Development of a pilot-plant process for a nevirapine analogue HIV NNRT inhibitor

Article abstract of DOI:10.1021/op8000756

The pilot-plant synthesis of nevirapine analogue 1 is described. The compound was prepared in eight steps from substituted pyridine raw materials and 4-hydroxyquinoline. The key transformation involves a novel one-pot conversion of an arylhalide to arylac

Full text of DOI:10.1021/op8000756

Organic Process Research & Development 2008, 12, 603–613
Development of a Pilot-Plant Process for a Nevirapine Analogue HIV NNRT Inhibitor
,
‡
‡
‡
‡
‡
‡
‡
Carl A. Busacca,* Mike Cerreta, Yong Dong, Magnus C. Eriksson, Vittorio Farina, XuWu Feng, Ji-Young Kim,
‡
‡
‡
‡
‡
†
†
Jon C. Lorenz, Max Sarvestani, Robert Simpson, Rich Varsolona, Jana Vitous, Scot J. Campbell, Mark S. Davis,
†
†
†
§
§
§
⊥
Paul-James Jones, Daniel Norwood, Fenghe Qiu, Pierre L. Beaulieu, Jean-Simon Duceppe, Bruno Hach e´ , Jim Brong,
⊥
⊥
⊥
⊥
⊥
Fang-Ting Chiu, Tom Curtis, Jason Kelley, Young S. Lo, and Tory H. Powner
Department of Analytical Sciences and Department of Chemical DeVelopment, Boehringer-Ingelheim Pharmaceuticals Inc.,
9
00 Ridgebury Road, Ridgefield, Connecticut 06877, U.S.A., Boehringer-Ingelheim Canada Limited, 2100 Cunard Street, LaVal,
Quebec H7S 2G5 Canada, and Boehringer-Ingelheim Chemicals Inc., 2820 North Normandy DriVe, Petersburg,
Virginia 23805, U.S.A.
Abstract:
Scheme 1. Retrosynthesis of 1
The pilot-plant synthesis of nevirapine analogue 1 is described.
The compound was prepared in eight steps from substituted
pyridine raw materials and 4-hydroxyquinoline. The key trans-
formation involves a novel one-pot conversion of an arylhalide to
arylacetic acid under palladium catalysis, followed by regioselective
reduction via in situ generated BH /THF to the arylethanol
3
intermediate 2. All stages were carried out on 10-150-kg scale.
Introduction
Nevirapine is a non-nucleoside reverse transcriptase inhibitor
marketed for the treatment of HIV infection. Compound 1 is
an analogue of nevirapine which was selected for further
1a
evaluation. The structures of nevirapine and analogue 1 are
shown in Scheme 1. The tricyclic cores of each are nearly
identical, the only differences being migration of the methyl
substituent from the 4- to the 5-position, and replacement of
the N-cyclopropyl substituent by N-ethyl in 1. The principal
change is the installation at the 8-position of the two carbon
ether linkage with a 4-quinoline-N-oxide terminus. Tricyclic
modichloroamide 5 in nearly quantitative yield. Nucleophilic
aromatic substitution with ethylamine then led regiospecifically
to amide 6. Cyclization and N-methylation were achieved in a
one-pot transformation with NaHMDS as base, furnishing
bromide 3 in 80% yield for the step. Removal of excess MeI
from the waste stream was achieved by distilling the volatiles
into a solution of 4-picoline, which was readily quaternized.
Haloaminopyridine derivatives are often highly colored by low
levels of impurities, and charcoal treatment of 3 was required
to provide the advanced intermediate as a yellow solid. The
route described in Scheme 2 was used to prepare 350 kg of
bromide 3.
1
halides similar to 3 were recognized early on as valuable
platforms for functionalization of the 8-position in these systems,
and thus bromide 3 was considered to be a potential advanced
intermediate for the synthesis of 1.
Results and Discussion
Condensation of the acid chloride of 5-bromo-2-hydroxyni-
1d
cotinic acid 4 with 2-chloro-3-aminopyridine furnished bro-
*
Author for correspondence. E-mail: carl.busacca@boehringer-ingelheim.
Attempts at halogen metal exchange using bromide 3 were
completely unsuccessful, including the use of alkyl lithiums,
magnesium metal, and organomagnesium reagents which have
com. Telephone: (203)-798-4126. Fax: (203)-791-6130.
†
Department of Chemical Development, Boehringer-Ingelheim Pharmaceu-
ticals Inc.
‡
Department of Analytical Sciences, Boehringer-Ingelheim Pharmaceuticals
2a,b
specifically been used for the generation of pyridyl Grignards.
Inc.
§
Boehringer-Ingelheim Canada.
Boehringer-Ingelheim Chemicals Inc.
The shortcomings of this approach are well illustrated by the
reaction of 3 with i-PrMgBr with electrophile quenching shown
in Scheme 3.
⊥
(
1) (a) Simoneau, B. U.S. Patent 6,420,359, 2002. (b) Klunder, J. M.;
Hoermann, M.; Cywin, C. L.; David, E.; Brickwood, J. R.; Schwartz,
R.; Barringer, K. J.; Pauletti, D.; Shih, C.-K.; Erickson, D. A.; Sorge,
C. L.; Joseph, D. P.; Hattox, S. E.; Adams, J.; Grob, P. M. J. Med.
Chem. 1998, 41, 2960. (c) Grozinger, K. G.; Byrne, D. P.; Nummy,
L. J.; Ridges, M. D.; Salvagno, A. J. Heterocycl. Chem. 2000, 37,
Rather than effecting the desired bromine-magnesium
exchange, directed metalation at the 7-position occurs, and the
2c
tricycle 4a with the bromine intact was isolated. Use of
stronger bases/higher temperatures or diorganomagnesium
reagents invariably led to complex mixtures where ring opening
of the lactam was a common reaction motif. Palladium-
catalyzed cross coupling of bromide 3, however, was found to
be extremely general, with standard Heck, Stille, Suzuki, and
2
29. (d) Beaulieu, P. L.; Duceppe, J.-S.; Haché, B. Synthesis 2002, 4,
5
28.
(
2) (a) Knochel, P.; Bonnet, V.; Mongin, F.; Tr e´ court, F.; Breton, G.;
Marsais, F.; Qu e´ guiner, G. Synlett 2002, 6, 1008. (b) Tr e´ court, F.;
Breton, G.; Bonnet, V.; Mongin, F.; Marsais, F.; Qu e´ guiner, G.
Tetrahedron 2000, 56 (10), 1349. For a very similar finding see: (c)
Pierrat, P.; Gros, P.; Fort, Y. Synlett 2004, 13, 2319.
1
0.1021/op8000756 CCC: $40.75

2008 American Chemical Society
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
603
Published on Web 06/28/2008

Scheme 2. Synthesis of bromide 3
Scheme 4. HST route
Scheme 3. Metalation of bromide 3
Since we could perform the cross coupling with a very
inexpensive monodentate ligand, more expensive ligands, such
as bidentate bisphosphines, were not examined. Silane 8 was
used crude in the following step in a one-pot conversion to
alcohol 2. The material was first converted to the intermediate
fluorosilane (not isolated) by treatment with 2 equiv of
4
BF
3
·2HOAc at 65 °C for 2-4 h, followed by cooling to
ambient temperature and charging 3 equiv of sodium perborate.
It was found that 18 h stirring at ambient temperature completed
the Tamao-Fleming oxidation to alcohol 2. We first evaluated
peracetic acid for this oxidation and found the reaction to be
very exothermic and hazardous. Calorimetry showed that a
significant portion of the energy involved was caused by
decomposition of peracetic acid by residual palladium (∼2000
ppm) contaminating silane 8. The reaction could be performed
by slow (∼8 h) addition of peracetic acid and triethylamine,
yet we sought a safer alternative. While sodium perborate has
1
Sonogashira reactions successful in nearly all cases. We
therefore looked to develop an efficient route from 3 to two-
carbon alcohol 2 using palladium C-C bond formation.
Hydroboration/Suzuki/Tamao (HST) Route. There is no
cross-coupling methodology to directly couple an ethanol
fragment to an arylhalide. We therefore sought a two-pot
method to effect this transformation. Silicon is a well-established
surrogate for oxygen, accessible through the Tamao/Fleming
3
oxidation of activated silanes. Our first pilot-plant route to
5
been used for the oxidation of organoboranes to alcohols, we
alcohol 2 therefore commenced with hydroboration/Suzuki
coupling utilizing vinylsilane 7 as raw material as shown in
Scheme 4. The vinylsilane chosen possessed the p-methoxy-
phenyl group to both facilitate protodesilation and to generate
anisole rather than the carcinogen benzene after its reaction.
Hydroboration with 9-BBN in toluene was regiospecific, and
following in situ formation of the borate with aqueous sodium
are not aware of any published use of this inorganic oxidant
for the Tamao oxidation. Although the adiabatic temperature
rise was similar for the two oxidants, the decomposition onset
temperature for the sodium perborate reaction was ∼55 °C,
significantly higher than peracetic acid containing residual Pd.
Sodium perborate proved to be a very safe reagent as direct
charging of the entire quantity led to no significant exotherm
using acetic acid as solvent. The reaction may well be controlled
hydroxide, Suzuki coupling with bromide 3 using Pd(OAc)
PPh proceeded smoothly in a one-pot operation to furnish silane
in 90% yield. Optimization of this step consisted primarily
2
/
3
8
(4) (a) Kn o¨ lker, H.-J.; Wanzl, G. Synlett 1995, 378. (b) Fleming, I.;
Henning, R.; Plaut, H. J. Chem Soc., Chem. Commun. 1984, 29 See
also ref 5a.
of minimizing the formation of reduced impurity 9. This was
achieved by performing the reaction in a nonpolar solvent
(5) (a) Kabalka, G. W.; Maddox, J. T.; Shoup, T. Org. Synth. 1996, 73,
1
16. (b) Kabalka, G. W.; Shoup, T. M.; Goudagon, N. M. J. Org.
(toluene), and especially by having a very high ligand-to-
Chem. 1989, 54, 5930. (c) McKillop, A.; Sanderson, W. R. Tetrahe-
dron 1995, 51, 6145.
palladium ratio. A plot of log (8/9) as a function of this
parameter is shown in Figure 1. Only at high (∼20:1) ratios
could reduction be held to 4-5%.
(
6) (a) Buchwald, S. L.; Moradi, W. A. J. Am. Chem. Soc. 2001, 123,
7
996. (b) Buchwald, S. L.; Fox, J. M.; Huang, X.; Chieffi, A. J. Am.
Chem. Soc. 2000, 122, 1360. (c) Buchwald, S. L.; Old, D. W.; Wolfe,
J. P.; Palucki, M.; Kamikawa, K. U.S. Patent 6,307,087, CAN 135:
318588, 2001. (d) Hartwig, J. F.; Hamann, B. C. U.S. Patent 6,057,456,
CAN 132:293568, 2000. (e) Hartwig, J. F.; Kawatsura, M. U.S. Patent
6,072,073, CAN 133:30376, 2000. (f) Busacca, C. A.; Eriksson, M. C.;
Kim, J.-Y. U.S. Patent 6,759,533, CAN 140:59670, 2004. (g) Beare,
N. A.; Hartwig, J. F. J. Org. Chem. 2002, 67, 541.
(
3) (a) Fleming, I.; Henning, R.; Parker, D. C.; Plaut, H. E.; Sanderson,
P. E. J. J. Chem. Soc., Perkin Trans. 1 1995, 317. (b) Jones, G. R.;
Landais, Y. Tetrahedron 1996, 52, 7599. (c) Hosomi, A.; Iijima, S.;
Sukarai, H. Chem. Lett. 1981, 243. (d) Tamao, K. AdVances in Silicon
Chemistry; JAI Press: Greenwich, CT, 1996; Vol. 3, pp 1-62.
6
04
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

under palladium catalysis using triphenylphosphine as ligand.8
We thus first examined an assortment of these materials as
shown in Chart 1, and all coupled in high yield to bromide 3
2 3
using Pd(OAc) /PPh and NaH as base, using essentially the
9
Beletskaya protocol. The malonate surrogate carbon acids were
thus more “forgiving” than the malonates themselves with
respect to ligand tolerance. We settled on cyanoisopropylacetate
as the carbon acid of choice for our synthesis of alcohol 2. As
shown in Scheme 6, slow addition (2 h) of the carbon acid at
60 °C to a mixture of all other components in toluene allowed
control over both the formation of the anion and its accompany-
ing evolution of hydrogen and heat. The mixture was then
heated at 100 °C for 1.5 h to complete the C-C bond-forming
event furnishing 15.
Figure 1. Suzuki chemoselectivity as a function of phosphine-
to-palladium ratio.
THF and dioxane could also be used as solvent, and tertiary
alkoxide bases could be substituted for NaH as well. We
ultimately chose NaH/oil beause of its lower unit cost, its
successful use in the manufacture of Nevirapine, and our
experience in the safe handling of NaH at our production
facility.
by the slow reaction of sodium perborate with acetic acid,5c
generating the active oxidant. We were able to reproducibly
prepare alcohol 2 in ∼65% yield using this procedure, and the
two-pot conversion of 3 to 2 described was used to produce 70
kg of the desired alcohol.
The hydroboration/Suzuki/Tamao route did present some
disadvantages, however. The reaction mass efficiency (RME)
was low, due to the high molecular weight of both BBN and
the silane substituents. In addition, the workup of the Tamao
oxidation required a large number of manipulations in the pilot
plant. Finally, the first crop yield was only ∼50%, and a silica
gel filtration was required for second crop isolation to give 65%
overall yield for the step. Therefore, we examined other efficient
routes to introduce the ethanol fragment of 2 from bromide 3.
Malonate Route. A three-step sequence from bromide 3 to
alcohol 2 utilizing malonate arylation was developed, as shown
in Scheme 5. Diethylmalonate was arylated under palladium
We then needed to carry out a complete hydrolysis/
decarboxylation to carboxylic acid 16. At low pH, the conver-
sion was poor and the reaction sluggish, yet under basic
conditions, 16 could be prepared in high yield. We found that
the hydrolysis reaction worked best under biphasic conditions,
and thus when the arylation was complete we simply quenched
residual NaH with i-PrOH, then added 1 M NaOH and heated
at 80 °C for 8 h. This furnished the desired acid 16, yet in
several laboratory runs with two different ester starting materials,
10-35% of a byproduct was observed. This material was
isolated and characterized, and then shown by independent
synthesis (palladium-catalyzed carbonylation of bromide 3 with
H O) to be the one-carbon carboxylic acid 17. This byproduct
2
6a
catalysis, although only the Buchwald biphenyl ligands and
appears to be formed by air oxidation of the enolate of 15 when
6g
t-Bu
ligand screen.
Krapcho monodecarboxylation of malonate 10 gave ester
3
P (12-14) were found to be successful after an extensive
10
reactor inertion was incomplete. We were able to control this
impurity to <1% by careful inertion of the reactor prior to the
hydrolysis/decarboxylation. The one-pot conversion of bromide
3 to acid 16 was then accomplished in 90-92% yield. Several
observations were made during the optimization of this step.
First, alcohol quenches of residual NaH were absolutely
essential. Quenching under aqueous conditions at any pH led
to insoluble organic “balls” that were broken up only with great
difficulty, while i-PrOH and MeOH quenches led to readily
stirrable slurries.
The concentration and quantity of base were also
carefully studied. At least 3.5 equiv of hydroxide was
required to effect complete conversion to acid 16, and
concentrations in excess of 1 M led to decomposition
7
1
1, which could be reduced with 6 equiv of sodium borohydride
in ethanol at 50 °C to the desired alcohol 2 in 63% overall
yield. Although this approach eliminated 9-BBN and the Tamao
workup problems, it presented many issues of its own. The
3
ligand t-Bu P is expensive and extremely pyrophoric and was
not considered suitable for plant use. Although the malonate
chemistry was successful, it requires the use of costly, special-
ized ligands and procedures which are also covered by
6c–e
patents. We also examined cross coupling of a Reformatsky
reagent to give an ester precursor to alcohol 2. The major
product observed, however, was the symmetric dimer of the
tricycle. In addition, the desired Reformatsky reagent was not
available in bulk, so this route would have required additional
steps, reactors, and time. For these reasons, the malonate and
Reformatsky routes were thus rejected, leading to the creation
(
8) (a) Uno, M.; Seto, K.; Ueda, W.; Masuda, M.; Takahashi, S. Synthesis
1
985, 506. (b) Yamanaka, H.; Sakamoto, T.; Katoh, E.; Kondo, Y.
Chem. Pharm. Bull. 1988, 36, 1664. (c) Yamanaka, H.; Sakamoto,
T.; Katoh, E.; Kondo, Y. Heterocycles 1988, 27, 1353. (d) Yamanaka,
H.; Sakamoto, T.; Katoh, E.; Kondo, Y. Chem. Pharm. Bull. 1990,
38, 1513. (e) Yamanaka, H.; Sakamoto, T.; Kondo, Y.; Suginome,
T.; Ohba, S. Synthesis 1991, 552. (f) Sakamoto, T.; Kondo, Y.;
Masumoto, K. Heterocycles 1993, 36, 2509 For a review see: (g) Prim,
D.; Campagne, J.-M.; Joseph, D.; Andrioletti, B. Tetrahedron 2002,
6f
of our current pilot-plant synthesis.
Malonate Surrogate Route. It has been known for some
time that carbon acids very similar to malonates can be arylated
5
8, 2041.
(
9) Beletskaya, I. P.; Kashin, A. N.; Mitin, A. V.; Wife, R. Tetrahedron
Lett. 2002, 43, 2539.
(
10) Treatment of the enolate of 15 in a pressure vessel under 100 psi O2
led to acid 17 in high yield.
(
7) Krapcho, A. P. Synthesis 1982, 805.
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
605

Scheme 5. Malonate route
Chart 1. Carbon acids arylated with bromide 3
Reduction of acid 16 to alcohol 2 could be readily achieved
under a variety of conditions. Using 1.4 equiv of BH /THF
3
furnished the product in 85% yield without any over-reduction
of the lactam to the tertiary amine. Alternatively, acid 16 could
be treated first with 1 equiv of catecholborane, followed by
addition of 1 equiv of sodium borohydride in the same pot to
generate the alcohol in similar yield. Borane and catecholborane
are far more expensive than sodium borohydride, though, and
require special material handling. Borane can be generated in
situ under a variety of conditions, typically involving the
interaction between a Lewis or protic acid and sodium boro-
products arising from hydrolysis of the lactam linkage.
The most surprising result was the low rate of hydrolysis
when performed under homogeneous conditions using, for
example, THF as solvent for both arylation and hydrolysis.
Despite the success of “classical” ester saponification
conditions using THF/aqueous base, this was a very poor
choice for the generation of carboxylic acid 16. Both 15
and 16 are soluble in 1 M NaOH, and the two-phase
conditions may simply sequester all organic impurities and
byproducts away from these components in the toluene
phase. Alternatively, the ternary azeotrope of i-PrOH/
13
hydride. Addition of a variety of Lewis acids (BF
ZnCl ) to acid 16 led to thick mixtures that were very difficult
to stir under any conditions, so we focused on protic acids.
When commercial solutions of concentrated HCl and H SO
3 3
, BCl ,
1
1
2
2
4
were used, the reaction mixtures were also extremely thick. We
found that simply treating a THF mixture of 16 with 1.5 equiv
of anhydrous HCl/THF, followed by slow addition (2 h) of 1.5
4
toluene/H
2
O might be involved, essentially removing some
equiv of NaBH /diglyme, led after a 3-h hold to complete
i-PrOH and thereby facilitating the hydrolysis. We were
curious what the mechanism of the hydrolysis might be,
specifically whether the ester or nitrile was hydrolyzed
first, or whether they were hydrolyzed at similar rates.
conversion to alcohol 2. NMR experiments in HCl/d -THF
established that no acid-promoted decomposition of THF would
occur for more than 48 h at ambient temperature. Slow addition
8
of NaBH was required to prevent lactam reduction. The
4
1
3
We therefore prepared the C-labeled isopropylcyano-
mixture was quenched with MeOH, volatiles were removed in
1
2a,b
13
acetate
1
and processed the material through to acid C-
vacuo, and the product then precipitated from the two-phase
mixture of 0.5 M NaOH/heptane and centrifuged to furnish the
product in 80-85% yield. This two-pot conversion of bromide
3 to alcohol 2 was successfully carried out on a 40kg scale.
A number of different strategies were initially examined to
convert alcohol 2 to target 1. A representative group of results
N
1
3
6. As shown in Scheme 7, C NMR and HRMS showed
1
3
greater than 95% C retention in the product. The
isopropylester is thus clearly hydrolyzed first, and appar-
ently then undergoes decarboxylation, followed by nitrile
1
2c
hydrolysis in a stepwise manner.
When the one-pot arylation/hydrolysis was completed, we
simply separated the aqueous phase, washed once with EtOAc,
and then adjusted the pH to 3.3 to cause crystallization of acid
is shown in Table 1, where S Ar reactions using both 4-nitro-
and 4-chloro-quinoline-N-oxide were studied. With NO as
2
leaving group (entries 1, 2), 1 was formed in low yield,
16 from water. Simple centrifugation and drying under vacuum
accompanied by elimination product 18. A range of bases and
solvents were screened with the chloro electrophile. In general,
organic bases were completely ineffective (entries 4-7), and
most inorganic bases gave low conversion. The best result found
then furnished 16 in high yield with HPLC purity >98%.
(
(
11) For recent examples, see: (a) Mannekens, E.; Tourwe, D.; Lubell,
W. D. Synthesis 2000, 9, 1214. (b) Springer, D. M.; Luh, B.-Y.;
Bronson, J. J.; McElhone, K. E.; Mansuri, M. M.; Gregor, K. R.;
Nettleton, D. O.; Stanley, P. L.; Tramposch, K. M. Bioorg. Med. Chem.
2
was with CsOH ·H O in DME (entry 12), where 61% conver-
2
000, 8, 1087.
(13) (a) Abiko, A.; Masamune, S. Tetrahedron Lett. 1992, 33, 5517. (b)
Periasamy, M.; Prasad, A. S. B.; Kanth, J. V. B. Tetrahedron 1992,
48, 4623. (c) Meyers, A. I.; Dickman, D. A.; Smith, G. A.; Gawley,
R. E. Organic Syntheses; Wiley & Sons: New York, 1990; Collect.
Vol. VII, p 530.
12) (a) Holmes, D. L.; Lightner, D. A. Tetrahedron 1996, 52, 5319. (b)
Silverman, R. B.; Lu, X.; Blomquist, G. D.; Ding, C. Z.; Yang, S.
Bioorg. Med. Chem. 1997, 5, 297. (c) In situ React-IR might be suitable
for monitoring this hydrolysis as well.
6
06
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

Scheme 6. Malonate surrogate route
Scheme 7. 13C-Labeling experiment
sion was obtained. We were unable to obtain full conversion,
however, and the use of reaction temperatures even slightly
above ambient led to significantly more formation of the
elimination product 18. 4-Chloroquinoline was also investigated
as an electrophilic partner. Conversions were far lower than
with the analogous N-oxide, however, and elimination was a
serious side reaction. Alcohol 2 was also converted to a variety
of leaving groups (mesylate, tosylate, brosylate, acetimidate,
etc.) to try and use this fragment as the electrophilic partner
with 4-hydroxyquinoline as nucleophile. The best isolated yield
of ether 19 under any conditions was ∼40%, however, with
significant elimination again plaguing the transformation. In all
amount of 20 produced was minimized to ∼4%. The product
was conveniently purified by simple precipitation of the
crystalline HCl salt, and 19 was isolated in 80% yield as a
nonhygroscopic, free-flowing solid. The Mitsunobu was suc-
cessfully executed on a 110 kg scale.
Target 1 was then produced from 19 with m-CPBA using a
two phase aqueous Na CO /CH Cl solvent system. It was
2 3 2 2
found to be critical to maintain high pH (∼10) throughout the
15
oxidation to ensure complete conversion. Since m-chloroben-
zoic acid is produced as the reaction proceeds, the pH steadily
drops over time, necessitating a large initial base charge. The
a
quinoline nitrogen (estimated pK 6) was selectively oxidized
N
S Ar attempts, conditions more vigorous than those in Table 1
with no competing oxidation of the less basic nitrogens (pK
a
led to elimination product 18 as the predominant species. We
decided therefore to optimize the Mitsunobu etherification
of alcohol 2 as shown in Scheme 8.
2.5, determined on the API after oxidation), and the reaction
was complete in 90 min at ambient temperature. The reaction
was then quenched (Na SO ), and organic solvents removed
2 3
1a,6f
Initial optimization showed the principal byproduct in the
in vacuo, leaving an aqueous slurry of 1. Analysis of the
impurities in the crude drug substance revealed two in particular,
lactam 21 and anilinoester 22. Typical levels of these two
impurities in the API were 0.20 A% and 0.15 A% by HPLC,
respectively. The lactam was formed via the known photore-
reaction to be the regioisomeric Mitsunobu adduct 20. 4-Hy-
14
droxyquinoline is of course an ambident nucleophile, and the
task rapidly reduced to one of establishing high regiocontrol
2 2
for ether 19. THF, DME, CH Cl , and DMF were evaluated
16
for the etherification at ambient temperature, and DME gave
far higher ratios of 19:20 (ca. 15:1) than any of the other
solvents. The amount of 20 was still unacceptably high, though,
so we lowered the initial reaction temperature to -25 °C and
ramped to ambient temperature over 12 h. In this way the total
arrangement of aromatic N-oxides, and photolysis of a solution
of 1 was shown to generate this material cleanly. The exact
(
(
15) For a possible explanation, see: (a) Houk, K. N.; Washington, I. Org.
Lett. 2002, 4, 2661.
16) (a) Hata, N. Bull. Chem. Soc. Jpn. 1986, 59, 2723. (b) Hata, N.;
Norisuke, Y.; Yagi, A. Chem. Lett. 1983, 309. (c) Kaneko, C.; Okuda,
W.; Karasawa, Y.; Somei, M. Chem. Lett. 1980, 547.
(
14) Black, T. H. Org. Prep. Proced. Int. 1989, 21, 179.
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
607

Table 1. S
N
Ar approaches to 1
procedure: Water (88 kg) and sodium bromide (44.53 kg, 432.8
mol, 1.505 equiv) were charged to a 120 L glass-lined reactor
and stirred under an inert atmosphere. Sodium hypochlorite
(269.2 kg of a freshly titrated 10.8 wt % solution, 345.1 mol,
1.2 equiv) was added to a 800 L glass-lined reactor and cooled
to 5 °C. The aqueous NaBr solution was then slowly transferred
to this reactor with cooling so that the internal temperature did
not rise above 20 °C. A second 800 L glass-lined reactor was
charged with 56.6 kg (707.4 mol, 2.46 equiv) of 50% aqueous
NaOH, followed by 40 kg (287.5 mol, 1.0 equiv) of 2-hydrox-
ynicotinic acid and the solution cooled to 5 °C. The aqueous
solution of sodium hypobromite was added to the solution of
2
-hydroxynicotininic acid at such a rate that the internal
entry
R
base
solvent
% conv.a
% 18
temperature did not rise above 20 °C. After the addition was
complete, the reaction mixture was stirred for 90 min. After
this time the reaction mixture was quenched by the sequential
addition of 120.3 kg of water, 116.3 kg (1002 mol, 3.5 equiv)
of 31.5% HCl, and 15.62 kg of isopropanol. The desired product
precipitated as a white solid which was collected by filtration.
The filter cake was then slurried with an additional 400 kg of
water and stirred for 30 min, then filtered a second time. The
filter cake and methanol (160 L) were then added to a 800 L
reactor and heated to reflux for 1 h, and allowed to cool to
between 0 and 5 °C over a 3 h period and maintained at that
temperature for an additional 4 h. The resultant slurry was then
filtered and dried at 70 °C at 20 torr to constant weight. This
afforded 39.1 kg (74%) of 5-bromo-2-hydroxynicotinic acid 4.
1
2
3
4
5
6
7
8
9
NO
NO
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
2
NaH
Cs CO
NaHCO
DBU
DIPEA
TEA
DME
8.5
20
0
5
2
2
3
MeCN
DMSO
DMSO
DMSO
DMSO
DMSO
3
NR
NR
NR
NR
NR
NR
35
NA
NA
NA
NA
NA
NA
ND
0
2,6-lutidine
NaOH/PTC
2 2
CH Cl
Cs
2
CO
3
DMSO
DME
DME
DME
1
1
1
0
1
2
NaOH
KOH
43
46
0
0
CsOH ·H
2
O
61
a
HPLC % conversion to 1.
mechanism of formation of 22 is not known, yet it might arise
from oxidative degradation of 21. The structure of 22 was
(
For characterization, see ref 1d).
5-Bromo-2-chloronicotinoyl Chloride, Solution in Aceto-
17
unequivocally established by independent synthesis. The drug
substance was isolated by centrifugation in 86% yield.
In summary, we have developed a four-step process for the
drug substance production of drug candidate 1 from advanced
intermediate bromide 3, which is itself produced in four steps
from bulk pyridine building blocks. The chemistry employs as
a key transformation the one pot conversion of aryl bromide 3
to arylacetic acid 16 under palladium catalysis using inexpensive
raw materials and nonproprietary ligands.
nitrile. An 800 L glass-lined reactor was charged with 416 L
of toluene, 1.37 kg of DMF (18.8 mol, 0.05 equiv), and 83.3
kg (382.0 mol, 1.0 equiv) of 5-bromo-2-hydroxynicotinic acid
4
, and the resulting slurry heated to 60 °C. Thionyl chloride
136.4 kg, 1146 mol, 1.5 equiv) was then added over a 4 h
period at a rate which kept the internal temperature below 70
C. After the addition was complete, the mixture was heated
to 100-105 °C for 3 h. The mixture was then heated to 125
C, and the volatile materials were removed by distillation.
(
°
°
When the distillation process slowed, the reaction was allowed
to cool to ∼20 °C, then the internal pressure was lowered to
50 torr and the mixture heated to 100 °C, and further volatile
materials were removed. Thirty kilograms of acetonitrile was
then added while distilling solvents from the reactor while
maintaining the temperature between 95-100 °C until no more
distillate was received at 100 °C. While maintaining the
temperature of the reactor at >40 °C, 37.04 kg of aceto-
nitrile was then added. The resultant solution was shown to
contain 97.4 kg (>99% yield, >380 mol) of 5-bromo-2-
chloronicotinoyl chloride. The solution was used in the subse-
quent step without further purification.
Experimental Section
General. HPLC analyses were performed under standard
gradient conditions on an Agilent 1100 using a Zorbax SB-CN
column, 150 mm × 4.6 mm, 5 µm particle size; flow rate 1.5
mL/min, temp 50 °C, detection @ 220 nm, 10 µL injection
volume, run time 57 min. Mobile phase A: 80/20 (v/v) H
MeOH with 0.05% TFA; Mobile phase B: 35/65 (v/v) H
MeOH with 0.05% TFA. Gradient: 100% A to 100% B in 43
min., 10 min. hold at 100%B. Retention times (RT, min) and
relative retention times (RRT) for the key compounds are as
follows: (RT, RRT): 4-OH-quinoline (3.8, 0.16); m-Cl-benzoic
acid (9.1, 0.39); alcohol 2 (9.9, 0.43); ether 19 (15.6, 0.67);
2
O/
2
O/
5
-Bromo-2-chloronicotinoyl-N-(2-chloropyridin-3-yl)ni-
cotinamide (5). An 800 L glass- lined reactor was charged with
60 L of acetonitrile, 88.0 kg (1047 mol, 3.21 equiv) of
NaHCO , and 41.68 kg (324.4 mol, 1.0 equiv) of 3-amino-2-
chloropyridine and the solution mixture stirred for 30 min at
5 °C. The solution of 5-bromo-2-chloronicotinoyl chloride in
Ph
.15); impurity 21 (27.2, 1.17).
-Bromo-2-hydroxynicotinic Acid (4). An aqueous solution
of sodium hypobromite was generated using the following
3
PdO (18.9, 0.81); API 1 (23.2, 1.00); impurity 22 (26.7,
1
4
5
3
2
(
(
17) See Experimental Section.
18) Busacca, C. A.; Cerreta, M.; Varsolona, R.; Smoliga, J.; Lorenz, J.
Vitous, J. U.S. Patent 7,309,700, CAN 143:353310, 2007.
acetonitrile, 140.95 kg of a 73% solution (85.6 kg, 334.1 mol,
1.03 equiv), was added over a 4 h period at a rate which
6
08
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

Scheme 8. Mitsunobu etherification
maintained the internal reaction temperature between 25 and
6
(100 MHz, DMSO-d ) δ: 165.88 (s), 156.10 (s), 152.28 (d),
34 °C. The mixture became thinner as the reaction progressed.
146.99 (d), 146.74 (s), 139.05 (d), 137.37 (d), 131.65 (s), 123.40
After the addition was complete, the mixture was allowed to
stir for 18 h at ambient temperature. After this time an aliquot
was removed and analyzed by HPLC. This showed 5.4 area %
residual 3-amino-2-chloropyridine remained. To consume the
remaining 3-amino-2-chloropyridine, 10.6 kg of the 5-bromo-
(d), 110.05 (s), 103.36 (s), 35.26 (t), 14.53 (q). Anal. Calcd for
C H12BrClN O: C, 43.91; H, 3.40; N, 15.75. Found: C, 43.99;
13 4
H, 3.29; N, 15.72.
2-Bromo-5-ethyl-10-methyl-5,10-dihydro-4,5,6,10-tetraaza-
dibenzo[a,d]cyclohepten-11-one (3). A 4000 L inerted reactor
(reactor 1) was charged with 114 kg of crude aminoamide 6
(321 mol, 1 equiv); 323 kg of THF were then added, and the
temperature of the contents was adjusted to 50 °C. To this
mixture was then added 650 kg of 1 M NaHMDS/THF (719
mol, 2.2 equiv) over 2 h, maintaining the temperature of the
contents below 55 °C. The reactor was maintained at ∼50 °C
for 1 h. HPLC showed incomplete consumption of aminoamide
6, so an additional 59.2 kg of 1 M NaHMDS (66 mol, 0.2 equiv)
was charged. After 1 h at 50 °C, cyclization was complete. The
contents were then cooled to 25 °C, and 119 kg of MeI (836
mol, 2.6 equiv) were then charged over 2 h while maintaining
the contents temperature below 30 °C. The contents were then
agitated at 25 °C for 10 h, at which time HPLC analysis showed
complete conversion to bromide 3. To a second, 6000 L reactor
(reactor 2) was then charged 600 g of 2,6-di-tert-butyl-4-
methylphenol, followed by 168 kg of 4-picoline (1803 mol,
5.6 equiv) and 88 kg of MeOH. The volatiles from reactor 1
were then distilled at 5 kPa into reactor 2 until distillation ceased.
Additional MeOH was then charged (240 kg each time) to
reactor 1, and two additional distillations were performed as
described above. The combined distillates were then heated at
50 °C for 5 h to complete destruction of MeI, in accordance
with earlier runs in 200 L equipment.
2
-chloronicotinoyl chloride solution was added over a 1.5 h
period. Analysis of this mixture showed that the level of
-amino-2-chloropyridine was <1%. Water (1200 kg) and
3
isopropanol (64.38 kg) were added to a separate, clean 2000 L
reactor, and this mixture was cooled to ∼5 °C. The acetonitrile
solution was added to this mixture over a 1 h period while
maintaining the temperature <10 °C, and the resultant slurry
was stirred for 1 h. The mixture was then filtered through a
paper filter, and the filter cake was then slurried with 320 kg
of water, stirred for 30 min, and then filtered again. The filter
cake was dried for 24 h at 50 °C under 25 torr. This afforded
9
7.9 kg (282 mol, 87% yield) of 5 as a tan solid. KF 0.27%
1
H
2
O. Mp 180-182 °C. H NMR (400 MHz, DMSO-d
0.64 (s, 1H), 8.72 (d, J ) 2.3 Hz, 1H), 8.43 (d, J ) 2.2 Hz,
H), 8.28 (m, 1H), 7.52 (dd, J ) 4.7, 7.9 Hz, 1H), 3.36 (br s,
6
) δ:
1
1
1
13
H). C NMR (100 MHz, DMSO-d ) δ: 162.97 (s), 151.28
6
(
1
C
d), 146.60 (d), 145.53 (s), 144.18 (s), 140.56 (d), 135.12 (d),
33.55 (s), 131.34 (s), 123.55 (d), 118.87 (s). Anal. Calcd for
BrCl O·[0.27% H O]: C, 37.88; H, 1.79; N, 12.05.
Found: C, 37.63; H, 1.51; N, 11.78.
-Bromo-N-(2-chloropyridin-3-yl)-2-ethylaminonicotina-
11
H
6
2 3
N
2
5
mide (6). To an inerted 6000 L reactor were charged 150 kg
of dichloroamide 5 (432 mol, 1 equiv) and 792 kg of THF,
and the mixture was cooled to 5 °C. Fifty-nine kilograms of
EtNH2(g) (1309 mol, 3 equiv) was then introduced subsurface
over 4 h, maintaining the contents temperature below 10 °C.
The mixture was agitated at this temperature for 15 min, then
ramped to 90 °C over 2 h, and held at this temperature for 8 h.
The mixture was then cooled to 25 °C, and an HPLC sample
To the contents of reactor 1 at 25 °C were then charged
930 kg of EtOAc and 369 kg of H O, and the contents agitated
2
well for 15 min at 40 °C. The lower aqueous phase was then
dropped to drums, and an additional 369 kg charge (performed
2
2×) of H O was added to the reactor, agitating and dropping
the lower phase to drums as described above. The organic
solvents were then distilled to minimal stirrable volume. To
the reactor were then charged 3.3 kg of Darco G60 and 3.3 kg
of Hyflo Supercel. The reactor was then inerted, and 399 kg of
EtOAc and 383 kg of isooctane were charged, and the contents
were heated at 45 °C for 2 h. The mixture was then filtered
warm through a Seitz filter/inline filter combination. The reactor
was then charged with 295 kg of EtOAc, the contents were
heated to reflux, and this hot solution was then filtered as
described above. The combined filtrates were then distilled at
50 °C and ∼10 kPa until the amount of distillate collected was
N
showed the S Ar reaction was complete. A 2050 kg amount
of H O was then charged, and the pH was adjusted by the
2
addition of 25 kg of HOAc to a final pH of 5.77. The mixture
was stirred 1 h, then centrifuged, using three ∼1000 L washes
of 1% aq HOAc to wash the cake. The crude solid was then
dried at 50 °C under full vacuum for 10 h to give 149.3 kg of
1
aminoamide 6 (97%). Mp 149-151 °C. H NMR (400 MHz,
6
DMSO-d ) δ: 10.33 (s, 1H), 8.30 (m, 2H), 8.11 (t, J ) 5.0 Hz,
1
3
H), 7.97 (d, J ) 7.8 Hz, 1H), 7.47 (dd, J ) 1.2, 7.8 Hz, 1H),
.38 (q, J ) 6.9 Hz, 2H), 1.11 (t, J ) 7.1 Hz, 3H). C NMR
13
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
609

∼
1025 L. The contents were then cooled to 5 °C over 1 h and
then held at 5 °C for 2 h. The resulting slurry was then
the mixture was cooled to 10 °C. The pH was then cautiously
adjusted with 5 M NaOH to a final pH of 8.0, requiring 79 kg
of the caustic solution. The mixture was warmed to 22 °C, 182
centrifuged, and the solid obtained was packed out and dried
at 45 °C under full vacuum for 4 h to give 86 kg of bromide 3
kg of CH
for 20 min. The two-phase mixture was then filtered through a
Celite pad which was previously wetted with CH Cl . The filter
pad was then washed with 71 kg of H O. From the combined
filtrates, the bottom organic layer was then transferred to a
second reactor, while the upper aqueous phase was re-extracted
2 2
Cl was charged, and the mixture was agitated well
1
(80%) as a yellow solid. Mp 144-145 °C. H NMR (500 MHZ,
DMSO-d
6
) δ: 8.51 (d, J ) 2.5 Hz, 1H, H9), 8.21 (dd, J ) 4.6,
.6 Hz, 1H, H2), 8.11 (d, J ) 2.5 H, 1H, H7), 7.84 (dd, J )
.0, 1.6 Hz, 1H, H4), 7.27 (dd, J ) 4.6, 8.0 Hz, 1H, H3), 4.04
2
2
1
8
(
2
13
q, J ) 7.0 Hz, 2H), 3.42 (s, 3H), 1.15 (t, J ) 7.1 Hz, 3H). C
NMR (125 MHz, DMSO-d ) δ: 164.93 (s), 157.74 (s), 153.10
6
with 150 kg of CH
2
Cl
2
. The combined CH
2 2
Cl layers were then
(s), 151.00 (d), 144.24 (d), 142.49 (d), 131.94 (d), 130.94 (s),
2
back-extracted with 100 kg of H O, and the organic solvents
1
21.96 (s), 120.61 (d), 113.04 (s), 40.63 (t), 36.70 (q), 13.23
O: C, 50.47; H, 3.93; N, 16.82;
Br, 23.98. Found C, 50.37; H, 3.59; N, 16.70; Br, 23.69.
-Ethyl-2-(2-hydroxyethyl)-10-methyl-5,10-dihydro-
,5,6,10-tetraazadibenzo[a,d]cyclohepten-11-one (2) via HST
Protocol. An inerted 400 L reactor was charged with 9.1 kg of
9-BBN) (74 mol monomer, 1.4 equiv) and then carefully
were removed by distillation at atmospheric pressure, collecting
201 L of distillate. The reactor was then cooled to 22 °C, and
13.4 kg of MTBE was charged, followed by 5 g of crystalline
alcohol 2. The resulting mixture was agitated at 22 °C for 8 h,
and the resulting slurry was filtered, washing the cake with 40
kg of MTBE. The solid was then packed out and dried at 35
°C to a constant weight of 9.15 kg of alcohol 2 (57.4% first
crop overall yield from bromide 3) as a light-yellow solid. Silica
gel filtration (EtOAc) of the mother liquors could then provide
a further 5-10% product after concentration and seeding as
described above. (For characterization of alcohol 2 see below.)
5-Ethyl-2-{2-[(4-methoxyphenyl)dimethylsilanyl]ethyl}-
10-methyl-5,10-dihydro-4,5,6,10-tetraazadibenzo[a,d]cyclo-
(q). Anal. Calcd for C14 H13BrN
4
5
4
(
2
reinerted before 66 kg of PhMe and 11.9 kg of vinyldimethyl-
p-methoxyphenylsilane (62 mol, 1.16 equiv) were added. The
resulting mixture was then heated to 70 °C and maintained at
that temperature for 2 h, at which time HPLC showed
hydroboration was complete. The solution was then cooled to
2
2 °C, and 46.9 kg of 3.2 M NaOH (134 mol, 2.5 equiv) was
added. The mixture was well agitated for 45 min at 22 °C, and
then 17.8 kg of bromide 3 (53 mol, 1 equiv), 4.22 kg of PPh
16 mol, 0.30 equiv), and 178 g of Pd(OAc) (1 mol, 0.015
1
hepten-11-one (8): mp 103.1-104.3 °C; H NMR (400 MHz,
CDCl ) δ: 8.18 (m, 2H), 7.09 (d, J ) 2.5 Hz, 1H), 7.46 (dd, J
3
3
(
2
) 1.6, 7.9 Hz, 1H), 7.41 (m, 2H), 7.06 (dd, J ) 4.7, 7.9 Hz,
1H), 6.89 (m, 2H), 4.16 (br s, 2H), 3.80 (s, 3H), 3.50 (s, 3H),
2.54 (m, 2H), 1.24 (t, J ) 7.1 Hz, 3H), 1.03 (m, 2H), 0.26 (s,
equiv) were then charged through the manway in the order
given. The resulting mixture was then agitated at 140 rpm at
13
40 °C for 30 min, then linearly heated to 70 °C over 3 h, and
3
6H). C NMR (100 MHz, CDCl ) δ: 167.76 (s), 160.34 (s),
maintained at 70 °C for 10 h. HPLC showed the Suzuki
coupling was complete. The mixture was then cooled to 22 °C,
and 9.1 kg of ethanolamine (149 mol, 2.8 equiv) was added
over 20 min, maintaining the internal temperature below 50
157.52 (s), 155.09 (s), 150.05 (d), 144.22 (d), 140.03 (d), 135.09
(s), 134.89 (d), 131.59 (s), 130.54 (d), 129.10 (s), 120.67 (s),
119.42 (d), 113.57 (d), 54.94 (q), 41.05 (t), 37.32 (q), 26.15
(t), 17.58 (t), 13.54 (q), -3.05 (q). Anal. Calcd for
°C. The mixture was then agitated at 22 °C for 1 h, and then
25 30 4 2
C H N O Si: C, 67.23; H, 6.77; N, 12.54; Found: C, 67.16;
the precipitate of 9-BBN-ethanolamine complex was filtered
off, washing the cake with 31 kg of PhMe. The phases were
then separated, and the lower aqueous layer was extracted with
H, 6.83; N, 12.66.
5-Ethyl-10-methyl-5,10-dihydro-4,5,6,10-tetraazadiben-
zo[a,d]cyclohepten-11-one (9): white solid, mp 99.0-100.5
1
2
0 kg of PhMe. The lower aqueous layer was then discarded
to waste, and the combined organic phases were washed once
with 50 kg of H O. The organic layer was then distilled to
°C; H NMR (400 MHz, CDCl
3
) δ: 8.38 (dd, J ) 2.0, 4.7 Hz,
1H), 8.19 (dd, J ) 1.6, 4.7 Hz, 1H), 8.08 (dd, J ) 2.0, 7.6 Hz,
1H), 7.47 (dd, J ) 1.6, 8.0 Hz, 1H), 7.07 (dd, J ) 4.7, 8.3 Hz,
1H), 6.98 (dd, J ) 4.8, 7.6 Hz, 1H), 4.18 (br s, 2H), 3.49 (s,
3H), 1.24 (t, J ) 7.1 Hz, 3H). C NMR (100 MHz, CDCl ) δ:
167.60 (s), 159.57 (s), 154.81 (s), 150.75 (d), 144.30 (d), 140.97
2
minimal stirrable volume, collecting 127 L of distillate. The
reactor was then cooled to 22 °C to give crude silane 8 as a
thick, light-yellow oil which crystallized on standing. To the
crude silane was then charged 56 kg of HOAc followed by
1
3
3
(d), 131.67 (s), 130.62 (d), 121.08 (s), 119.64 (d), 118.49 (d),
2
0.0 kg of BF
3
·2HOAc (106 mol, 2 equiv). The resulting
41.12 (t), 37.25 (q), 13.55 (q). Anal. Calcd for C14
O]: C, 65.66; H, 5.59; N, 21.88. Found: C, 65.66; H, 5.32;
N, 21.76.
14 4
H N O·[0.1
mixture was then heated to 65 °C and maintained at that
temperature for 4 h and then cooled to 22 °C. To the mixture
H
2
was then added 24.6 kg of NaBO
3
·4H
2
O (160 mol, 3 equiv)
2-(5-Ethyl-10-methyl-11-oxo-10,11-dihydro-5H-4,5,6,10-
tetraaza-dibenzo[a,d]cyclohepten-2-yl)-malonic Acid Diethyl
Ester (10; see ref 6f). A 500 mL two-neck round-bottom flask
fitted with mechanical stirrer, reflux condenser, and thermo-
couple was purged with argon for 20 min, and then 404 mg
through the manway. The mixture was then well agitated at
this temperature for 12 h, and HPLC showed the Tamao
oxidation was complete. To the reactor was then charged 39.8
kg of 1 M K
of 2 M Na
3
PO
4
solution (34.3 mol, 0.65 equiv), then 104 kg
(160 mol, 3 equiv) was added slowly over 45
2
S
2
O
3
2
Pd(OAc) (1.80 mmol, 0.03 equiv), 1.08 g of ligand 12 (3.60
min while maintaining the internal temperature below 30 °C.
The mixture was agitated well for 45 min, and then the volatiles
were removed by distillation at 40 °C/10 mm to give 94 L of
mmol, 0.06 equiv), and 20.0 g of bromide 3 (60.0 mmol, 1
equiv) were charged. To this mixture in an inerted flask were
3 4
added 25.5 g of K PO (120 mmol, 2 equiv), 100 mL of PhMe,
2
distillate. Then to the reactor was charged 107 kg of H O, and
and 13.8 mL of diethylmalonate (90.0 mmol, 1.5 equiv). The
6
10
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

Scheme 9. Synthesis of 1
stirred suspension was placed in an oil bath and heated at reflux
under nitrogen for 24 h until the bromide was consumed by
HPLC. The reaction mixture was cooled to room temperature
and diluted with water (150 mL), and toluene (50 mL) was
added. The mixture was stirred for 1 h at room temperature
until all solids had dissolved. The two-phase solution was then
transferred to a separatory funnel, and the layers were separated.
The aqueous layer was extracted with toluene (50 mL). The
combined organic layers were filtered through a pad of Celite
and concentrated under vacuum to give a viscous oil, which
was evaporated to dryness to furnish an orange solid. HPLC
analysis of this material indicated 83% 10, 8.8% monoester,
and 8.8% reduced impurity 9. Silica gel chromatography (1:1
was cooled to 25 °C, and the phases were separated. The upper
phase was discarded to waste, while the lower aqueous phase
was returned to the reactor; 218 kg of EtOAc was then charged,
and the mixture agitated for 15 min; then the phases were
allowed to separate. The lower aqueous phase was returned to
the reactor and cooled to 10 °C. To this mixture was then slowly
added 6 N H
2
SO
4
with continuous monitoring of pH until a
SO
final pH of 3.3 was achieved: this required 133 kg of H
2
4
solution. The resulting slurry was then warmed to 25 °C and
maintained there for 1 h. The batch was then centrifuged and
washed twice with 40 kg of H O. The solid was then dried at
2
full vacuum at 40 °C until LOD <1% to give 36.0 kg of
carboxylic acid 16 (96%) as an off white solid; mp 226-228
1
EtOAc/Hex) provided 20.5 g of pure 10 (83%) as a white solid;
°C. H NMR (500 MHZ, DMSO-d
6
) δ: 12.47 (s, 1H), 8.32 (d,
1
mp 122-123 °C. H NMR (400 MHz, CDCl
3
) δ: 8.38 (d, J )
J ) 2.0 Hz, 1H), 8.20 (dd, J ) 4.6, 1.5 Hz, 1H), 7.96 (d, J )
2.3 Hz, 1H), 7.81 (dd, J ) 8.0, 1.6 Hz, 1H), 7.24 (dd, J ) 4.6,
8.0 Hz, 1H), 4.08 (q, J ) 7.0 Hz, 2H), 3.63 (s, 3H), 3.43 (s,
2.5 Hz, 1H), 8.19 (dd, J ) 1.5, 4.6 Hz, 1H), 8.16 (d, J ) 2.4
Hz, 1H), 7.47 (dd, J ) 1.5, 8.0 Hz, 1H), 7.09 (dd, J ) 4.7, 8.0
Hz, 1H), 4.56 (s, 1H), 4.20 (m, 6H), 3.49 (s, 3H), 1.25 (t, J )
13
3H), 1.17 (t, J ) 7.0 Hz, 3H). C NMR (125 MHz, DMSO-
) δ: 172.22 (s), 166.26 (s), 157.47 (s), 153.96 (s), 151.32 (d),
13
7
.1 Hz, 9H). C NMR (100 MHz, CDCl
3
) δ: 167.30 (s), 167.23
d
6
(s), 159.31 (s), 154.50 (s), 151.18 (d), 144.34 (d), 141.93 (d),
144.06 (d), 141.59 (d), 131.72 (d), 131.14 (s), 126.01 (s), 120.28
131.63 (s), 130.74 (d), 123.52 (s), 120.57 (s), 119.78 (d), 62.17
(d), 120.12 (s), 40.45 (t), 36.74 (q), 36.34 (t), 13.42 (q). Anal.
(
t), 54.80 (d), 41.24 (t), 37.30 (q), 13.96 (q), 13.58 (q). Anal.
Calcd for C21 : C, 61.15; H, 5.87; N, 13.58. Found: C,
1.14; H, 5.66; N, 13.51.
5-Ethyl-10-methyl-11-oxo-10,11-dihydro-5H-4,5,6,10-tet-
raazadibenzo[a,d]cyclohepten-2-yl)acetic Acid (16. See ref
f). An inerted 800 L glass-lined reactor fitted with an H vent
was charged with 12.0 kg of 60% NaH/oil (300 mol, 2.5 equiv),
.27 kg of Pd(OAc) (1.2 mol, 0.01 equiv), 1.26 kg of PPh
4.8 mol, 0.04 equiv), and 40 kg of bromide 3 (120 mol, 1
16 4 3
Calcd for C16H N O : C, 61.53; H, 5.16; N, 17.94. Found: C,
61.28; H, 4.94; N, 17.56.
H
24
N
4
O
5
6
5-Ethyl-10-methyl-11-oxo-10,11-dihydro-5H-4,5,6,10-tet-
(
raazadibenzo[a,d]cycloheptene-2-carboxylic acid (17): mp
1
251-253 °C. H NMR (400 MHz, CDCl
3
) δ: 8.84 (d, J ) 2.3
6
2
Hz, 1H), 8.40 (d, J ) 2.3 Hz, 1H), 8.19 (dd, J ) 4.6, 1.6 Hz,
1H), 7.82 (dd, J ) 8.0,1.6 Hz, 1H), 7.27 (dd, J ) 8.0, 4.6 Hz,
1H), 4.11 (q, J ) 7.0 Hz, 2H), 3.39 (s, 3H), 1.15 (t, J ) 7.0
0
(
2
3
13
3
Hz, 3H). C NMR (100 MHz, CDCl ) δ: 166.18 (s), 165.80
equiv). The reactor was again inerted, 70 kg of PhMe was
charged, and the mixture was heated to 60 °C. To this mixture
was then added a solution of 17.4 L of cyanoisopropylacetate
(s), 161.84 (s), 152.89 (s), 152.19 (d), 144.81 (d), 142.49 (d),
132.54 (d), 131.60 (s), 121.87 (s), 121.43 (d), 119.94 (s), 41.51
(t), 37.16 (q), 13.84 (q). Anal. Calcd for C15H N O : C, 60.40;
14 4 3
(
138 mol, 1.15 equiv) in 20 L of PhMe via metering pump at
H, 4.73; N, 18.78. Found C, 60.27; H, 4.64; N, 18.50.
5-Ethyl-2-(2-hydroxyethyl)-10-methyl-5,10-dihydro-
4,5,6,10-tetraazadibenzo[a,d]cyclohepten-11-one (2) via Ma-
lonate Surrogate Protocol. An 800 L reactor (reactor 1)
a fixed rate over 2 h (CAUTION: H gas evolution). The reactor
2
contents were then heated to 100 °C and maintained for 1.5 h.
HPLC of an aliquot showed the arylation was complete. The
mixture was then cooled to 35 °C and quenched by the addition
of 19 kg of i-PrOH (308 mol, 2.6 equiv) over 20 min
4
was inerted, then charged with 4.70 kg of NaBH (124
mol, 1.5 equiv) and reinerted, and 90 kg of diglyme was
added. The mixture was agitated at 25 °C for 3 h. In a
separate inerted 800 L reactor (reactor 2) was charged 27
kg of THF, and the mixture was cooled to 0 °C. Then
4.60 kg of HCl(g) was introduced subsurface over 1 h.
Titration of an aliquot showed 15.3% HCl by weight. In
(
CAUTION: H
to a stainless steel reactor, and 438 kg of 1 M NaOH solution
417 mol, 3.5 equiv) was then added. The mixture was heated
2
gas evolution). The batch was then transferred
(
to 80 °C and maintained there for 8 h. HPLC of an aliquot
showed the hydrolysis was complete to acid 16. The mixture
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
611

a third inerted 800 L reactor (reactor 3) equipped with a
hydrogen sensor were charged 26.0 kg of acid 16 (83.2
mol, 1 equiv) and 90 kg of THF, and the resultant slurry
was then cooled to 0 °C. To this slurry was then added
the contents of reactor 2 (HCl/THF) under N pressure
2
over 30 min, causing the formation of a thicker slurry.
This mixture was stirred for 30 min, then the contents of
added to reactor 2 over 2 h, through the hard pipe riser,
maintaining the reactor 2 contents temperature below -15 °C.
Reactor 1 was then rinsed with 154 kg of DME, and this wash
was transferred to reactor 2. The batch contents were then
warmed linearly at 0.07 °C/min to 22 °C, requiring 8 h. Reactor
1 was then cooled to -5 °C, and distillation of DME from
reactor 2 to reactor 1 was carried out at 1-25 °C/1-7 kPa,
collecting 1880 L of DME, requiring 14 h. To reactor 2 was
then charged 1564 kg of i-PrOH; then azeotropic distillation
of i-PrOH was carried out at 15-23 °C/2-13 kPa, collecting
2000 L distillate, requiring 9 h. Then charged to reactor 2 was
737 kg of i-PrOH, and to reactor 1 (empty) was charged 463
kg of i-PrOH. Reactor 1 contents were then cooled to -10 °C,
and HCl(g) was charged subsurface to reactor 1, using 10.7 kg
of HCl (293.5 mol, 0.97 equiv), requiring 1 h. The batch in
reactor 2 was then warmed to 74 °C, and the HCl/i-PrOH
solution in reactor 1 was then transferred to reactor 2 through
the bottom valve over 30 min. The batch contents (reactor 2)
were then cooled linearly to 60 °C at 0.2 °C/min, and held at
reactor 1 (NaBH
rate over 2 h (CAUTION: H
temperature of this reactor was then ramped to 25 °C over
h and held at that temperature for 2.5 h. HPLC of an
4
/diglyme) were added to it at a fixed
2
gas evolution). The contents
1
aliquot showed the reduction was complete. The reactor
contents were then cooled to 0 °C, and 206 kg of MeOH
was added at a fixed rate over 30 min and then agitated
vigorously for 30 min. The temperature of the contents
was slowly raised to 50 °C, and THF and MeOH were
removed via distillation under vacuum (60 mm). The
distillate was dropped from the receiver and discarded,
and then vacuum was reintroduced (14 mm); the contents
were then heated to a maximum of 69 °C as diglyme was
distilled off. The contents were then cooled to 25 °C, and
60 °C for 2 h. The batch was then cooled linearly to 20 °C at
0.1 °C/min. requiring 7 h. The resultant slurry was then aged
12 h at 20 °C and then centrifuged at 740 rpm, performing
4
2
37 kg of H O was charged. Azeotropic distillation of
three 82 kg of i-PrOH washes of the cake. The cake was then
cut out, placed in a dryer, and dried at 38 °C under full vacuum
this mixture at full vacuum and ∼50 °C was then carried
out until minimum stirrable volume was achieved. The
contents were again cooled to 25 °C; then 39 kg of 0.5 M
NaOH and 57 kg of heptane were charged to the reactor.
The resultant two-phase slurry was agitated for 10 h and
with a nitrogen purge to constant weight: 111.69 kg of 19,
1
8
0.2% yield, as a white solid; mp 206.2-207.9 °C. H NMR
(
400 MHz, MeOH-d
4
) δ: 8.95 (d, J ) 6.7 Hz, 1H), 8.48 (d, J
)
2.4 Hz, 1H), 8.40 (d, J ) 8.5 Hz, 1H), 8.15 (m, 2H), 8.07
then centrifuged, washing the solids with 40 kg of H
The material was packed out to a dryer and dried at 35
C and 10-50 mBar vacuum for 6 h to give 20.5 kg of
2
O.
(d, J ) 3.5 Hz, 2H), 7.84 (m, 1H), 7.75 (dd, J ) 1.5, 8.0 Hz,
1H), 7.47 (d, J ) 6.7 Hz, 1H), 7.19 (dd, J ) 4.7, 8.0 Hz, 1H),
°
4
)
.76 (t, J ) 5.6 Hz, 2H), 4.10 (m, 2H), 3.49 (s, 3H), 3.34 (t, J
6.1 Hz, 2H), 1.18 (t, J ) 7.1 Hz, 3H). C NMR (100 MHz,
alcohol 2 (80%) as an off-white solid. KF water analysis
13
showed 0.55%, and HPLC purity 99.1%; mp 139.0-140.5
1
MeOH-d ) δ: 170.23 (s), 169.21 (s), 160.16 (s), 156.04 (s),
4
°
C. H NMR (400 MHz, CDCl
3
) δ: 8.21 (d, J ) 2.4 Hz,
1
(
1
(
52.79 (d), 147.58 (d), 145.86 (d), 142.84 (d), 140.46 (s), 136.13
d), 133.26 (s), 133.14 (d), 130.27 (d), 130.00 (s), 124.67 (d),
22.28 (s), 122.19 (s), 121.72 (d), 121.35 (d), 103.80 (d), 72.91
t), 42.18 (t), 37.77 (q), 32.18 (t), 14.00 (q). Anal. Calcd for
: C, 65.00; H, 5.24; N, 15.16. Found: C, 64.86;
H, 5.09; N, 14.87.
-Ethyl-10-methyl-2-[2-(1-oxyquinolin-4-yloxy)ethyl]-5,10-
1
1
4
H), 8.15 (dd, J ) 4.6, 1.6 Hz, 1H), 7.91 (d, J ) 2.4 Hz,
H), 7.43 (dd, J ) 8.0, 1.6 Hz, 1H), 7.05 (dd, J ) 8.0,
.6 Hz, 1H), 4.12 (b, 2H), 3.75 (b, 2H), 3.46 (s, 3H), 2.76
(
3
t, J ) 6.5 Hz, 2H), 2.63 (br s, 1H), 1.21 (t, J ) 7.0 Hz,
H). C NMR (100 MHz, CDCl ) δ: 167.97 (s), 158.40
3
1
3
25 5 2
C H24ClN O
(
(
(
s), 155.17 (s), 151.48 (d), 144.52 (d), 141.25 (d), 131.78
s), 130.87 (d), 129.47 (s), 120.78 (s), 119.76 (d), 62.90
t), 41.23 (t), 37.55 (q), 35.32 (t), 13.73 (q). Anal. Calcd
5
dihydro-4,5,6,10-tetraazadibenzo[a,d]cyclohepten-11-one (1).
for C16
6
H
18
N
4
O
2
: C, 64.41; H, 6.08; N, 18.78. Found C,
4.26; H, 6.15; N, 18.58.
-[2-(5-Ethyl-10-methyl-11-oxo-10,11-dihydro-5H-4,5,6,10-
4
tetraazadibenzo[a,d]cyclohepten-2-yl)ethoxy]quinolinium
Chloride (19). An inerted 2000 L reactor (reactor 1) was
charged with 90 kg of alcohol 2 (302 mol, 1 equiv). A separate,
inerted, 3000 L reactor (reactor 2) was charged with 118.7 kg
All operations were performed in the dark. An inerted 400
L reactor was charged with 9.24 kg of 19 (19.8 mol, 1
of PPh
3
(452.6 mol, 1.5 equiv), and 47.7 kg of 4-hydroxyquino-
equiv) and then reinerted by two vacuum/N cycles. MeOH
2
line (329 mol, 1.09 equiv). Reactor 1 was reinerted; 1202 kg
of DME was charged, and agitation was started with heating
to 50 °C. After 2 h at 50 °C, the mixture was cooled to 25 °C
and held there. Reactor 2 was then reinerted, and 475.3 kg of
DME was charged, and the contents agitated at 25 °C for 20
min. To this mixture was then added 85.4 kg of DIAD (422.3
mol, 1.4 equiv) via metering pump over 90 min. The resulting
mixture was held at 30 °C for 40 min and then cooled to -22
(3.4 kg) and CH Cl (42.3 kg) were then charged, and
2
2
the internal temperature of the agitated slurry was adjusted
to 25 °C. Over 45 min was charged 133 kg of 2.7 M
Na CO , giving a batch pH of 10.0. To this mixture was
2
3
then added an aqueous slurry of 26.5 kg of 60% m-CPBA
(92 mol, 4.6 equiv) + 22.3 kg of H O over 45 min,
2
maintaining the batch temperature not more than 30 °C,
and using 3 kg of H O to rinse the charging lines. H
2
2
O
°C. The contents of reactor 1 (containing alcohol 2) were then
(13.1 kg) and CH Cl (7.7 kg) were then charged, and
2
2
6
12
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

the mixture was agitated at 120 rpm for 90 min, when
HPLC indicated the reaction was complete. Over 30 min,
7.11 (t, J ) 7.8 Hz, H27), 5.85 (s, H22), 4.30 (br s, H20), 4.08
(br s, H17), 3.27 (s, H16), 3.16 (t, J ) 6.6 Hz, H19), 1.21 (t, J
13
was charged 64.4 kg of 2 M Na
2
SO
3
, maintaining the batch
) 7.0 Hz, H18). C NMR (100 MHz, DMSO-d ) δ: 166.25
6
temperature not more than 30 °C. The mixture was then
(C15), 163.07 (C23), 162.95 (C21), 157.64 (C15), 153.94 (C12),
aged 50 min with agitation, and volatiles were removed
by distillation (38 °C/98 kPa), collecting 42 L. H O (153
2
151.09 (C9), 144.00 (C2), 141.16 (C7), 138.56 (C30), 131.67
(C4), 131.11 (C13), 130.83 (C26), 128.94 (C8), 122.12 (C28),
121.21 (C27), 120.25 (C14), 120.22 (C3), 115.08 (C25), 114.39
kg) was then charged to the reactor, and the batch was
agitated for 8 h at 40 °C and cooled to 30 °C; the solids
were isolated by centrifugation, at high speed, until mother
liquors ceased to flow (∼3 h). The cake was then washed
(
C29), 97.24 (C22), 68.25 (C20), 40.37 (C17), 36.71 (C16),
0.49 (C19), 13.39 (C18). Anal. Calcd for C25 ·[0.5
O]: C, 66.65; H, 5.37; N, 15.55. Found C, 66.47; H, 5.02;
N, 15.33.
-Aminobenzoic acid 2-(5-ethyl-10-methyl-11-oxo-10,11-
3
23 5 3
H N O
H
2
2
in the centrifuge twice with 30 kg of H O and dried to
constant weight: 9.85 kg of 1 as an off-white solid, 10.7%
O by KF (weakly crystalline trihydrate by XRPD.
CH Cl < 400 ppm. For a discussion of the crystalline
forms of compound 1, see ref 18). HPLC purity 98.1A%,
6% yield. CAUTION: This solid is a dust explosion
2
H
2
dihydro-5H-4,5,6,10-tetraazadibenzo[a,d]cyclohepten-2-yl)-
ethyl ester (22).
2
2
8
hazard! DSC showed an endotherm at ∼94 °C, followed
1
by a melt at ∼136 °C. H NMR (400 MHz, CDCl
3
) δ:
8
.69 (d, J ) 7.0 Hz, 1H,), 8.62 (d, J ) 8.7 Hz, 1H), 8.39
d, J ) 2.4 Hz, 1H), 8.21 (d, J ) 8.4 Hz, 1H), 8.18 (dd,
J ) 4.7, 1.6 Hz, 1H), 7.8 (dd, J ) 5.4, 5.4 Hz, 1H), 7.70
dd, J ) 7.2, 7.2 Hz, 1H), 7.47 (dd, J ) 8.0, 1.6 Hz, 1H),
(
A flask was charged with 596 mg of alcohol 2 (2.10 mmol, 1
equiv), 1.22 g of DMAP (10.0 mmol, 5 equiv), 1.63 g of isatoic
anhydride (10.0 mmol, 5 equiv) and 15 mL DMF. The mixture
(
7
4
3
7
.07 (dd, J ) 7.9, 4.7 Hz, 1H), 6.78 (d, J ) 7.0 Hz, 1H),
.42 (dd, J ) 6.2, 6.2 Hz, 2H), 4.18 (q, J ) 6.9 Hz, 2H),
.50 (s, 3H), 3.22 (dd, J ) 6.2, 6.2 Hz, 2H), 1.24 (t, J )
was heated to 60 °C under N and maintained there for 2 h.
2
The cooled mixture was then diluted with 50 mL of EtOAc
1
3
.0 Hz, 3H). C NMR (125 MHz, DMSO-d
6
) δ: 166.25
and 50 mL of H O, and the layers were separated. The aqueous
2
(
(
(
(
(
C6), 157.68 (C15), 153.95 (C12), 151.31 (C21), 151.13
C9), 144.02 (C2), 141.21 (C7), 140.47 (C30), 135.36
C23), 131.70 (C4), 131.11 (C13), 130.58 (C26), 128.86
C8), 128.02 (C27), 122.28 (C28), 122.07 (C29), 120.24
C3 + C14), 119.24 (C25), 101.92 (C22), 68.87 (C20),
phase was extracted with EtOAc (2 × 50 mL), and the
combined organic phases were washed with saturated NaCl (2
×
2 4
100 mL) and dried (Na SO ), and the solvents were removed
in vacuo to give a solid. The solid was dissolved in EtOAc and
filtered through a pad of silica gel to remove polar material,
and then the filtrate was chromatographed on silica gel eluting
with 1:1 Hex/EtOAc to give 600 mg of anilinoester 22 (72%)
4
0.36 (C17), 36.71 (C16), 30.59 (C19), 13.40 (C18). Anal.
Calcd for C25 · [10.7% H O]: C, 60.60; H, 5.90;
N, 14.13. Found C, 60.46; H, 5.45; N, 14.22.
-Ethyl-10-methyl-2-[2-(2-oxo-1,2-dihydroquinolin-4-yloxy)-
H
23
N
5
O
3
2
23 5 3
as a white solid; mp 107 °C. HRMS Calcd for C23H N O :
5
1
4
18.1879; Found: 418.1878 (error 0.2 ppm); H NMR (500
MHz, DMSO-d ) δ: 8.38 (d, J ) 2.2 Hz, H9), 8.19 (dd, J )
.6, 4.6 Hz, H2), 8.01 (d, J ) 2.3 Hz, H7), 7.61 (dd, J ) 1.6,
.1 Hz, H27), 7.25 (m, H3 + H25), 6.74 (dd, J ) 0.8, 8.3 Hz,
), 6.47 (t, J ) 7.1 Hz, H26), 4.36 (t, J )
.6 Hz, H20), 4.06 (m, H17), 3.42 (s, H16), 3.01 (t, J ) 6.6
ethyl]-5,10-dihydro-4,5,6,10-tetraazadibenzo[a,d]cyclo-
hepten-11-one (21).
6
1
8
H24), 6.58 (br s, NH
6
2
13
Hz, H19), 1.16 (t, J ) 7.1 Hz, H18). C NMR (125 MHz,
DMSO-d ) δ: 167.11 (C21), 166.23 (C6), 157.64 (C15), 153.92
C12), 151.35 (C23), 150.94 (C9), 144.01 (C2), 140.99 (C7),
6
(
A solution of 0.48 g (1.00 mmol, 1 equiv) of 1 in 0.7 L of
MeOH was photolyzed with a 200 W floodlamp for 7 days in
a glass flask. The solvent was then evaporated and the residual
133.98 (C25), 131.69 (C4), 131.12 (C13), 130.41 (C27), 128.98
(C8), 120.28 (C14), 120.23 (C3), 116.46 (C24), 114.67 (C26),
108.54 (C22), 63.82 (C20), 40.38 (C17), 36.73 (C16), 30.41
solid dissolved in CH
silica gel. This material was then chromatographed on silica
gel using 5% MeOH/CH Cl to give 240 mg (50%) lactam 21
as a white solid; mp 281 °C. H NMR (400 MHz, DMSO-d
δ: 8.45 (d, J ) 1.8 Hz, 1H, H9), 8.18 (d, J ) 4.0 Hz, H2), 8.08
d, J ) 1.8 Hz, H7), 7.82 (d, J ) 7.8 Hz, H4), 7.71 (d, J ) 7.8
Hz, H28), 7.47 (t, J ) 7.8 Hz, H26), 7.25 (m, H3 + H25),
2 2
Cl /MeOH and adsorbed onto a plug of
(C19), 13.40 (C18). Anal. Calcd for C23
H N O
23 5 3
2
·[0.27 H O]:
C, 65.41; H, 5.62; N, 16.58. Found: C, 65.79; H, 5.52; N, 16.18.
2
2
1
6
)
Received for review March 29, 2008.
OP8000756
(
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
613