the reaction. In our hands there was
complete consumption of the tosylate 7 when the reaction was
carried out at ambient temperature with a 20% molar excess of
both thiol 4 and cesium carbonate, to give 8 in good quality
after crystallisation. Laboratory experiments on 5-10 g scale
also showed that the volume of acetonitrile could be reduced
from 43 volumes to 15 volumes. The cesium tosylate that
precipitates out of the reaction mixture, however, causes
thickening of the medium and poor mixing of the heterogeneous
components at higher concentrations; thus, a 22-volume process
was deemed optimum. The improved process was successfully
demonstrated on 300 g scale in the kilo-lab and product 8 was
obtained in 78% yield. Stretching experiments had shown that
prolonged reaction times caused the product to degrade under
the reaction conditions, and therefore, an assay method was also
developed to establish the point at which the product concentra-
tion had reached a maximum. The revised conditions coupled
with the assay method formed the protocol we adopted to
prepare more than 6.0 kg of 8 in 79% yield using 50 L reactors
in the plant. The apparent lack of robustness of this process
however prompted us to look deeper into the details of this
transformation.
Interestingly, during this work, some reactions unexpectedly
failed to go to completion, with up to 60% 7 remaining
unreacted. Addition of excess of each of the reagents and/or
base failed to restart these reactions. From these runs 8 could
only be isolated in low yield after chromatography because the
presence of 7 compromises the crystallisation process. Since
this is a heterogeneous reaction we initially looked into the
particle size and water content of the hygroscopic Cs2CO3, speed
of agitation and related mass transfer/mixing effects but despite
our systematic approach the stalling effect not only persisted
but also eluded our understanding.
misleading, as this was actually isolated in unexpectedly low
amounts despite the severe decomposition of 8. Not surprisingly,
the Cbz-protected amino thiol component 10 was found in the
reaction mixture together with its oxidised form namely disulfide
11.
Since the elimination of the acrylate is a reversible process
we expected that the corresponding conjugate addition of 10
to 9 could act as a mechanism of racemisation of the stereogenic
centre R to the ester group. The optical purity of the surviving
8 however had not deteriorated appreciably (96% ee), implying
that under these conditions when deprotonation R to the ester
occurs, elimination is much faster than reprotonation (epimer-
ization). The isolation of thiol 10 also indicates that its conjugate
addition to acrylate 9 is not a facile process and in fact 10,
being an oil, remains in the liquors at the end of the process
unless exposed to oxygen at elevated pH in which case it
oxidatively dimerises to the corresponding disulfide 11 (Scheme
4). The major fate of 9 became evident upon characterisation
of compound(s) 12 (Scheme 4), which accounted for 20% of
the mass of the crude reaction mixture. The diastereomeric
mixture of sulfides 12 arises from the conjugate addition of the
excess primary thiol 4 to the eliminated acrylate 9 and had
escaped our attention because of its poor UV absorbance.
Apparently, the anion of 10, being secondary and less reactive,
cannot compete for 9 against the thiolate derived from the
primary and more abundant 4.9,10 The latter pathway essentially
renders 10 the sink for the overall process. Other impurities
include disulfide 3, arising from oxidative dimerization of the
starting thiol 4 in the same way as 10 converts to 11. Although
3 and 11 are formed in small amounts, they are not tolerated
well in the crystallisation and therefore their formation had to
be suppressed. Since the auto-oxidation of thiols takes place in
the presence of oxygen and at high pH, the solvent had to be
In an attempt to understand the intrinsic aspects of this
unusually sensitive SN2 process, a stalled reaction was allowed
to run for longer than usual, and its mass balance was
subsequently examined. The first impurity isolated was acrylate
9 (Scheme 4), arising from base-induced elimination of thiol
10 from 8.8 The high HPLC response factor of 9 was
(9) Chiba, J.; Tanaka, K.; Ohshiro, Y.; Miyake, R.; Hiraoka, S.; Shiro,
M.; Shionoya, M. J. Org. Chem. 2003, 68 (2), 331–338.
(10) Chittiboyina, A. G.; Venkatraman, M. S.; Mizuno, C. S.; Desai, P. V.;
Patny, A.; Benson, S. C.; Ho, C. I.; Kurtz, T. W.; Pershadsingh, H. A.;
Avery, M. A. J. Med. Chem. 2006, 49 (14), 4072–4084.
(8) For a related reaction see Narayan, R. S.; VanNieuwenhze, M. S. Org.
Lett. 2005, 7 (13), 2655–2658.
Vol. 13, No. 4, 2009 / Organic Process Research & Development
•
775

group within 7 (Scheme 5).11,12 Consequently, 14 may be ring-
opened by the (R)-thiolate at either the primary carbon atom,
resulting in the unwanted isomer 13, or at the secondary carbon
atom with inversion, resulting in the also unwanted (R,S)-
diastereoisomer 15 through an overall double inversion [note
that 8 has the (R,R) stereochemistry]. We also looked for
oxazoline 16 which could arise from the intramolecular
displacement of the tosylate by the oxygen atom of the
carbamate carbonyl group (Scheme 5), but none was detected.13
Oxazoline 16 may be attacked at the benzyl group by nucleo-
philes (4 being the best nucleophile in the reaction mixture) to
generate the parent Evans-type oxazolidinone, but neither this
nor the S-benzyl derivative of 4 was found. The absence of
these impurities suggested that the pathway leading to 16 was
not occurring.
The aziridine hypothesis was put to the test, and we were
able to synthesise (R)-N-Cbz-2-methyl aziridine (14) by treating
tosylate 7 with either Cs2CO3 or t-BuOK. Subsequent treatment
of the aziridine with either K or Cs thiolate of 4 revealed an
interesting trend. The kinetic base, which ensures product
stability, was completely consumed in the reaction and gave
an approximately 1:1 mixture of 15 and 13, whereas by using
the usual excess of Cs2CO3 reasonably pure 15 was isolated,
albeit at half the overall yield. This suggests that the latter
reaction probably exhibits the same regioselectivity for the ring-
opening of the aziridine 14, but isomer 13 is degraded faster
than 15 by Cs2CO3 due to the elimination of a primary thiolate
instead of the less likely leaving group secondary thiolate. This
may explain the negligible amounts of 13 found in the
re-examined crude samples of 8 prepared in our first campaign
using the Cs2CO3 method.
The countermeasure we included in our new process was
to preform the thiolate species and then add the tosylate 7, thus
avoiding exposure of 7 to excess of base. This was successful,
provided that a short time was allowed for the deprotonation
because the thiolate species decomposes with time into the Boc
protected R-amino acrylate 9 and potassium hydrogen sulfide.
In fact all of 4, 7, and 8 are sensitive to base, and their
decomposition is concentration-, temperature-, and time-de-
pendent. Thiolate formation is essentially complete at 10 °C
by the time the base is added (over 10 min) as indicated by
monitoring the exotherm which is actually allowed to raise the
batch temperature to 22 °C. Rapid addition of the tosylate to
the potassium thiolate at this point allows the reaction to proceed
at a useful rate. At less than 15 °C the SN2 reaction is slow,
and proton transfer from the NHCbz group to the thiolate and
subsequent aziridine formation becomes the dominant pathway.
A tight control of t-BuOK between 1.05-1.10 equiv in the
presence of 1.2 equiv of thiol offers a robust process with high
conversion, minimum product degradation, and reaction comple-
tion within 10-12 h. Chiral analysis on the crude reaction
mixtures indicates that the regioisomer 15 forms at about 3-5%,
suggesting deprotonation of the NHCbz moiety by the thiolate
species, and therefore aziridine formation occurs unavoidably
Scheme 5. Degradation of 7 and mechanism for formation
of impurities
degassed and we found that refluxing the solvent under nitrogen
or distilling 1-2 volumes prior to dissolving the thiol. Both of
these methods are superior to degassing by sparging with
nitrogen gas. Quenching the reaction with 15% aqueous NH4Cl
also helped to reduce disulfide formation.
The major problem however was the base sensitivity of the
product and the side reactions of the degradants. At this point
it became clear that the continuous exposure of the product to
Cs2CO3 was not offering a sufficiently robust process for use
in our next campaign. Thus a wide range of both inorganic and
organic bases were screened in pursuit of a better coupling
process but only K3PO4, Cs2CO3 and DBU appeared successful.
The latter, although very effective in bringing about the desired
transformation, caused extensive epimerization R to the ester
group in product 8. K3PO4 is comparable to Cs2CO3 in rate
and profile but less consistent due to the irregular particle size
and water content. Phase transfer catalysis offered no advantage
with these heterogeneous systems; hence, we committed our
efforts to kinetic organic bases. Thus, all Li, Na, and K salts of
TMSOH, HMDS, and t-BuOH as well as the corresponding
metal hydrides were tested. The reaction performance parallels
the solubility of the metal thiolate, with the potassium species
being completely soluble in a range of solvents unlike its Li
and Na counterparts. The reaction is fastest in MeCN but not
as clean as in acetone, THF, TBME, or mixtures of the latter
solvents. We opted for stoichiometric quantities of t-BuOK
(20% w/w in THF) with TBME as the bulk reaction solvent,
and rewardingly, in contrast to the Cs2CO3 method, the reaction
is homogeneous, never stalls, completes within 12 h, and can
be left for several days without deterioration of either the ee or
the chemical yield.
Some samples of 8 prepared by this method however
contained a product-like impurity as high as 20% by NMR
although no molecular ions were observed by LC/MS other
than that corresponding to the product. This new impurity did
not match any of the components corresponding to the four
possible diastereoisomers of 8 previously prepared via the
racemic synthesis. Fractional crystallisation of the enriched
samples afforded good quantities of this new impurity, and 1H
NMR studies confirmed that this is the topological isomer 13
(Scheme 5), arising from the transposition of the methyl group
to the adjacent carbon atom.
(11) Trost, B. M.; O’Boyle, B. M. Org. Lett. 2008, 10 (7), 1369–1372.
(12) Hodgson, D. M.; Humphreys, P. G.; Miles, S. M.; Brierley, C. A. J.;
Ward, J. G. J. Org. Chem. 2007, 72 (26), 10009–10021.
(13) Ghosh, A.; Sieser, J. E.; Caron, S.; Watson, T. J. N. Chem. Commun.
2002, (15), 1644.
A rationale for the production of isomer 13 involved the
ring-opening of aziridine 14, formed by deprotonation of the
CbzNH moiety and intramolecular displacement of the tosylate
776
•
Vol. 13, No. 4, 2009 / Organic Process Research & Development

Scheme 6. Final steps in the synthesis of 1a
Figure 2
in the background. Subsequent crystallisation, however, fur-
nishes the desired thioether 8 typically in 80% isolated yield
with regioisomer levels consistently in the range of 1-2%.
Moreover, the unwanted isomer 13 is not only tolerated but is
completely removed in the next stages.
a Reagents and conditions: (a) Pd(OH)2 EtOH, aq NH4HCO2, EtOAc, HCO2H,
86%; (b) EtOAc, S-naphthyl methyl thioacetamidine hydrochloride, 85%; (c)
toluene, K2CO3, then solid H3PO4, water, acetone, 82%.
Finally, traces of impurities 17 (diastereoisomers of 8 and
13) and 18 (Figure 2) were also isolated from the crystallisation
liquors. This further supported our understanding which can
be summarised in the following way: A family of product-like
compounds 17 (all isomers, Figure 2) may result from 4
engaging both 7 and 14. Elimination in isomers 17 can in turn
provide 9 and several thiols that may also attack 7 and 14 to
furnish the group of thioethers 18. In addition, the reversibility
of the elimination of isomers 17 via conjugate additions to 9
offers an epimerization pathway that enhances the population
of group 17. According to the argument above, the stalling effect
in the Cs2CO3 method may be due to base starvation since part
of the base is consumed in deprotonating the CbzNH group.
Base added during the process induces elimination in the already
formed base-sensitive product 8 (and/or 13), while the resulting
electrophile 9 may offer a more reactive partner than tosylate
7 to any thiolate anion present. This mechanistic proposal is in
accord with the impurity profile and explains adequately both
the stalling effect and why the reaction cannot be revived once
derailed. To transform such a sensitive reaction into a robust
process we believe is a remarkable achievement. More than
500 kg of 8 was produced in seven batches using the t-BuOK
process in the plant.
absorb the catalyst and filter the reation mixture prior to aqueous
workup. The solution of the crude amine obtained after
extraction of the free base into ethyl acetate was found to be
contaminated by approximately 2% of the N-ethyl analogue of
the expected amine (possibly due to acetaldehyde generated by
hydrogen transfer involving the solvent and the Pd species
present). Because the desired free amine is not crystalline, after
a screen of appropriate acids we decided to isolate the product
as the formate salt 19 (Scheme 6). Rewardingly, the crystalli-
sation of this salt from ethyl acetate removed the N-ethyl
impurity completely, but the corresponding amine arising from
Cbz removal from impurity 13 could not be eliminated (1-2%).
The crystallisation of the formate salt proved sensitive to the
water content carried over from the aqueous workup. In plant,
we employed multiple distillations of ethyl acetate to azeotro-
pically dry the solution of the amine prior to adding formic
acid. In this way, more than 500 kg of 19 was made in five
pilot-plant batches. The palladium levels in these batches were
consistently below 20 ppm and were reduced to zero after the
final stage.
Preparation of the Amidine Intermediate. In the subse-
quent stage the amidine hydrochloride 20 (Scheme 6) is
prepared by the action of a slight excess of S-(1-naphthyl
methyl) thioimidate on the amine formate 19.14 The reaction,
as inherited from medicinal chemistry, required 18 h in 17
volumes of THF and excess of the amidinating agent followed
by aqueous workup and chromatographic purification. Apart
from the offensive odor of the thiol liberated during the
destruction of S-(1-naphthylmethyl)thioacetimidate, the latter
was also flagged as too toxic an impurity to carry forward in
the final step. The same applied for acetamide which is also
generated from the hydrolysis of the amidinating agent. For
these reasons, significant efforts were devoted into limiting these
undesirable impurities in our penultimate intermediate. We
achieved this by using stoichiometric amounts of 19 and S-(1-
naphthylmethyl)thioacetimidate suspended in ethyl acetate
which allows direct precipitation of the amidine hydrochloride
product 20.15 Addition of methanol at this stage decomposes
the small amount of residual amidinating agent and helps to
solubilise the impurities. The product is isolated via a simple
Cbz Deprotection. In the next reaction we pursued the
removal of the Cbz group in 8 via hydrogenolysis. Despite
screening a variety of conditions and a number of catalysts,
this transformation proved unusually problematic as it frequently
required up to 1.2 wt equiv of Perlman’s catalyst [20% Pd(OH)2
on wet charcoal]. Below 1 wt equiv of the catalyst the reaction
did not proceed to completion. In the past we had successfully
carried out Pd-catalysed hydrogenation/lysis of sulfurous sub-
strates; however, in this case we believe the nature of the
reaction product 19 (1,2-amino sulfide with pendant amino acid
moiety) may be responsible for strong ligation to palladium
species, inhibiting thus the metal to re-enter the catalytic cycle.
Even with catalysts known for their resistance to sulfur
poisoning, we were unsuccessful in achieving a catalytic
reaction superior to that obtained with palladium hydroxide.
We examined alternative protocols for removal of the Cbz
group, but these met either with failure or epimerization or
required significant development work. In order to meet
agressive project timelines we decided to develop the existing
transfer hydrogenolysis process in ethanol using aqueous
ammonium formate as the hydrogen source (Scheme 6). Next,
we investigated a number of methods to remove the palladium
residue from the reaction mixture, and the best way for doing
this was to add Harbolite clay (filtering aid) in the rector to
(14) Barker, P. L.; Gendler, P. L.; Rapoport, H. J. Org. Chem. 1981, 46
(12), 2455.
(15) Box, D.; Colclough, D.; Gillies, I.; Loft, M. S.; Moore, R., PCT Int.
Appl. 2002, WO 2002050021 A1 20020627, Appl. WO 2001-GB5596
20011217 CAN 137:47442, AN 2002:487518, 2002.
Vol. 13, No. 4, 2009 / Organic Process Research & Development
•
777

filtration in 85% yield and excellent purity, and more impor-
tantly, the amine and amidine derived from 13 are not detected.
Final Deprotection and Drug Substance Isolation. The
final stage constitutes a combined global deprotection and salt-
forming process where drug substance 1 is produced. The
original version inherited from medicinal chemistry targeted the
monohydrochloride salt via deprotection of the Boc group and
hydrolysis of the tert-butyl ester with the use of anhydrous HCl
in 1,4-dioxane. This process was undesirable due to the toxicity
of 1,4-dioxane and the generation of the flammable isobutylene
gas from the tert-butyl residues in the Boc and ester groups.
Finally, the hygroscopic nature of both the mono- and dihy-
drochloride salts rendered these versions unacceptable for active
pharmaceutical ingredient (API) development. From the version
screen initiated, the monophosphate monohydrate salt emerged
as the most suitable version for further development. Initially,
80% aqueous H3PO4 solution at 65 °C was used for the
combined removal of the Boc and tert-butyl ester groups.
During this process the desired phosphate salt was also
generated. The use of solid phosphoric acid in the aqueous
deprotection/salt-forming step allowed for greater control of the
amount of reagent charged and consequently on the salt
stoichiometry (API). Generation of the free base of 20 into
toluene and treatment with a freshly prepared H3PO4 solution
facilitated the desired reactions, and rewardingly no isobutylene
gas evolution was detected in this process with online mass
spectrometry. The fate of the tert-butyl group associated with
the Boc and the ester functionalities is its conversion into
tert-butanol.15-17 Addition of the aqueous phase into acetone
with seeding facilitates robust crystallisation of 1. Upon drying,
however, 1 dehydrates, and significant efforts were made to
meet the challenge of consistent preparation of the monophos-
phate monohydrate salt. In collaboration with our Particle
Science and Physical Properties departments no other hydrated
forms were identified by standard solid form screening methods.
Subsequently, the dehydration of the monohydrate was shown
to be fully reversible by DSC, GVS, XRPD, and Raman
spectroscopy. At scale, a reconditioning process was imple-
mented using a Boltz dryer to achieve dehydration followed
by the application of a moist nitrogen stream through the dryer
to rehydrate to the monohydrate. This method proved robust
and consistent for the manufacture of iNOS inhibitor 1 which
was also analysed by a variety orthogonal techniques including
chiral HPLC (99.9% PAR), metal analysis (Pd not detected),
phosphate and water analysis (consistent with the monophoshate
mononohydrate stoichiometry), and XRPD (consistent with
desired polymorph and crystallinity).
toxicity and the overall cost of goods. These are described in a
separate communication.18
4. Experimental Section
1H NMR and 13C NMR data were acquired on a 400 MHz
instrument.
Tosylate (7). Phenylmethyl[(2S)-2-hydroxypropyl]carbam-
ate-1-methyl-methylsulfonylbenzene. Benzylchloroformate (72.64
kg, 2.27 wt, 1 equiv) was added to an aqueous solution (320
L, 10.0 vol) of (S)-1-amino-2-propanol 5 (32 kg, 1.0 wt) and
potassium carbonate (64.96 kg, 2.03 wt) at <20 °C. The mixture
was stirred at ambient temperature for 18 h and then extracted
twice with dichloromethane (320 L, 10.0 vol and 128 L, 4.0
vol). The dichloromethane solution was washed with 5 M
hydrochloric acid (128 L, 4.0 vol) and then diluted with
additional dichloromethane (192 L, 6.0 vol). The organic phase
was then concetrated by vacuum distillation, during which
appoximately 200 L of dichloromethane were removed. The
concentrated solution of the Cbz-protected amino alcohol 6 was
then added in one charge to a suspension of tosyl chloride
(137.92 kg, 4.31 wt) in dichloromethane (160 L, 5 vol), and
the transfer line was washed with 32 L of the same solvent.
Trimethylamine hydrochloride (48.96 kg, 1.53 wt) was added,
follwed by the addition of triethylamine (160 L, 5.0 vol) at such
a rate wherein the temperature would be maintained below 20
°C. The resultant slurry was stirred at ambient temperature for
∼0.5 h and then quenched by the addition of water (320 L,
10.0 vol). The phases were separated, and the organic phase
was washed successively with 5 M hydrochloric acid (320 L,
10.0 vol), 10% w/w aqueous sodium carbonate (320 L, 10.0
vol), and water (320 L, 10.0 vol). The volume was reduced by
vacuum distillation to 160 L, and the product was crystallised
by the addition of heptane (480 L, 15.0 vol). The product was
isolated by filtration and dried at 40 °C under vacuum; 130.2
kg, 84% th. yield, 99.2% pure by 1H NMR.
1H NMR in CDCl3 (CHCl3 as reference at 7.26 ppm) δ
1.24(3H) d J ) 6.4 Hz, 2.40 (3H) s, 3.17-3.26 (1H) m,
3.39-3.47 (1H) m, 4.65-4.74 (1H) m, 4.99-5.11 (3H) m, 7.30
(2H) d J ) 8.2 Hz, 7.31-7.39 (5H) m, 7.78 (2H) d J ) 8.2
Hz. 13C NMR (CDCl3 as reference at 77.00 ppm) δ 18.14,
21.58, 45.72, 66.82, 78.72, 127.71, 127.92, 128.12, 128.50,
129.87, 133.75, 136.28, 144.82, 156.33. HRMS (Electrospray
+) calculated for the protonated molecular ion (MH+)
C18H22N1O5S1 364.1219, found 364.1210.
Thioether (8). 1,1-Dimethylethyl-N-{[(1,1-dimethylethyl)oxy]-
carbonyl}-S-[(1S)-1-methyl-2-({[(phenylmethyl)oxy]carbonyl}ami-
no)ethyl]-L-cysteinate. TBME (1500 L, 20 vol) was degassed
by distillation, and the total volume was adjusted to 1350 L,
18 vol. Thiol 4 (75 kg, 1 wt) was added, and the mixture was
stirred under nitrogen to form a solution. The solution was
cooled to 10 ( 3 °C, and potassium tert-butoxide in THF
(20.5% w/w, 120 kg, 1.6 wt, 1.06 equiv) was added in less
than 10 min (exothermic, ∼12 °C), and the reaction mixture
was stirred for a maximum of 5 min to allow the exotherm to
subside. Immediately after, a solution of tosylate (7) (75 kg, 1
wt) in THF (187.5 L, 2.5 vol), was then added as fast as
possible, followed by a THF (37.5 L, 0.5 vol) line-wash. After
3. Conclusions
In conclusion, we have described our efforts to deliver an
efficient supply route for the iNOS inhibitor 1. Although this
route was capable of delivering multikilogram quantities of
material for ongoing studies, we have further investigated
alternative routes which address the current issues of reagent
(16) Li, B.; Bemish, R.; Buzon, R. A.; Chiu, C. K.-F.; Colgan, S. T.; Kissel,
W.; Le, T.; Leeman, K. R.; Newell, L.; Roth, J. Tetrahedron. Lett.
2003, 44 (44), 8113.
(17) Dias, E. L.; Hettenbach, K. W.; Am Ende, D. J. Org. Process Res.
DeV. 2005, 9 (1), 39–44.
(18) Submitted for publication (Tetrahedron. Lett.).
778
•
Vol. 13, No. 4, 2009 / Organic Process Research & Development

the addition was complete, the reaction mixture was warmed
to 22 ( 3 °C and stirred for 12 h before it was quenched with
600 L, (8 vol) of 15% w/v aqueous ammonium chloride
solution. The mixture is stirred for ∼10 min until a clear
biphasic solution is formed and then allowed to settle. The
aqueous layer was removed, and the organic phase was washed
successively with 450 L (6 vol) of 8% w/v aqueous sodium
bicarbonate solution and 450 L (6 vol) of 30% w/v aqueous
sodium chloride solution. The aqueous layer was removed, and
the organic phase was circulated for at least 30 min through a
bed of sodium sulfate (∼75 kg) before transfer to a clean vessel.
TBME [225 L (3 vol)] was used as vessel rinse and transferred
into the vessel containing the bulk of the filtrate. The filtrate
was concentrated by vacuum distillation to approximately 187.5
L (2.5 vol) while the temperature was maintained below 50
°C. TBME (450 L, 6 vol) was added, and the solution was
reconcentrated to approximately 187.5 L (2.5 vol) by vacuum
distillationwithinthesametemperaturerange.Thisdilution-distillation
cycle was repeated once more, and the solution was finally
diluted with TBME to a total of 300 L (4 vol). Isooctane (450
L, 6 vol) was then added at 20 ( 3 °C over 1 h, and the resulting
suspension was aged at 20 ( 3 °C for 1.5 h. The product was
filtered, washed with a 1:5 mixture of TBME/isooctane (225
L, 3 vol), and dried under vacuum at 30 ( 3 °C; 82.2 kg, 85%
th. yield, 98.2% pure by 1H NMR (1.4% of 13 present, solvents
<0.2%).
(204 L, 3 vol), and then the combined organic solution was
washed with water (408 L, 6 vol). The ethyl acetate solution
was concentrated under reduced pressure to approximately 204
L (3 vol). Two cycles of diluting with ethyl acetate (578 L, 8.5
vol) and concentrating to approximately to 204 L and 408 L,
respectively, were performed. KF analysis of the final solution
confirmed acceptable water content (<2% w/w), the solution
was then cooled to 5 °C, and formic acid (6.8 kg, 0.1 wt, 1
equiv) was added. The resulting slurry was aged for 1 h at 5
°C. The product was filtered, washed with ethyl acetate (2 ×
102 L), and dried at 40 °C under vacuum to constant probe
temperature; 45.48 kg, 86.0% th. yield, 97.6% pure by 1H NMR
(0.9% N-ethyl analogue and 1.1% of deprotected 13 present,
solvents <0.2%).
1H NMR in DMSO-d6 (TMS as reference at 0.0 ppm) δ
1.22(3H) d J ) 6.9 Hz, 1.39 (9H) s, 1.41 (9H) s, 2.68-2.85
(4H) m, 2.85-2.95 (1H) m, 3.92-4.00 (1H) m, 7.24 (1H) d J
) 7.7 Hz, 8.34 (1H) s. 13C NMR (DMSO-d6 as reference at
39.50 ppm) δ 18.80, 27.60, 28.13, 30.85, 40.37, 45.09, 54.85,
78.29, 80.87, 155.31, 165.06, 170.09. HRMS (Electrospray +)
calculated for the protonated molecular ion (MH+) C15H31N2O4S
335.2005, found 335.1994.
Amidine Hydrochloride (20). (2S)-N-[(1Z)-1-Aminoethy-
lidene]-2-{[(2R)-3-[(1,1-dimethylethyl)oxy]-2-({[(1,1-dimethyl-
ethyl)oxy]carbonyl}amino)-3-oxopropyl]thio}-1-propanamini-
um Chloride. To a suspension of amine formate salt 19 (80.0
kg, 1 wt) in ethyl acetate (800 L, 10 vol) was added S-(1-
naphthylmethyl)thioacetimidate hydrochloride (54.4 kg, 0.68 wt,
1.03 equiv) followed by an ethyl acetate (8 L, 0.1 vol) line wash.
The mixture was stirred 2 h at 20 °C before it was quenched
by the addition of methanol (0.4 L, 0.005 vol). Subsequently,
the reaction mixture was heated to 55 °C for 2.5 h, cooled to
20 °C, and stirred for 1 h. The resulted slurry was filtered, and
the cake was washed first with ethyl acetate (2 × 120 L) and
then with MIBK (2 × 120 L). The product was dried at 45 °C
under reduced pressure to constant probe temperature; 83.85%,
1H NMR in CDCl3 (CHCl3 as reference at 7.26 ppm) δ 1.26
(3H) d J ) 6.9 Hz, 1.43 (9H) s, 1.47 (9H) s, 2.85-3.02 (3H)
m, 3.18-3.27 (1H) m, 3.32-3.41 (1H) m, 4.35-4.42 (1H) m,
5.11 (2H) s, 5.27-5.34 (1H) m, 5.39 (1H) broad, 7.28-7.38
(5H) m. 13C NMR (CDCl3 as reference at 77.00 ppm) δ 19.10,
27.96, 28.27, 33.28, 41.40, 45.84, 53.89, 66.72, 77.21, 80.04,
82.72, 128.07*, 128.47, 136.53, 155.30, 156.48, 169.83 (* )
two overlapping signals). HRMS (electrospray +) calculated
for the sodiated molecular ion (MNa+): C23H36N2O6SNa
491.2192, found 491.2186.
1
Amine Formate Salt (19). (2S)-2-{[(2R)-3-[(1,1-Dimeth-
ylethyl)oxy]-2-({[(1,1-dimethylethyl)oxy]carbonyl}amino)-3-
oxopropyl]thio}-1-propanaminium Formate. A mixture of
thioether 8 (68 kg, 1 wt), 20% palladium hydroxide on carbon
paste (68 kg, 1 wt) and IMS (680 L, 10 vol) was stirred and
heated to 40-45 °C. A solution of ammonium formate (27.4
kg, 0.403 wt, 3 equiv) in 1:1 water/IMS (136 L, 2 vol) was
added over ∼30 min, and the reaction mixture was kept at
40-45 °C for a further 30 min before it was cooled to 20 °C.
Harbolite 800 filter aid (40.8 kg, 0.6 wt) was added, and the
mixture was stirred for 15-30 min before the solids (catalyst
and filter aid) were removed by filtration. The filter cake was
washed with IMS (340 L, 5 vol), and the combined filtrates
were concentrated under vacuum to approximately (204 L, 3
vol). Ethyl acetate (578 L, 8.5 vol) was added, and the mixture
was concentrated under reduced pressure vacuum to ap-
proximately (204 L, 3 vol). This dilution-distillation cycle was
97% th. yield, 99.5% pure by H NMR (solvents <0.2%).
Chloride by IC 8.5% w/w, th. 8.6% w/w.
1H NMR in DMSO-d6. (TMS as reference at 0.0 ppm) δ
1.25(3H) d J ) 6.8 Hz, 1.39 (9H) s, 1.41 (9H) s, 2.19 (3H) s*,
2.78-2.91 (2H) m, 3.00-3.09 (1H) m, 3.35 (2H) d J ) 6.9
Hz, 3.92-3.99 (1H) m, 7.21 (1H) d J ) 8.1 Hz, 8.85 (1H)
broad, 9.37 (1H) broad, 9.57 (1H) broad (* ) This signal has
an associated minor rotameric peak at 2.24 ppm). 13C NMR in
DMSO-d6 (DMSO-d6 as reference at 39.5 ppm) δ 18.56, 18.91,
27.60, 28.14, 31.27, 38.53, 47.18, 54.85, 78.33, 80.92, 155.30,
164.23, 170.03. HRMS (Electrospray +) calculated for the
protonated molecular ion (MH+) C17H34N3O4S 376.2270, found
376.2262.
Drug Substance (1). (2S)-2-{[(2R)-2-Amino-2-carboxyethyl]-
thio}-N-[(1Z)-1-aminoethylidene]-1-propanaminium Phosphate
Monohydrate. To a solution of 20 kg of amidine salt 20 in water
(70 L, 3.5 vol) was added toluene (70 L, 3.5 vol) and the
mixture was stirred vigorously and cooled to 5 ( 2 °C. A
solution of potassium carbonate (24 kg, 1.2 wt) in water (24 L,
1.2 vol) was added over 15 min at 5 ( 2 °C, and the mixture
was stirred at the same temperature for 30 min. The phases
were allowed to settle, and the aqueous phase was removed.
1
repeated once more before H NMR was run to ensure that
ethanol levels were acceptable (<5% w/w). Ethyl acetate (340
L, 5 vol) and 10% sodium carbonate solution (748 L, 11 vol)
were added, and after a 10 min stirring period, the phases were
separated. The aqueous phase was extracted with ethyl acetate
Vol. 13, No. 4, 2009 / Organic Process Research & Development
•
779

To the organic phase was added a solution of phosphoric acid
(5 kg, 0.25 wt) in water (48 L, 2.4 vol) over 10 min with the
temperature being kept below 5 ( 3 °C. The biphase was then
stirred at 72 ( 2 °C for 18 h and cooled to 20 °C, and then the
layers were separated. The aqueous phase was added over 2 h
to a preheated mixure of acetone (80 L, 4 vol) and water (16
L, 0.8 vol) at 40 ( 2 °C containing 60 g (0.003 wt) of seed of
1. The temperature was maintained at 40 ( 2 °C throughout
the addition, and the resulting suspension was stirred at 40 (
2 °C for 15 min. Further acetone (48 L, 2.4 vol) was added
over 60 min to the suspension at 40 ( 2 °C, and the mixture
was stirred at 40 ( 2 °C for 15 min before it was cooled to 5
( 2 °C and held at this temperature for 1 h. The slurry was
filtered, and the cake was washed with acetone/water 3:1 (2 ×
50 L) and then with acetone (3 × 40 L). The cake was
deliquored for 60 min, and the product was dried in a Boltz
dryer at 60 °C and at 10 mbar to constant probe temperature.
Anhydrous 1 was then rehumidified over 2 h by “tumbling” in
the presence of wet nitrogen (wet nitrogen stream was produced
by passing nitrogen gas through 20% w/v aqueous sodium
chloride solution). Isolated was 12.2 kg of 1; 74.1% th. yield,
99.7% pure by 1H NMR, >99.95% a/a by chiral HPLC, water
by KF 5.35% w/w, th. 5.36 w/w, phoshate by IC 29.85% w/w,
th. 29.22% w/w.
1H NMR in D2O (HDO as reference at 4.80 ppm) δ 1.35(3H)
d J ) 6.9 Hz, 2.31 (3H) s*, 3.14-3.29 (3H) m, 3.42-3.56
(2H) m, 4.00-4.04 (1H) m (* ) This signal has an associated
minor rotameric peak at 2.34 ppm). 13C NMR in D2O (TSP as
reference at 0.00 ppm) δ 20.92, 21.35, 33.10, 40.73, 49.95,
56.60, 168.17, 175.62. HRMS (Electrospray +) calculated for
the protonated molecular ion (MH+) C8H18N3O2S 220.1120,
found 220.1111.
Acknowledgment
We acknowledge all GSK staff involved from Strategic
Technologies, Physical Properties, Analytical Sciences and the
Pilot Plants at Stevenage (UK), Tonbridge (UK), and Cork
(Ireland) that helped to make this possible in less than 2 years.
Received for review April 29, 2009.
OP900108B
780
•
Vol. 13, No. 4, 2009 / Organic Process Research & Development