Facile reductive amination of aldehydes with electron-deficient anilines by acyloxyborohydrides in TFA: Application to a diazaindoline scale-up

Article abstract of DOI:10.1021/jo900157z

A scale-up of diazaindoline 1 was achieved in four stages and 32% overall yield. The key step involved rapid reductive animation of aldehyde 8 with aniline 5 by sodium triacetoxyborohydride (STAB-H) and TFA followed by ring closure of intermediate amine 9

Full text of DOI:10.1021/jo900157z

with NaBH4 afforded the primary alcohol 7. Finally, formation
of the mesylate of 7 and cyclization in situ provided 1.
From a scale-up perspective, we considered use of NaH to
be undesireable owing to handling issues and significant
hydrogen gas production. The ozonolysis of 6 was of even
greater concern in view of the highly energetic ozonide
intermediates3 and cryogenic conditions involved. For these
reasons, we pursued an alternative synthesis of 1 for scale-up
applications with the goals of shortening the synthesis, avoiding
chromatography, and eliminating hazardous procedures.
We proposed a new route to 1 from aniline 5 and the known
aldehyde 84 via the reductive amination/ring-closure sequence
shown in Scheme 1. We felt that the putative amine intermediate
(9) would readily cyclize to 1 in situ under the appropriate
conditions. A few accounts of one-pot reductive amination/
cyclization syntheses of diazaindolines with sodium triacetoxy-
borohydride (STAB-H) and AcOH are known in the literature,5
although yields are variable. Encouraged by these reports, we
prepared a nominal amount of allyldichloropyrimidine 4 by the
route shown in Scheme 1 and oxidatively cleaved the double
bond to afford the requisite aldehyde 8.
Facile Reductive Amination of Aldehydes with
Electron-Deficient Anilines by
Acyloxyborohydrides in TFA: Application to a
Diazaindoline Scale-Up
Eric E. Boros,* James B. Thompson, Subba R. Katamreddy,
and Andrew J. Carpenter
GlaxoSmithKline Research & DeVelopment, FiVe Moore
DriVe, Research Triangle Park, North Carolina 27709
eric.e.boros@gsk.com
ReceiVed January 27, 2009
Reductive amination of 8 with 5 was investigated using
STAB-H6 in CH2Cl2 both with and without AcOH cosolvent.
Reactions were monitored by a combination of HPLC and
LCMS. Adding STAB-H portionwise to a mixture of 5 and 8
in CH2Cl2/AcOH resulted predominantly in reduction of 8 to
the corresponding primary alcohol. The results were improved
by adding a CH2Cl2 solution of 8 dropwise to a mixture of 5
and STAB-H in CH2Cl2/AcOH. Under these conditions, some
formation of 9 (ca. 30% by HPLC) occurred overnight at rt;
however, none of the cyclized product 1 was observed.
Performing the same reaction without AcOH produced very little
of 9 (<5% conversion).
We next evaluated TFA as cosolvent (10-30 equiv) and
found that it gave a much cleaner and faster reaction. Reasonable
formation of 9 was achieved by cooling a mixture of 5 and 8 in
TFA followed by portionwise addition of STAB-H; however,
we found that dropwise addition of a CH2Cl2 solution of 8 to a
mixture of 5 and STAB-H in TFA gave a cleaner reaction and
avoided dialkylation of 5.
Accordingly, STAB-H (1.5 equiv) was dissolved in neat TFA,
and the solution was charged with 5 (1 equiv). After the mixture
was stirred for a few minutes7 at rt, it was cooled to -15 °C,
and a solution of 8 (1.1 equiv) in CH2Cl2 was added dropwise.
The reaction mixture was immediately checked by HPLC and
LC-MS which showed clean formation of 9 (>92%); moreover,
little or no aldehyde reduction was observed. After the mixture
was stirred for 10 min, workup afforded 9 in 83% yield (Table
A scale-up of diazaindoline 1 was achieved in four stages
and 32% overall yield. The key step involved rapid reductive
amination of aldehyde 8 with aniline 5 by sodium triac-
etoxyborohydride (STAB-H) and TFA followed by ring
closure of intermediate amine 9 to compound 1 in the same
pot. These reaction conditions were also applied to facile
reductive aminations with anilines known to have little
reactivity under STAB-H/AcOH conditions. Spectral data
supported the tris(trifluoroacetoxy)borohydride anion (16) as
the active reducing agent.
Diazaindoline 1 is a useful intermediate in the synthesis of
preclinical GPR119 receptor agonists.1,2 Our metabolic pathways
drug discovery team recently required ca. 200 g of 1 for
analogue synthesis and scale-up campaigns. The original five-
step synthesis of 1 (used for medicinal chemistry purposes) and
the four-stage scale-up route are shown in Scheme 1. The first
two steps of both syntheses were similar and involved cycliza-
tion of commercially available diethyl allylmalonate (2) to
pyrimidine 3 followed by deoxychlorination to 4. At this point,
the two routes diverged.
In the original synthesis, commercially available sulfonyla-
niline 5 was deprotonated with NaH, and the resulting sodium
salt was reacted with 4 to give anilinopyrimidine 6. Low
temperature (-78 °C) ozonolysis of 6 followed by treatment
(3) Ragan, J. A.; am Ende, D. J.; Brenek, S. J.; Eisenbeis, S. A.; Singer,
R. A.; Tickner, D. L.; Teixeira, J. J., Jr.; Vanderplas, B. C.; Weston, N. Org.
Proc. Res. DeV. 2003, 7, 155–160.
(4) Montgomery, J. A.; Hewson, K. J. Med. Chem. 1967, 10, 665–667.
(5) Kempson, J.; Pitts, W. J.; Barbosa, J.; Guo, J.; Omotoso, O.; Watson,
A.; Stebbins, K.; Starling, G. C.; Dodd, J. H.; Barrish, J. C.; Felix, R.; Fischer,
K. Bioorg. Med. Chem. Lett. 2005, 15, 1829–1833.
(1) Katamreddy, S. R.; Caldwell, R. D.; Heyer, D.; Samano, V.; Thompson,
J. B.; Carpenter, A. J.; Conlee, C. R.; Boros, E. E.; Thompson, B. D. Patent
WO 2008/008887 A2; Chem. Abstr. 2008, 148, 144797.
(2) For recent reviews of GPR119 receptor agonists, see: (a) Lauffer, L.;
Iakoubov, R.; Brubaker, P. L. Endocrinology 2008, 149, 2035–2037. (b) Fyfe,
M. C. T.; McCormack, J. G.; Overton, H. A.; Procter, M. J.; Reynet, C. Expert
Opin. Drug DiscoV. 2008, 3, 403–413.
(6) Abdel-Magid, A. F.; Mehrman, S. J. Org. Proc. Res. DeV. 2006, 10, 971–
1031.
(7) Solutions prepared from STAB-H and TFA lose their effectiveness over
time and should be used immediately following preparation for optimum results.
10.1021/jo900157z CCC: $40.75
Published on Web 04/07/2009
2009 American Chemical Society
J. Org. Chem. 2009, 74, 3587–3590 3587

SCHEME 1
TABLE 1. Reaction Conditions and Results for Reductive
Aminations of Aldehydes 8, 11, and 14 with Electron-Deficient
Anilines 5, 10, and 13 by Solutions Prepared from STAB-H and
TFAa
of 14 with 13 occurred within minutes (>87% by HPLC); the
product (15) was isolated as its hydrocholoride salt in 82% yield
(Table 1).
The mechanism of reductive aminations by solutions prepared
from STAB-H in TFA is difficult to determine; however, the
actual reducing agent does not appear to be STAB-H. For
example, dissolving STAB-H (a white powder) in excess TFA
followed by rotary evaporation and drying under high vacuum
afforded a viscous oil; mass balance of this material together
with its 13C NMR spectrum suggested significant substitution
of the STAB-H acetate groups by trifluoroacetic acid; moreover,
electrospray mass spectral analysis of a solution of 13 and
STAB-H in TFA showed strong base peaks matching aniline
13 and the tris(trifluoroacetoxy)borohydride anion (16).
Interestingly, treating excess TFA with NaBH4 (caution:
significant hydrogen gas evolution occurs along with a mild
exotherm) is known to generate 16.10 Treatment of this solution
with 13 and subsequent dropwise addition of a solution of 14
in CH2Cl2 also effected rapid reductive amination to 15;
however, not all of 13 was consumed, and the reaction produced
more byproducts compared to the STAB-H/TFA method. This
result also supported the intermediacy of 16 as the reducing
agent in the reductive amination reactions summarized in
Table 1.
aniline
(equiv)
aldehyde
(equiv)
STAB-H
(equiv)
%
time
product
yieldb
5 (1)
8 (1.05)
8 (1.1)
11 (2.3)
14 (1)
(1.5)
(1.4)
(4)
5 minc
9
1
12
15
83
69
92
82g
5 (1)d
10 (1)
13 (1)
18 he
10 minf
10 minf
(2)
a A solution of aniline and STAB-H in neat TFA (10-30 equiv) was
treated dropwise with a solution of aldehyde in CH2Cl2. b Isolated yield.
c -15 °C. d 0.96 mol scale. e -15 °C to rt overnight. f rt. g Hydro-
chloride salt.
1). Peforming the same procedure at -15 °C and allowing the
reaction mixture to warm to rt with stirring overnight effected
complete cyclization of 9 to diazaindoline 1 in the same pot.
Interestingly, whereas the small amount of 9 generated under
STAB-H/AcOH conditions did not readily cyclize to 1; this
cyclization occurred smoothly with TFA as cosolvent. Thus,
TFA facilitated both the reductive amination and ring-closure
steps.
To determine if our results with STAB-H/TFA were unique
for reactants 5 and 8 or were applicable to other weakly basic
amines, we explored the reaction of 2,4-dinitroaniline (10) with
cyclohexanecarboxaldehyde (11) (Scheme 2). This reaction is
reported8 to occur slowly under STAB-H/AcOH conditions; the
published yield of 12 is 61% after 4 d with 5 equiv of STAB-H
and 4-5 equiv of 11. In contrast, the STAB-H/TFA conditions
(Table 1) produced complete reaction immediately following
aldehyde addition. The product (12) was isolated in 92% yield
following workup and trituration with hexanes.
We also investigated reaction of 2,4,6-trichloroaniline (13)
with benzaldehyde (14) (Scheme 2). There is speculation8 that
STAB-H-mediated reductive aminations of aliphatic aldehydes
with 13 proceed through an enamine intermediate; this is
primarily based on the lack of reactivity of 138,9 (under STAB-
H/AcOH conditions) with benzaldehyde (14) which cannot form
an enamine. Remarkably, the STAB-H/TFA reductive amination
Assuming the reductive amination reactions shown in Table
1 proceed by a mechanism involving hydride reduction of an
iminium ion, it appears that TFA solutions of acyloxyborohy-
drides promote this process, especially with severely electron
deficient anilines.
With practical procedures in hand for synthesis of 1, we
turned our attention to its scale-up. All of the scaled reactions
were performed in graduated 6 or 16 L fixed-glass, jacketed
laboratory reactors (quench volumes ca. 10-30 L). Workups
and phase splits were performed in the reactor; however, for
(8) Abdel-Magid, A. F.; Carson, K. G.; Harris, B. D.; Maryanoff, C. A.;
Shah, R. D. J. Org. Chem. 1996, 61, 3849–3862.
(9) Bae, J. W.; Lee, S. H.; Cho, Y. J.; Yoon, C. M. J. Chem. Soc., Perkin
Trans. 1 2000, 145–146.
(10) Yamada, K.; Takeda, M.; Iwakuma, T. J. Chem. Soc., Perkin Trans. 1
1983, 265–270.
3588 J. Org. Chem. Vol. 74, No. 9, 2009

SCHEME 2
fluoroacetoxy)borohydride anion (16) as the active reducing
agent in these transformations. Details of the medicinal chem-
istry synthesis of 1 and the structure-activity relationships of
GPR119 receptor agonists derived from this compound are
planned for a future publication.
Experimental Section
(4,6-Dichloro-5-pyrimidinyl)acetaldehyde (8). A 16 L fixed-glass
jacketed laboratory reactor (32 L quench volume) was charged with
4 (274 g, 1.45 mol), acetone (3.8 L), and water (3.8 L). The solution
was cooled to 15 °C (jacket control ) 15 °C), and K2OsO4 ·(2 H2O)
(18 g, 49.3 mmol) was added followed by the addition of NaIO4
(1.24 kg, 1.8 mol) in eight portions over 1 h while maintaining the
reaction temperature below 40 °C. Following this addition, the
reaction mixture was stirred at rt for 1 h. The resulting suspension
was filtered; the filtrate was returned to the reactor, and the acetone
was removed by distillation at reduced pressure. The aqueous
phase was extracted with CH2Cl2 (5 × 1 L), and the combined
organic layers were washed with 10% Na2S2O3 (2 × 3.5 L) and
brine, dried over Na2SO4, filtered, and concentrated to afford 84 as
a light amber solid (203.7 g, 74% yield). This material was used
without purification. An analytical sample was obtained by recrys-
tallization from CH2Cl2/heptane: mp 89-91 °C (lit.4 mp 84-86
°C); 1H NMR (DMSO-d6, 400 MHz) δ 9.72 (s, 1H), 8.86 (s, 1H),
4.21 (s, 2H); Anal. Calcd for C6H4Cl2N2O: C, 37.73; H, 2.11; N,
14.67. Found: C, 37.81; H, 2.02; N, 14.77.
convenience, the final organic phase was often removed from
the reactor, dried, and concentrated by rotovap.
Allyldihydroxypyrimidine 34 was prepared on 2.2 mol scale
from formamidine hydrochloride and diethyl allylmalonate 2
in the presence of sodium methoxide/methanol. Solvent removal
and neutralization with 6 N HCl afforded 3 which was collected
by filtration in 76% yield. Deoxychlorination of 3 with POCl3
at reflux temperature effected clean conversion to the dichlo-
ropyrimidine 44 in 82% yield.
We next evaluated oxidative cleavage of allylpyrimidine 4
to aldehyde 8.4 The two most common methods employed for
this type of transformation are ozonolysis (which we wished to
avoid) or use of catalytic osmium tetroxide in the presence of
stoichiometric oxidants11 such as NaIO4 at ambient temperatures.
We chose to evaluate the latter conditions, but replaced OsO4
with K2OsO4 ·(2 H2O), which is a more stable, less volatile,
and easier to handle solid.12 The scaled synthesis of 8 was
accomplished by portionwise addition of NaIO4 to a mixture
of 4 (1.45 mol) in acetone/water containing 3 mol % of
K2OsO4 ·(2 H2O). Following an aqueous reductive workup,
aldehyde 8 was obtained as a solid in 74% yield.
Reductive amination of 8 with 5 was performed on a ca. 1
mol scale. We chose to first dissolve aniline 5 in TFA (12 equiv)
with cooling and then add the STAB-H portionwise at -15 °C.
Following the STAB-H charge, the mixture was stirred for a
few minutes, and a CH2Cl2 solution of 8 was added dropwise;
after the reaction mixture was warmed to rt and stirred overnight,
formation of 1 was complete. Following workup, the crude
material was triturated with EtOAc/MeOH to afford 1 as a
crystalline solid (216 g, 69% yield).
[2-(4,6-Dichloro-5-pyrimidinyl)ethyl][2-fluoro-4-(methylsulfo-
nyl)phenyl]amine (9). Compound 5 (1 g, 5.29 mmol) was dissolved
in TFA (5 mL) at rt, and the solution was cooled to -15 °C;
STAB-H (1.68 g, 7.93 mmol) was added portionwise. The mixture
was stirred for 10 min, and a solution of 8 (1.06 g, 5.55 mmol) in
CH2Cl2 (10 mL) was added dropwise. After 5 min, the mixture
was poured into ice-cold satd NaHCO3 solution. The layers were
separated, and the aqueous layer was extracted with CH2Cl2. The
combined CH2Cl2 layers were washed with satd NaHCO3, dried
over MgSO4, filtered, and concentrated to afford a solid which was
triturated with hexanes and filtered to afford 9 as a light beige solid
(1.6 g, 83% yield). An analytical sample was obtained by
1
recrystallization from CH3CN: H NMR (DMSO-d6, 400 MHz) δ
8.77 (s, 1H), 7.51 (dd, J ) 8, 2 Hz, 1H), 7.48 (dd, J ) 11, 2 Hz,
1H), 6.94 (br t, J ) 8 Hz, 1H), 6.72 (br m, 1H), 3.50 (br q, J ) 7
Hz, 2H), 3.08 (s, 3H), 3.06 (t, J ) 7 Hz, 2H); 13C NMR (DMSO-
d6, 101 MHz) δ 162.2, 156.8, 150.0 (d, JCF ) 243 Hz), 141.6 (d,
JCF ) 12 Hz), 131.2, 126.5 (d, JCF ) 6 Hz), 125.7 (d, JCF ) 2 Hz),
114.0 (d, JCF ) 21 Hz), 110.8 (d, JCF ) 4 Hz), 44.8, 40.0 (obscured
by solvent multiplet), 29.9; ESI MS m/z 364 [M + H]+. Anal. Calcd
for C13H12Cl2FN3O2S: C, 42.87; H, 3.32; N, 11.54. Found: C, 42.79;
H, 3.22; N, 11.51.
In conclusion, a practical laboratory scale-up of diazaindoline
1, a useful intermediate in the synthesis of novel GPR119
receptor agonists,1 was achieved in four stages and 32% overall
yield without chromatography. The new route circumvented an
ozonolysis reaction in the first-generation approach and short-
ened the synthesis by one step. A key synthetic discovery
involved the facile reductive amination of aldehyde 8 with
aniline 5 by STAB-H and TFA. The TFA cosolvent facilitated
both the formation of intermediate amine 9 and its subsequent
ring closure to diazaindoline 1. In addition, the STAB-H/TFA
methodology was useful for rapid reductive amination of
aldehydes with other weakly basic anilines (e.g., 10 and 13)
known to have little or no reactivity under STAB-H/AcOH
conditions. Mass and NMR spectral data supported the tris(tri-
4-Chloro-7-[2-fluoro-4-(methylsulfonyl)phenyl]-6,7-dihydro-5H-
pyrrolo[2,3-d]pyrimidine (1). A 6 L fixed-glass jacketed laboratory
reactor (12 L quench volume) was charged with TFA (850 mL).
The contents of the reactor were cooled to -10 °C, and 5 (181 g,
0.96 mol) was added. The reactor’s jacket temperature was set to
-25 °C, and when the solution reached -15 °C, STAB-H (293.7
g, 1.39 mol) was added. After the solution was stirred for a few
minutes, a solution of 8 (203.7 g, 1.07 mol) in CH2Cl2 (550 mL)
was added dropwise over 1 h while maintaining the reaction temp
at ca. -15 °C. Stirring was continued at -15 °C for 30 min, and
the reaction mixture was then allowed to warm to rt overnight.
The reaction mixture was diluted with 1,2-dichloroethane (1 L),
and the majority of CH2Cl2 and TFA was removed by distillation
at reduced pressure (ca. 50 mbar; jacket temperature ) 25-30 °C).
The reaction mixture was diluted with CH2Cl2 (1 L) and neutralized
by slow addition of 10% Na2CO3 (6 L) to pH 7-8. The layers
were separated, and the aqueous layer was extracted with CH2Cl2
(3 × 1 L). The combined organic layers were washed with brine
(1.5 L), dried over Na2SO4, filtered, and concentrated. The crude
(11) Pappo, R.; Allen, D. S., Jr.; Lemieux, R. U.; Johnson, W. S. J. Org.
Chem. 1956, 21, 478–479.
(12) Whitehead, D. C.; Travis, B. R.; Borhan, B. Tetrahedron Lett. 2006,
47, 3797–3800.
J. Org. Chem. Vol. 74, No. 9, 2009 3589

material was stirred with 1:1 EtOAc/MeOH (250 mL) for 2 h, and
mL) at rt; STAB-H (424 mg, 2mmol) was added, and the mixture
was stirred for 5 min at rt. A solution of 14 (106 mg, 1 mmol) in
CH2Cl2 (1 mL) was then added slowly dropwise over 5 min. The
reaction mixture was stirred for 10 min at rt and then poured into
a mixture of ice and satd NaHCO3 solution. The layers were
separated, and the aqueous phase was extracted with CH2Cl2. The
combined organic layers were dried over MgSO4, filtered, and
concentrated to afford a colorless oil. This material was dissolved
in heptane (10 mL); 4 N HCl/p-dioxane (0.22 mL, 0.88 mmol)
was added, and the mixture was stirred for 10 min at rt. The product
(15·HCl) was collected by filtration as a white solid (265 g, 82%
yield): HPLC purity (LCMS) > 99%; 1H NMR (CDCl3, 400 MHz)
δ 9.35 (br, 2H), 7.39 (d, J ) 7 Hz, 2H), 7.27 (s, 2H), 7.30-7.20
(m, 3H), 4.80 (s, 2H); 13C NMR (CD3OD, 101 MHz) δ 134.4, 133.9,
131.7, 129.7, 129.3, 129.1, 128.9, 128.6, 52.5; ESI MS m/z 286
[M + H]+ (100), 288 [M + 2 + H]+ (97), 290 [M + 4 + H]+
(32).
compound 1 was collected by filtration as a light beige crystalline
1
solid (216 g, 69% yield): mp 165-167 °C; H NMR (DMSO-d6,
400 MHz) δ 8.29 (s, 1H), 7.94 (br t, J ) 8 Hz, 1H), 7.89 (dd, J )
11, 2 Hz, 1H), 7.79 (dd, J ) 8, 2 Hz, 1H), 4.20 (t, J ) 8 Hz, 2H),
3.27 (s, 3H), 3.22 (t, J ) 8 Hz, 2H); 13C NMR (DMSO-d6, 101
MHz) δ 166.2, 158.0, 155.5 (d, JCF ) 254 Hz), 152.7, 139.3 (d,
JCF ) 6 Hz), 132.5 (d, JCF ) 11 Hz), 126.6 (d, JCF ) 2 Hz), 124.2
(d, JCF ) 3 Hz), 120.1, 116.5 (d, JCF ) 23 Hz), 50.7 (d, JCF ) 5
Hz), 44.0, 24.7; ESI MS m/z 328 [M + H]+. Anal. Calcd for
C13H11ClFN3O2S: C, 47.64; H, 3.38; N, 12.82. Found: C, 47.54;
H, 3.32; N, 12.60.
(Cyclohexylmethyl)(2,4-dinitrophenyl)amine (12). STAB-H (847
mg, 4 mmol) was dissolved in TFA (7 mL) at rt. Compound 10
(183 mg, 15 wt % water, 0.852 mmol) was added, and the mixture
was stirred for 10 min at rt. A solution of 11 (224 mg, 2 mmol) in
CH2Cl2 (5 mL) was added dropwise. After 10 min, the mixture
was poured slowly into ice-cold satd NaHCO3 solution. The layers
were separated, and the aqueous phase was extracted with CH2Cl2.
The combined CH2Cl2 layers were washed with satd NaHCO3, dried
over MgSO4, filtered, and concentrated to afford a dark orange oil
which was triturated with hexanes and stirred for 1 h at rt. The
mixture was filtered, washed with hexanes, and dried under high
vacuum to afford 128 as a bright yellow solid (218 mg, 92% yield):
HPLC purity (LCMS) >99%; mp 80-83 °C (lit.8 mp 76-78 °C);
1H NMR (CDCl3, 400 MHz) δ 9.12 (d, J ) 3 Hz, 1H), 8.64 (br,
1H), 8.23 (dd, J ) 10, 3 Hz, 1H), 6.89 (d, J ) 10 Hz, 1H), 3.23
(t, J ) 6 Hz, 2H), 1.89-1.65 (m, 6H), 1.36-1.11 (m, 3H),
1.13-0.98 (m, 2H); ESI MS m/z 278 [M - H]-.
Acknowledgment. We thank David M. Black for performing
the ESI MS analyses on solutions prepared from TFA and
STAB-H.
Supporting Information Available: Procedures for synthesis
of 3 and 4; 1H and 13C NMR spectra (including expansions) of
1, 9, and 15·HCl; analytical HPLC conditions; HPLC chro-
matogram (LCMS) of 15·HCl; ESI mass spectrum of 16. This
material is available free of charge via the Internet at
http://pubs.acs.org.
2,4,6-Trichloro-N-(phenylmethyl)aniline9 Hydrochloride (15 ·HCl).
Compound 13 (196 mg, 1 mmol) was dissolved in neat TFA (1
JO900157Z
3590 J. Org. Chem. Vol. 74, No. 9, 2009