A scaleable synthesis of fiduxosin

Article abstract of DOI:10.1021/op049889k

Fiduxosin (1) has been under development at Abbott Laboratories for the treatment of benign prostatic hyperplasia. A convergent strategy required methodologies for preparation of an enantiomerically pure 3,4-cis-disubstituted pyrrolidine and a 2,3,5-trisubstituted thienopyrazine in a regiospecific manner. A [3+2] cycloaddition of an enantiopure azomethine ylide followed by a diastereoselective crystallization was employed to prepare the benzopyranopyrrolidine in high diastereomeric and enantiomeric purity. Conditions for reduction of an O-aryl lactone susceptible to epimerization were developed, and cyclization of the alcohol/phenol to the ether was accomplished in high yield. The thienopyrazine was prepared by condensation of methyl thioglycolate and a regiospecifically prepared 2-bromo-3-cyano-5-phenylpyrazine. Conditions for effective halogen substitutive deamination to prepare regiospecific trisubstituted pyrazines will be described.

Full text of DOI:10.1021/op049889k

Organic Process Research & Development 2004, 8, 897−902
A Scaleable Synthesis of Fiduxosin
Anthony R. Haight,* Anne E. Bailey,† William S. Baker, Michael H. Cain, Richard R. Copp,‡ John A. DeMattei,§
Kelley L. Ford, Rodger F. Henry, Margaret C. Hsu, Robert F. Keyes, Steven A. King, Maureen A. McLaughlin,
Laura M. Melcher, William R. Nadler,| Patricia A. Oliver, Shyamal I. Parekh, Hemant H. Patel,# Louis S. Seif,
Mike A. Staeger, Gregory S. Wayne, Steven J. Wittenberger, and Weijiang Zhang3
GPRD Process Research and DeVelopment, Abbott Laboratories, Building R8/1, 1401 Sheridan Road,
North Chicago, Illinois 60064-6285, U.S.A.
Abstract:
Fiduxosin (1) has been under development at Abbott Labora-
tories for the treatment of benign prostatic hyperplasia. A
convergent strategy required methodologies for preparation of
an enantiomerically pure 3,4-cis-disubstituted pyrrolidine and
a 2,3,5-trisubstituted thienopyrazine in a regiospecific manner.
A [3+2] cycloaddition of an enantiopure azomethine ylide
followed by a diastereoselective crystallization was employed
to prepare the benzopyranopyrrolidine in high diastereomeric
and enantiomeric purity. Conditions for reduction of an O-aryl
lactone susceptible to epimerization were developed, and cy-
clization of the alcohol/phenol to the ether was accomplished
in high yield. The thienopyrazine was prepared by condensation
of methyl thioglycolate and a regiospecifically prepared 2-bromo-
3-cyano-5-phenylpyrazine. Conditions for effective halogen
substitutive deamination to prepare regiospecific trisubstituted
Figure 1. BPH antagonists.
pyrazines will be described.
blocking of the R1-adrenoreceptor sites can successfully treat
the physiological effects of BPH.3
The first R1-antagonists approved for the treatment of BPH
(terazosin,3 prazosin,4 and doxazosin5) had previously been
Introduction
introduced as therapies for hypertension. It is not surprising
then that cardiovascular side effects limited the dosing of
these agents for the treatment of BPH. Binding and functional
assays have demonstrated that the R1-adrenoreceptors are
composed of at least three native subtypes: R1a-, R1b-, and
R1d-adrenoreceptor sites with the R1a-site attributed to the
uroselectivity.6 Tamsulosin (Flomax), 2,7 was the first
compound designed with improved R1a/R1b-selectivity for the
treatment of BPH (Figure 1).
At Abbott, the R1a-antagonist fiduxosin (1) was found to
have improved uroselectivity relative to tamsulosin.8 To
further advance development of fiduxosin, a scaleable
synthetic route was necessary. We describe here our efforts
Benign prostatic hyperplasia (BPH), or the nonmalignant
enlargement of the prostate gland, is a disorder associated
with up to 80% of all males over the age of 60.1 Typical
symptoms of BPH include nocturia, increased frequency of
urination, and low urine flow caused, in part, by the
enlargement of the prostate gland and contraction of the
smooth muscle tissue surrounding the bladder neck. The
finding by Caine2 that the smooth muscle of the prostate
and bladder neck contains large numbers of R-adrenoreceptor
sites led to the experimental observation that the nonselective
R2/R1-adrenoreceptor antagonist, phentolamine, could ef-
fectively address the symptoms of BPH. Clinical evidence
with the selective R1-adrenoreceptor antagonist prazosin and
further functional studies have led to the recognition that
(3) Lepor, H.; Henry, D.; Laddu, A. R. Prostate 1991, 18, 345.
(4) Hedlund, H.; Andresson, K.-E.; Ek, A. J. Urol. 1983, 130, 275.
(5) Holme, J. B.; Christensen, M. M.; Rasmussen, P. C.; Jacobsen, F.; Nielsen,
J.; Norgaard, J. P.; Olesen, S.; Noer, L.; Wolf, H.; Husted, S. E. Scand. J.
Urol. 1994, 28, 77.
(6) Hieble, J. P.; Ruffolo, R. R. Exp. Opin. InVest. Drugs 1997, 6, 367. Meyer,
M. D.; Altenbach, R. J.; Bai, H.; Basha, F. Z.; Carroll, W. A.; Kerwin, J.
F.; Lebold, S. A.; Lee, E.; Pratt, J. K.; Sippy, K. B.; Tietje, K.; Wendt, M.
D.; Brune, M. E.; Buckner, S. A.; Hancock, A. A.; Drizin, I. J. Med. Chem.
2001, 44, 1971 and references therein.
* To whom correspondence should be addressed. E-mail: anthony.haight@
abbott.com.
† Present address: Hospira Inc., 275 N. Field Dr., Lake Forest, IL 60045.
‡ Present address: Quintessence Biosciences, 505 S. Rosa Rd., Madison, WI
53719.
§ Present address: Array Biopharma 1885 33rd St., Boulder, CO 80301.
| Present address: Industria Farmaceutica Serono, Via Luigi Einaudi, 11,
00012 Guidonia-Montecelio, Italy.
Present address: Vertex Pharmaceuticals, Boston, MA.
# Present address: Gujarat Industrial Chemicals, 388 620 Gujarat, India.
3 Present address: Theravance Inc., 901 Gateway Blvd. South, San Francisco,
CA 94080.
(7) Yamada, S.; Suzuki, M.; Tanaka, C.; Mori, R.; Kimura, R.; Inagaki, O.;
Honda, K.; Asano, M.; Takenaka, T.; Kawabe, K. Clin. Exp. Pharmacol.
Physiol. 1994, 21, 405.
(1) Bock, M. G.; Patane, M. A. Ann. Rep. Med. Chem. 2000, 35, 221. Garraway,
W. M.; Collins, G. N.; Lee, R. J. Lancet 1991, 338, 469.
(2) Caine, M.; Pfau, A.; Perlberg, S. Brit. J. Urol. 1976, 48, 255.
(8) Hancock, A. A.; Buckner, S. A.; Brune, M. E.; Esbenshade, T. A.; Ireland,
L. M.; Katwala, S.; Milicic, I.; Meyer, M. D.; Kerwin, J. F., Jr.; Williams,
M. J. Pharmacol. Exp. Ther. 2002, 300, 478.
10.1021/op049889k CCC: $27.50 © 2004 American Chemical Society
Published on Web 10/16/2004
Vol. 8, No. 6, 2004 / Organic Process Research & Development
•
897

Scheme 1. 5-Methoxycoumarin synthesisa
Figure 2. Retrosynthesis of fiduxosin.
a Reaction conditions: (a) (1) BuLi, 0 °C, THF/TMEDA; (2) DMF; (3) 4 M
HCl (68%). (b) AlCl3, CH2Cl2 (93%). (c) (1) Ac2O, DMF, K2CO3, 70 °C; (2)
H2O (10 mol%), 120 °C (68%).
of 9. Cannizarro reaction byproducts were prevalent if the
mixture was not first acidified.11,12 Following the addition
of heptane to the slurry to remove residual 8, aldehyde 9
was isolated in 68% yield and 99.3% GC purity as an off-
white solid. The mass balance was accounted for in the
filtrate, which contained unreacted 8 (6%), aldehyde 9 (5%),
and the Cannizarro acid 12 (8%) and alcohol 13 (8%).
Selective dealkylation was carried out by addition of 9
to a -15 °C methylene chloride slurry of 1.5 equiv of
AlCl3.13 Slowly warming the reaction mixture to ambient
temperature gave mono-demethylation in 4-6 h with <5%
bis-demethylation as detected by GC. If the reaction was
allowed to warm further to 30 °C, significant bis-dealkylation
to 11 would occur. Fortunately, this undesired impurity sub-
limed during the drying of the product (45 °C under vacuum)
giving aldehyde 10 in 93% yield and >99% GC purity.
Preparation of coumarin 6 from aldehyde 10 was carried
out by the Perkin reaction.14 Employing Cs2CO3 and DMF
as the base and solvent, respectively, acetic anhydride
condensed with 10 to yield coumarin 6 in 86% isolated yield.
Although the reaction proceeded in high yield and good
purity, the expense of cesium carbonate caused us to evaluate
alternative bases.15 Among these, potassium carbonate was
found to be the best, but the procedure required some
modifications. The reaction was carried out in a stepwise
fashion by first treating 10 with 5 equiv of Ac2O and 1.1
equiv of K2CO3 in DMF at 70 °C to generate the intermediate
acetate 14. Addition of a catalytic amount of water (10 mol
%) then generated a more soluble mixed base system of
potassium acetate and carbonate. After increasing the reaction
temperature to 120 °C for 10 to 12 h, the conversion was
complete and the product was precipitated by slow addition
of water. Coumarin 6 was obtained in 68% isolated yield.
The coumarin was isolated with up to 25% weight inorganic
salts. Since no adverse consequences were found in the
Figure 3. [3+2] Cycloaddition strategy.
to prepare multi-kilogram quantities of 1 through a conver-
gent strategy.
Results
Fiduxosin can be assembled via the coupling of the
thienopyrazine isocyanate 3 with the benzopyranopyrrolidyl
butyryl diamine, 4. Consequently, we required a regiospecific
preparation of 3 and a synthesis that would control both the
relative and absolute stereochemistry in the 3,4-disubstituted
pyrrolidyl ring system of 5 (Figure 2).
A diastereoselective cycloaddition between 5-methoxy-
coumarin, 6, and a nonracemic azomethine ylide was en-
visaged as the key step in preparing the benzopyranopyrro-
lidine (Figure 3).9,10 This approach offers the advantage of
establishing both the relative and absolute stereochemistry
in one step. In the original synthesis of 1,9 the use of the
(R)-R-methylbenzylamine derived azomethine ylide 7 pro-
vided a facile means to key tricyclic amine intermediate 5
with high enantiomeric excess. The ease of isolation and the
robustness of the chemistry encouraged us to use this route
for the initial clinical deliveries.
Preparation of 5-methoxycoumarin 6 was accomplished
in three steps from 1,3-dimethoxybenzene. Regioselective
lithiation of 8 with n-BuLi at 0 °C in THF/TMEDA followed
by treating the aryllithium with DMF gave the desired
2,6-dimethoxybenzaldehyde, 9 (Scheme 1). Although TME-
DA was not necessary for the metalation, it did stabilize the
DMF adduct. Without TMEDA, the tetrahedral intermediate
would collapse to the aldehyde, which we had observed to
dealkylate under the reaction conditions. Addition of aqueous
HCl followed by distillation of the THF left an aqueous slurry
(9) Meyer, M. D.; Altenbach, R. J.; Basha, F. Z.; Carroll, W. A.; Drizin, I.;
Kerwin, J. F., Jr.; Wendt, M. D.; Haight, A. R.; Zhang, W. U.S. Patent
6,046,207, 2000. Lavielle, G.; Dubuffet, T.; Millan, M.; Newman-Tancredi,
A., U.S. Patent 5,663,191, 1995.
(11) Geissman, T. A. Org. Reactions 1944, 2, 94.
(12) Cannizarro type products were still seen with the pH < 2; however they
were greatly reduced: >80% conversion to 12 and 13 without HCl versus
<20% conversion after neutralization to pH <2.
(10) Padwa, A.; Chen, Y.-Y.; Chiacchio, U.; Dent, W. Tetrahedron 1985, 41,
3529. Negron, G.; Roussi, G.; Zhang, J. Heterocycles 1992, 34, 293. Cottrell,
I. F.; Hands, D.; Kennedy, D. J.; Paul, K. J.; Wright, S. H. B.; Hoogsteen,
K. J. Chem. Soc., Perkin Trans 1 1991, (5), 1091.
(13) Horie, T.; Tominaga, H.; Kawamura, Y.; Yamada, T. J. Org. Chem. 1992,
57, 3343.
(14) Johnson, J. R. Org. React. 1942, 1, 210.
(15) The cost of cesium carbonate is 20 times the cost of potassium carbonate.
898
•
Vol. 8, No. 6, 2004 / Organic Process Research & Development

Table 1. Cycloaddition optimization
the supernatant. The minor lactone 15b proved more soluble
than 15a in both solvent systems, where, at 0-5 °C, the ratio
of 15a to 15b was 1:5 and 1:6, respectively. Using the
concentration of 15b in these crystallization solvents (60
mg/mL in EtOAc; 47 mg/mL in EtOAc/MtBE 50/50), we
were able to determine the minimum amount of solvent
required to remove isomer 15b in a single crystallization.
The optimized procedure entailed addition of a 2-fold
molar excess of the ylide precursor 16 to a cooled THF
solution containing 6 and catalytic TFA which resulted in
the formation of a 1.5:1 ratio of the lactones 15a/15b with
high conversion (Scheme 2). The isolation was carried out
by concentrating the reaction mixture and solvent switching
to EtOAc for crystallization to give 15a in >99:1 diaster-
eomeric purity and in 40-43% isolated yield from 5-meth-
oxycoumarin. The mother liquors had 10 mg/mL of 15a
(15-20 wt % of product) and 60 mg/mL of the undesired
15b isomer.
temp
(°C)
dr
entrya solvent
additive
conversionb (15a/15b)
1
2
3
4
THF
TFA (10%)
8 to 10
8 to 10
8 to 10
8 to 10
>90%
>90%
>90%
>90%
1.7/1
1.8/1
1.6/1
1.7/1
toluene TFA (10%)
EtOAc TFA (10%)
MtBE/ TFA (10%)
THF
5
THF
BF3-Et2O
8 to 10
8 to 10
<50%
1.7/1
(10%)
6
7
8
9
MtBE AlCl3 (10%)
10%
<50%
36%c
2.1/1
1.9/1
1.5/1
1.7/1
THF
THF
THF
TBAF (10%) -1 to 23
TFA (10%) -10 to -9
TFA (10%)
55
<25%
The carbonyl group was removed by a reduction/cycliza-
tion sequence. Addition of a THF solution of lithium
borohydride to lactone 15a resulted in formation of diols
17c and 17t in a 9:1 ratio. The epimerization results from
enolization of the intermediate aldehyde (21a/21b) generated
during the reduction (Scheme 3).18 Epimer 17t could be
minimized by reversing the order of addition to maintain an
excess of the reducing agent. On multi-kilogram scale,
controlled inverse addition was successful in minimizing the
amount of 17t observed (55:1 for 17c/17t).
Further improvement on the selectivity was realized after
the preparation of the phenol/alcohol on a large scale. The
alternative reduction methodology involved slowly adding
methanol to a mixture of lactone 15a and excess NaBH4 in
toluene at 0 °C. As the methanol is added, the reduction
initiates to cleanly give 17c with less than 0.2% of the epimer
17t.19 Quenching of the reduction with methanol followed
by refluxing and distillation effectively hydrolyzed the
boronate ester and removed residual boron as trimethyl
borate19,20 to give the desired diol in >95% yield and >98%
purity. Although this methodology is superior to the LiBH4
reduction protocol, it was not employed in the initial scale-
up as the conditions were not optimized prior to the first
delivery of fiduxosin.
To complete the cyclization, crude diol prepared from the
LiBH4 reduction sequence was treated with excess MsCl and
TEA. Reaction with 1 equiv of MsCl gave a mixture of 17c,
18a, and 18b (5:92:2 respectively). We were unable to
selectively react the primary alcohol in the presence of the
phenol. To consume the diol completely required 1.3 equiv
of MsCl which resulted in a mixture of mono- and bis-
mesylates, 18a and 18b in ∼7:1 ratio. On exposure to
potassium tert-butoxide, 18b undergoes aryl deprotection,21
and in one pot, the mesylate mixture was cleanly converted
to the tricyclic freebase of 19c in >98% assay yield.
a Reaction conditions: 16 (2 equiv) added over 30 to 60 min to the mixture
of 6, additive, and solvent. b HPLC ratio (15a + 15b)/(6 + 15a + 15b) at 220
nm. c Ylide (1.3 equiv) added.
subsequent steps due to the inorganics, the crude solid was
used as obtained.16
The original synthesis of 1 utilized a [3+2] cycloaddition
of the ylide derived from 16 and coumarin 6.9 Although there
was practically no diastereoselectivity observed, the cycload-
ducts 15a and 15b could easily be separated by crystalliza-
tion. The poor selectivity observed was not suprising,10
because the preferred exo-transition state of the cycloaddition
precludes efficient stereodifferentiating interactions between
the chiral auxiliary and the developing stereocenters.17
Nonetheless, we were attracted by the efficient separation
of the diastereomers 15a and 15b. Practically, this cycload-
dition can be thought of as a classical resolution. Any
enhancement in diastereoselectivity would result in a yield
improvement, so we first embarked on an investigation to
improve the reaction selectivity and yield.
Solvent, azomethine ylide activation methods, and tem-
perature were all evaluated for their influence on the
diastereoselectivity and efficiency of the cycloaddition.
Dilution with MtBE or replacement of the THF with toluene
or ethyl acetate did not greatly influence either the diaster-
eomeric ratio or the product assay (Table 1, entries 1-4).
Alternative methods of activating the ylide precursor 16
failed to give good conversion and, additionally, offered little
promise for improving the diastereoselectivity (entries 5-7).
Changing the temperature from -10 °C to 55 °C did affect
the amount of product observed with little influence on the
stereoselectivity (entries 8 and 9).
Given the poor diastereoselectivity, it was important to
optimize the isolation of the desired diastereomer, 15a.
EtOAc and 50/50 MtBE/EtOAc slurry mixtures of 15a and
15b were analyzed by HPLC for the diastereomeric ratio in
(18) Caille, J.-C.; Govindan, C. K.; Junga, H.; Lalonde, J.; Yao, Y. Org. Proc.
Res. DeV. 2002, 6, 471.
(19) Repic, O. Principles of Process Research and Chemical DeVelopment in
the Pharmaceutical Industry; Wiley: New York, 1998; pp 72, 89-93.
(20) Trimethyl borate forms an azeotrope with methanol: Cunningham, G. L.
U.S. Patent 3004058.
(16) Latter scale-up syntheses employed steam distillation to remove inorganics
from 6.
(17) Padwa, A.; Chen, Y. Y.; Chiacchio, U.; Dent, W. Tetrahedron 1985, 41,
3529.
(21) Dinsmore, C. J.; Zartman, C. B. Tetrahedron Lett. 1999, 40, 3989.
Vol. 8, No. 6, 2004 / Organic Process Research & Development
•
899

Scheme 2. Benzopyranopyrollidine synthesisa
a Reaction conditions: (a) TFA (10 mol%), THF/MtBE, 15 °C, 16 (2.0 equiv). (b) EtOAc x’tal (2 mL/g 6) (43%). (c) LiBH4, THF. (d) MeOH, reflux (95%). (e)
THF/toluene, 0 °C, TEA (2.4 equiv), MsCl (1.3 equiv), 1 h. (f) KOtBu (3 equiv), 10 °C, 1 h (98%). (g) HCl, iPrOH (76% from 6). (h) Pd/C, H2, MeOH (92%).
Scheme 3. Lactone reduction
Figure 4. X-ray structure of 19c.
5-phenylpyrazine. We made use of chemistry by E. C. Taylor
involving the condensation of aminomalononitrile 22 with
phenylglyoxal oxime 23 in 2-propanol to give 24 in 95%
The low melting waxy freebase ether of 19c proved
difficult to purify by crystallization. We turned to preparing
isolated yield by simply cooling the reaction to 0 °C and
the hydrochloride salt 19c by crystallization from 2-propanol,
collecting the solid (Scheme 4).24 2-Amino-3-cyano-5-
which provided the compound in greater than 98.0% purity
phenylpyrazine N-oxide 24 could be reduced by treatment
with the trans isomer 19t as the largest impurity (<1% by
with (EtO)3P at 100 °C. Calorimetry of the reaction showed
HPLC) in 76% yield from 6.
the presence of two large exotherms; the first initiating at
The relative and absolute stereochemistry was confirmed
115 °C and the second at 210 °C. To alleviate safety
by a single-crystal X-ray structure of 19c (Figure 4).
concerns, the reaction was carried out by slowly adding 24
Hydrogenolysis of 19c over Pd(OH)2/C at 50 °C in ethanol
to neat (EtO)3P at 100 °C. Upon completion, the reaction
was uneventful in yielding tricycle 20c in 92% isolated yield
and >99.5% purity (>500:1 20c/20t).
was quenched with water to hydrolyze unreacted triethyl
phosphite and precipitate 25 in 92% yield and 98% HPLC
Phenyl-Thieno[2,3-b]pyrazine. The most general method
purity. As an alternative to the slow addition of 24 to the
of synthesizing thieno[2,3-b]pyrazines has been to treat
3-cyano-2-chloropyrazines with thioglycolates.22 We were
acutely aware that one of the more troublesome aspects of
the original synthesis of 1 was the separation of the regio-
isomers of 28.23 With this in mind, we sought efficient
methods for the regiospecific preparation of 2-halo-3-cyano-
(EtO)3P, it was found that the reaction could be run in
refluxing ethanol. This was considered safer to scale-up as
the ethanol provided an effective heat sink and assured the
temperature would remain sufficiently below the onset of
the observed exothermic events. Residual (EtO)3P in the
ethanol was then quenched with water, and the product
isolated, as before, by filtration.
(22) Taylor, E. C.; Reiter, L. C. J. Org. Chem. 1982, 47, 528. Bourguignon, J.;
Lemarchand, M.; Quequiner, G. J. Heterocycl. Chem. 1980, 17, 257.
(23) Attempts to separate at regioisomeric mixtures of pyrazine intermediates
were not successful. See ref 9.
(24) Taylor, E. C.; Kobayashi, T. J. Org. Chem. 1973, 38, 2817.
900
•
Vol. 8, No. 6, 2004 / Organic Process Research & Development

Scheme 4. Thienopyrazine synthesisa
a Reaction conditions: (a) iPrOH, rt, 24 h (95%). (b) P(OEt)3, EtOH, 78 °C (92%). (c) See text (73%). (d) HSCH2CO2Me, Na2CO3, MeOH (92%). (e) Triphosgene,
THF, 68 °C (95%).
Scheme 5. Final assembly stepsa
reaction media at 50-60 °C, while lower temperatures
resulted in higher levels of competitive reaction byproducts
such as the hydroxypyrazine. To ameliorate this, immediately
upon completion, the reaction mixture was rapidly quenched
into a -20 °C acidic solution to precipitate the product. The
crude product was dissolved in CH2Cl2 and clarified through
a pad of Celite. Crystallization from heptane/CH2Cl2 gave
27 in 73% yield and >98% purity.
Coupling of 27 with methyl thioglycolate in methanol with
sodium carbonate as the base proceeded uneventfully. Crude
28 was purified by filtering through a pad of alumina or silica
with either methylene chloride or toluene to yield 28 as a
bright yellow solid in 92% yield and >99% purity. Conver-
sion of thienopyrazine 28 to isocyanate 29 was accomplished
by heating a THF solution of 28 in the presence of 3 equiv
of triphosgene followed by precipitation of 29 with heptane.
a Reaction conditions: (a) NaHCO3, CH2Cl2. (b) Br(CH2)3CN, iPr2NEt,
CH3CN (95%). (c) Ra-Ni, MeOH/NH3, H2 (93%). (d) (1) 29 (0.98 equiv),
toluene, rt, 2 h; (2) reflux 2 h. (e) Toluene/EtOH, HCl (70-85% from 29).
This gave 29 in 90-95% yield and >99.0% purity.
Final Assembly. Hydrochloride salt 20c was neutralized
in methylene chloride with aqueous NaHCO3. The solution
was solvent exchanged into acetonitrile, and the freebase was
N-alkylated with 4-bromobutyronitrile in the presence of
Hunig’s base. The reaction mixture was concentrated and
diluted with toluene. Aqueous extraction removed unreacted
freebase and a bis-alkylated ammonium salt to give nitrile
30 in >95% yield and >98% purity. The nitrile was reduced
with Raney nickel in a mixture of methanol and ammonia.27
The catalyst was filtered, and the solvent exchanged into
toluene. Treatment of the toluene solution with aqueous
NaOH precipitated residual nickel from the green-tinted
biphasic mixture as a flocculant white solid which could be
removed by filtration.28 This yielded a toluene solution of
the primary amine 31 in 93% assay yield with less than 5
ppm of residual nickel by ICP (Scheme 5).
In our attempts to convert the 2-amino substituent of 25
to a leaving group, Sandmeyer conditions (CuCl, aqHCl,
NaNO2)25 failed to give the desired chloride 26. Careful
analysis of in-process HPLC chromatograms showed a new
component (thought to be the diazonium intermediate)
forming rapidly and then slowly decomposing during the
course of the reaction to give low yields of 2-hydroxypyra-
zine as the only identified component. Preparation of the
diazonium salt under anhydrous conditions using t-BuONO
as the oxidant still resulted in poor conversion with cuprous
chloride. However, formation of the diazonium species in
the presence of cupric chloride (CuCl2, t-BuONO, CH3CN)
gave chloride 26 as the major component of the reaction
mixture.26 Byproducts such as the hydroxypyrazine and the
acetamide resulting from a Stetter reaction with the solvent
(acetonitrile) suggested chloride was not reactive enough.
Switching to DMF as the solvent and cupric bromide as the
halide source resulted in consumption of the diazonium salt
in less than 30 min at 50-60 °C. Unfortunately, HPLC assay
indicated that the bromopyrazine 27 was unstable in the
Amine 31 was reacted with isocyanate 29 in toluene to
generate fiduxosin via the transient urea 32. Temperature
control during the stages of the reaction was important. If
the intermediate urea was allowed to cyclize in the presence
(26) Doyle, M. P.; Siegfried, B.; Dellaria, J. F., Jr. J. Org. Chem. 1977, 42,
2426. Doyle, M. P.; Van Lente, M. A.; Mowat, R.; Fobare, W. F. J. Org.
Chem. 1980, 45, 2570. Taylor, E. C.; Reiter, L. A. J. Am. Chem. Soc. 1989,
111, 285.
(27) Freifelder, M.; Ng, Y. H. J. Pharm. Sci. 1965, 54, 1204.
(28) Svehla, G. Vogel’s QualitatiVe Inorganic Analysis; Svehla, G., Ed.; p 132.
(25) For examples of Sandmeyer reactions with similar compounds, see: Taylor,
E. C.; Reiter, L. A. J. Am. Chem. Soc. 1989, 111, 285. Favini, G.; Simonetta,
M. Gazz. Chim. Ital. 1959, 89, 2222.
Vol. 8, No. 6, 2004 / Organic Process Research & Development
•
901

of unreacted 29, the methanol produced would compete with
31 and generate the methyl carbamate of 29. To inhibit the
generation of this impurity, the reaction was kept at ambient
temperature until the isocyanate was consumed (2 h). The
mixture was then heated to reflux to drive the cyclization
reaction to completion. Precipitation with heptane gave
fiduxosin freebase as a free flowing solid. The free base was
dissolved in 50% ethanolic toluene solution and treated with
concentrated HCl (1.75 equivalents) to give fiduxosin HCl
in 70-85% isolated yield from 29.
azomethine ylide [3+2] cycloaddition strategy was developed
as a means of generating the 3,4-cis-disubstituted pyrrolidine
in high enantiomeric excess. We have also described
methodology to regiospecifically prepare the trisubstituted
bromopyrazine and its conversion to thienopyrazine.
Acknowledgment
We thank Alicia Kuckkan and James Morley for analytical
support.
The unusual solubility characteristics of the freebase
precluded formation of the hydrochloride salt directly from
either an ethanol or a toluene solution. While the freebase
showed only limited solubility in either ethanol or toluene
(<10 mg/mL in either), a 50/50 mixture of the solvents
improved the solubility immensely (>65 mg/mL at 65 °C).
In conclusion, a convergent synthesis of fiduxosin HCl
has been demonstrated. A robust and scaleable chiral
Supporting Information Available
Experimental procedures and characterization. This
material is available free of charge via the Internet at
http://pubs.acs.org.
Received for review June 2, 2004.
OP049889K
902
•
Vol. 8, No. 6, 2004 / Organic Process Research & Development