Development of a scalable and safe procedure for the production of (3R)-3-(2,3-dihydro-1-benzofuran-5-yl)-1,2,3,4-tetrahydro-9H-pyrrolo[3,4-b] -quinolin-9-one, an intermediate in the synthesis of PDE-V inhibitors RWJ387273 (R301249) and RWJ444772 (R290629

Article abstract of DOI:10.1021/op060099f

A scalable and safe process for the oxidative rearrangement of β-carboline to quinolone derivatives, intermediates in the synthesis of PDE-V inhibitors RWJ387273 (R301249) and RWJ444772 (R290629), has been developed.

Full text of DOI:10.1021/op060099f

Organic Process Research & Development 2006, 10, 1275−1281
Lecture Transcripts
Development of a Scalable and Safe Procedure for the Production of
(3R)-3-(2,3-Dihydro-1-benzofuran-5-yl)-1,2,3,4-tetrahydro-9H-pyrrolo[3,4-b]-
quinolin-9-one, an Intermediate in the Synthesis of PDE-V Inhibitors
RWJ387273 (R301249) and RWJ444772 (R290629)
Bert Willemsens,*,† Ivan Vervest,† Dominic Ormerod,† Wim Aelterman,† Christine Fannes,† Narda Mertens,†
Istva´n E. Marko´,‡ and Sebastien Lemaire‡
Chemical DeVelopment, Johnson & Johnson Pharmaceutical Research & DeVelopment, Turnhoutseweg 30, 2340 Beerse,
Belgium, and UniVersite´ Catholique de LouVain, Laboratoire de Chimie Organique, Place Louis Pasteur, 1, 1348
LouVain-la-NeuVe, Belgium
Abstract:
A scalable and safe process for the oxidative rearrangement of
â-carboline to quinolone derivatives, intermediates in the
synthesis of PDE-V inhibitors RWJ387273 (R301249) and
RWJ444772 (R290629), has been developed.
Introduction
R301249 (RWJ387273) and R290629 (RWJ444772),
depicted in Figure 1, are two PDE-V inhibitors active in the
treatment of erectile disfunction.1 An important issue in the
Figure 1. Structures of PDE-V inhibitors R301249 and
R290629.
treatment of erectile disfunciton with PDE-V inhibitors is
the selectivity versus other enzymes of the phosphodiesterase
family. Selectivity versus other enzymes can cause adverse
cardiovascular side effects as well as visual disturbances by
PDE-1 and PDE-6 inhibition. R301249 and R290629 are
potent compounds with a greater selectivity for PDE-5 versus
other PDE isozymes.
rearrangement was performed on the benzyl-protected com-
pound 8. The resulting quinolone 9 was then deprotected
and subsequently arylated. As the final selection of the
candidate still had to be made during the course of the
investigation for a scalable method, it was decided a common
penultimate would be pursued which in the final step could
then be arylated to the desired API. This resulted in the
approach as depicted in Scheme 3. In this approach, the
oxidative rearrangement is performed on a protected â-car-
boline. Among the protective groups benzyl, tert-butoxy-
carbonyl, and benzyloxycarbonyl were considered.
Compounds of the R290629 and R301249 type were
identified by medicinal chemistry as minor byproducts during
the N-alkylation of the corresponding indoles as a result of
a Winterfeldt2 oxidation due to the presence of small amounts
of oxygen.3 Winterfeldt oxidation of 1,2,3,4-tetrahydro-â-
carbolines is well-known.4 With acyl-â-carbolines, however,
Winterfeldt oxidation failed to give the desired quinolones,5
Results and Discussion
Medicinal chemistry routes for both compounds are shown
in Schemes 1 and 2. The â-carboline derivative 6 was syn-
thesized starting from 2,3-dihydrobenzofurane which, after
bromination and formylation by a Grignard reaction with
dimethylformamide, gives the aldehyde 3. Condensation with
tryptamine then leads to the imine 4 which is closed under
Pictet-Sprengler conditions to the racemic â-carboline 5.
Resolution of 5 was performed either by diastereomeric salt
formation with acetyl-D-leucine or by chiral chromatography.
The aim of this publication is to focus on the most crucial
step of the synthesis namely the oxidative rearrangement of
the indole derivatives to the quinolone derivatives. To
synthesize R301249 this oxidative rearrangement was per-
formed on the arylated indole derivative 7 yielding directly
the final compound, whereas for R290629 the oxidative
(2) Winterfeldt, E. Liebigs Ann. Chem. 1971, 745, 23-30. Warneke, J.;
Winterfeldt, E. Chem. Ber. 1972, 105, 2120-2115.
(3) Sui, Z.; Macielag, M. J.; Guan, J.; Jiang, W.; Lanter, J. C.; Walsh, S. P.;
Fiordeliso, J. J.; Qiu, Y.; Kraft, P.; Bhattacharjee, S.; Craig, E.; Haynes-
Johnson, D.; John, T. M.; Clancy, J. J. Med. Chem. 2003, 46, 441-444.
Sui, Z.; Macielag, M. J.; Guan, J.; Jiang, W.; Zhang, S.; Qiu, Y.; Kraft, P.;
Bhattacharjee, S.; Haynes-Johnson, D.; Clancy, J. J. Med. Chem. 2002,
45, 4094-4096.
(4) Boch, M.; Korth, T.; Nelke, J. M.; Pike, D.; Radunz, H.; Winterfeldt, E.;
Chem. Ber. 1972, 105, 2126-2142.
(5) Jiang, W.; Zhang, X.; Sui, Z. Org. Lett. 2003, 5, 43-46.
* Author for correspondence. E-mail: bwillem2@prdbe.jnj.com.
† Chemical Development, Johnson & Johnson Pharmaceutical Research and
Development.
‡ Universite´ Catholique de Louvain.
(1) Sui, Z.; Macielag, M. J.; Guan, J.; Jiang, W.; Lanter, J. C. PCT Int. Appl.
2001 WO0187882.
10.1021/op060099f CCC: $33.50 © 2006 American Chemical Society
Published on Web 07/12/2006
Vol. 10, No. 6, 2006 / Organic Process Research & Development
•
1275

Scheme 1. Synthesis of the â-carboline derivative T2513
Table 1. Initiation temperature of the decomposition of
KO2-solvent mixtures with and without ET4NBr
worst case the temperature would rise 92 °C resulting in an
MTSR8 of 147 °C. This means that in case of a cooling
failure the decomposition reaction will be triggered. As the
boiling point of the solvent DMF is higher than the MTSR,
evaporative cooling cannot serve as a safety barrier.
To reduce the risk involved with this reaction other
solvents were screened. As shown in Table 1, the initiation
temperature for the decomposition of KO2-solvent mixtures
with or without Et4NBr is considerably higher in DMI as
compared to DMF, CH3CN, and DMA. Although the severity
of the decomposition remained the same, the probability for
the decomposition reaction to occur was lowered. To
decrease the severity of a possible decomposition, further
process development on the reaction in DMI was performed.
The excess of KO2 was reduced from 8 to 6 equiv, the
amount of Et4NBr was reduced from 8 to 1 equiv, and the
reaction was performed more diluted, i.e., 7 L/mol instead
of 2.3 L/mol and at 50 °C. As a result of this, the reaction
time became longer, and it now took 24 h for complete
conversion.
Additional RC1 and Phi-tec experiments on the adapted
reaction conditions were performed. The RC1 experiment
revealed that the reaction enthalpy was about 500 kJ/mol,
but as the reaction is slow the specific heat release is only
1.5 W/L and can easily be controlled by cooling in a
production plant. However, as shown in Figure 2 there is
about 90% of thermal accumulation, so this is in fact a batch
process and batch processes are not easy to control.
The graph of the Phi-tec run, as depicted in Figure 3,
shows that the reaction or exothermic decomposition starts
without Et4NBr (°C) with Et4NBr (°C)
dimethylformamide
acetonitrile
dimethylacetamide
dimethylimidazolidinone
80
90
75
70
85
80
120
100
and a more general applicable method to oxidize 1,2,3,4-
tetrahydro-â-carbolines was needed.
Oxidative cleavage of the 2,3 bond of indoles with KO2
in the presence of a phase transfer catalyst is well described
in the literature.6 So it was decided to also test this reagent
on the N-benzyl-substituted â-carboline 11. With 8 equiv of
KO2 and 8 equiv of Et4NBr in DMF at 55 °C, quinolone 14
was obtained in low yield (20-25%). In a typical experiment
a slurry of KO2 and Et4NBr in DMF was heated to 55 °C,
and a solution of 11 in DMF was added over 20 min after
which the reaction mixture was stirred for another 3 h.
Earlier safety investigations on the reaction mixture
however had shown a thermal instability at relatively low
temperatures. In an ARC-experiment on the reaction mixture
with 11, an exothermic decomposition starting at 39 °C with
increased decomposition and pressure buildup at 60 °C was
observed.7
An RC1 experiment on the desired reaction revealed a
75% heat accumulation after all 11 was added. The reaction
generates a moderate heat flow over about 3 h, and in the
(6) Baloch-Hergovich, E.; Spier, G. Tetrahedron Lett. 1982, 23, 4473-4475.
Itakura, K.; Uchida, K.; Kawakishi, S. Tetrahedron Lett. 1992, 33, 2567-
2570.
(8) MTSR: maximum temperature of the synthetic reaction. In our case 55
°C + 92 °C (adiabatic temperature rise) ) 147 °C. The MTSR is dependent
on the amount of unreacted reactants at the moment of the cooling failure.
(7) Li, X.; Branum, S.; Russell, R.; Jiang, W.; Sui, Z. Org. Process Res. DeV.
2005, 9(5), 640.
1276
•
Vol. 10, No. 6, 2006 / Organic Process Research & Development

Scheme 2. Medicinal chemistry route to R301249 and R290629
at 52 °C with a total, Phi corrected, adiabatic temperature
rise of 192 °C. Already at 70 °C the self-heat rate is 0.56
°C/ min with a maximum of 102 °C/min at 150 °C. The
conclusion of this adiabatic experiment was that the reaction
temperature should not exceed 70 °C, the temperature at
which the control on the reaction can be lost and a runaway
cannot be avoided. Besides these safety aspects, the isolated
yield of quinolone 15 was only 18-20%. Thus, it was
decided to abandon the KO2 approach and to look for safer
oxidizing agents.
Oxidation of 11 with m-CPBA mainly resulted in the
N-oxide of the starting material. To avoid this, the nonbasic
BOC and Cbz derivatives 12 and 13 were considered.
Typically, the m-CPBA oxidation was performed by dosing
3 equiv of a 0.166 molar solution of m-CPBA in dichlo-
romethane to a 0.166 molar solution of 12 or 13 in
dichloromethane at 20 °C over a 1 h period.
From this experiment we may conclude that the reaction
with m-CPBA is dose controlled with respect to the dosing
end point and the immediate heat release will be dependent
on the dosing time so that this reaction can be performed
safely on a pilot plant scale. Moreover, the reflux of
dichloromethane will serve as a safety barrier before the
MTSR is reached. After workup, the residue of the reac-
tion contained 20-30% of the keto-lactam 17 (Scheme 4)
which on treatment with KOH was quantitatively converted
to quinolone 15. Attempts to isolate the keto-lactam 17
or quinolone 15 by crystallization failed, and a chromato-
graphic purification was required to obtain 15 with an
acceptable purity. This drawback was solved by using the
benzyloxycarbonyl as a protecting group. Thus performing
the m-CPBA oxidation on the Cbz-protected â-carboline
13 allowed us to isolate the keto-lactam 20 by crystal-
lization from ethyl acetate. However due to the poor
selectivity of m-CPBA, formation of byproducts was
substantial resulting in a keto-lactam yield of only (20%.
The graph of an RC1 experiment of the m-CPBA
oxidation of the BOC-derivative 12 is depicted in Figure 4.
Vol. 10, No. 6, 2006 / Organic Process Research & Development
•
1277

Scheme 3. Approach to common penultimate
Scheme 4. Keto-lactam formation after mcpba-oxidation of T2542
So it was decided a more selective oxidizing agent should
be looked for.
Oxidation of 2,3-substituted indoles has been reported by
Hino et al.,9 and one of the formed intermediates was the
hydroxy-imine 18 as shown in Scheme 5. Hino showed that
this intermediate could be converted to the keto-lactam 19
in good yields by further oxidation with m-CPBA. Epoxi-
dation of the indole could be done with dimethyldioxirane
(DMD) as this is known to be a good epoxidizing agent. It
can be formed in situ from acetone and oxone at controlled
pH.10
(10) (a) Adam, W.; Chan, Y.-T.; Cremer, D.; Gauss, J.; Scheutzow, D.; Schlinder,
M.; J. Org. Chem. 1987, 52, 2800-2803. (b) Murray, R. W.; Jeyaraman,
R. J. Org. Chem. 1985, 50, 2847-2853.
(9) Hino, T.; Yamaguchi, H.; Matsuki, K.; Nakano, K.; Sodeoka, M.;
Nakagawa, M. J. Chem. Soc., Perkin Trans. 1 1983, 141-147.
1278
•
Vol. 10, No. 6, 2006 / Organic Process Research & Development

Figure 2. Graph of RC1-experiment of KO2 oxidation of T2514 in DMI.
Figure 3. Phi-tec run on mixture of T2514, KO2, and ET4NBr in DMI.
A Phi-Tec experiment on in situ generated DMD solutions
in water/ethyl acetate/acetone, NaHCO3 showed that these
solutions were thermally stable up to 70 °C. So a stepwise
oxidative rearrangement was developed in which a solution
of 13 in ethyl acetate/acetone was first reacted with oxone
followed by m-CPBA oxidation to the keto-lactam 20.
Starting from â-carboline 6, the keto-lactam 20 was isolated
by crystallization from ethyl acetate with an overall yield of
36% on a 46 mol scale (Scheme 6). Though the yield of
this reaction was nearly doubled it still seemed that the
m-CPBA sequence was responsible for the still relatively
low yield. After the DMD-oxidation step the reaction mixture
contained up to 60-70% (HPLC area percent) of the
hydroxy-imine 22. Also in this type of oxidation most of
the impurities started to form upon addition of m-CPBA.
The most important impurity was the hydroxy-ketone 23
(identification in reaction mixture by LC-MS). This impu-
rity, formed for 25 mol %, could possibly be formed by
attack of m-CPBA on the enamine rather than on the imine
as shown in Scheme 7.
Vol. 10, No. 6, 2006 / Organic Process Research & Development
•
1279

Figure 4. Graph of RC1-experiment of mcpba-oxidation of T2542 in CH2Cl2.
Scheme 5. Oxidation of 2,3-substituted indoles as reported by Hino et al.
Scheme 6. Reaction scheme as applied in pilot plant
deprotection provided us with a sufficient penultimate for
delivery of the targeted amounts of APIs.
Conclusion
In conclusion, we have developed a process for the
oxidative rearrangement of â-carboline derivatives to qui-
nolone derivatives that has been successfully applied on a
•
40 to 50 mol scale. As compared to the original KO2
oxidation, the yield has been doubled, and even more
important is that the process can be performed safely.
Experimental Section
General. Starting materials, reagents, and solvents were
obtained from commercial suppliers and were used without
any further purification.
1H NMR spectra were recorded at 400 MHz on a Bruker
Avance-400 instrument.
Benzyl (3R)-3-(2,3-Dihydro-1-benzofuran-6-yl)-2,7-di-
oxo-1,2,3,5,6,7-hexahydro-4H-1,4-benzodiazonine-4-car-
boxylate (20). A 250 mL vessel was charged successively
with 1-(R)-1-(2,3-dihydrobenzofuran-5-yl)-2,3,4,9-tetrahydro-
1H-â-carboline (6) (5.22 g, 0.018 mol), ethyl acetate (150
mL), and triethylamine (2.1 g, 0.0208 mol). The mixture was
heated to 40-45 °C. Then the heating was stopped, and
benzylchloroformate (3.55 g, 0.0208 mol) was added drop-
wise. The addition was slightly exothermic, and the tem-
perature rose from 42 to 51 °C. The mixture was stirred for
Using this stepwise oxidation approach, six pilot plant
batches were performed successfully to give keto-lactam 20
which after ring closure to the quinolone 21 and subsequent
1280
•
Vol. 10, No. 6, 2006 / Organic Process Research & Development

Scheme 7. Possible pathway for the formation of the
hydroxy-ketone impurity
mixture was stirred for another 1.5 h at 0-5 °C and then
washed successively with water (76 mL) and with a solution
of NaHCO3 (6 g) in water (76 mL). After checking for
peroxides, the organic layer was evaporated to nearly dryness,
and the resulting oily residue was crystallized from ethyl
acetate (40 mL). After stirring for 18 h at 10-25 °C the
precipitate was filtered, washed with ethyl acetate (2 mL),
and dried in vacuo at 50 °C for 20 h yielding 3 g (36.5%) of
the title compound.
1H NMR (400 MHz, DMSO-d6) showed a mixture of
rotamers: δ ppm [2.75-2.90] (m, 2H); [3.08-0.019] (m,
3H); [4.45-4.53] (o, 3H); [5.01-5.24] (4 × d (o), 2H); 5.93,
5.96 (2 × s, 1H); [6.67-6.72] (2 × d (o), 1H); [6.84-6.90]
(2 × d (o)n 1H); 7.01, 7.03 (2 × s, 1H); [7.19-7.67] (o.m.
9H); 10.84, 10.86 (2 × s, 1H)
Probably other rotamers visible: [3.38-4.28] (m); [4.60-
4.77] (2 × d); 5.54, 5.61 (2 × s); 9.68, 9.73 (2 × s). o:
overlapping.
(3R)-3-2,3-Dihydro-1-benzofuran-5-yl)-1,2,3,4-tetrahy-
dro-9H-pyrrolo[3,4-b]quinolin-9-one (10). A 1 L flask was
charged successively with keto-lactam 20 (45.6 g, 0.1 mol)
and ethanol (400 mL). To the resulting suspension was added
a NaOH 50% aqueous solution (5.8 mL, 0.11 mol), and the
mixture was stirred for 1 h at 20-25 °C. Subsequently
concentrated HCl (9.9 mL, 0.11 mol) was added dropwise,
and after stirring for another 30 min the precipitated salts
were filtered and washed with ethanol (100 mL). The
combined filtrates were transferred into a hydrogenation
vessel, and concentrated HCl (9 mL, 0.1 mol) was added.
The solution was flushed with nitrogen, and palladium on
charcoal (5%-50% aqueous, 5 g) was added and subse-
quently hydrogenated at 25-30 °C. When deprotection was
complete the mixture was filtered. NH4OH (10 mL, 0.13 mol)
was added to filtrate, followed by a dropwise addition of
water (550 mL). The solution was then seeded with qui-
nolone (10) (0.1 g). When crystallization had started, another
portion of water (550 mL) was added dropwise in 1 h. The
mixture was stirred for another 18 h after which the
precipitate was filtered and washed with water (50 mL). After
drying at 50 °C for 18 h, 22.7 g (75%) of the title compound
was obtained.
another 2 h at 51 to 42 °C. Water (18 mL) was added, and
the layers were separated while keeping the temperature at
35-40 °C. To the organic layer acetone (107 mL) was added,
and the mixture was cooled to 20-25 °C. Subsequently,
water (36 mL) and NaHCO3 (4.82 g, 0.057 mol) were added,
and the mixture was further cooled to 0-5 °C. A solution
of oxone (15.5 g, 0.025 mol) in water (60 mL) was added
by means of a dosing pump over a 2 h period at 0-5 °C.
After the addition was complete, the mixture was stirred for
another 2 h at 0-5 °C. Subsequently a solution of m-
chloroperbenzoic acid 71% (5.34 g, 0.217 mol) in dichlo-
romethane (62 mL) was added in 15 min at 0-5 °C. The
1H NMR (400 MHz, DMSO-d6) δ ppm 3.13 (t, J ) 8.69
Hz, 2 H), 4.02 (dd, J ) 12.97, 1.38 Hz, 1 H), 4.18 (dd, J )
12.97, 2.90 Hz, 1 H), 4.50 (t, J ) 9.06 Hz, 2 H), 5.37 (dd,
J ) 2.77, 1.76 Hz, 1 H), 6.73 (d, J ) 8.31 Hz, 1 H), 7.07
(dd, J ) 8.31, 1.76 Hz, 1 H), 7.17 (d, J ) 1.26 Hz, 1 H),
7.26-7.31 (m, J ) 8.06, 5.79, 2.01 Hz, 1 H), 7.53-7.57
(m, 1 H), 7.56-7.60 (m, 1 H), 8.11-8.15 (m, 1 H), 11.58
(s, 1 H).
Received for review May 10, 2006.
OP060099F
Vol. 10, No. 6, 2006 / Organic Process Research & Development
•
1281