Carboxylate-directed Kumada coupling of an acetaldehyde synthon with 2-bromobenzoates used towards the synthesis of isochromanes

Article abstract of DOI:10.1055/s-2007-985572

This letter describes the Kumada coupling of bromobenzoic acid derivatives with the acetaldehyde enolate synthon (1,3-dioxolan-2-ylmethyl)magnesium bromide. The lithium carboxylate salt shows the highest reactivity in the coupling reaction while addition

Full text of DOI:10.1055/s-2007-985572

LETTER
2179
Carboxylate-Directed Kumada Coupling of an Acetaldehyde Synthon with
2-Bromobenzoates Used towards the Synthesis of Isochromanes
Kumada
o
C
oupling of
a
an
A
cetald
n
e
hyde Synth
n
on with 2-Br
i
omoben
s
z
oates N. Houpis,*1a Jean-Pierre Van Hoeck, Ulf Tilstam1b
Chemical Product Research and Development, Lilly Research Laboratories, rue Granbonpre 11, 1348 Mont-st-Guibert, Belgium
Fax +32(14)605755; E-mail: yhoupis@prdbe.jnj.com; E-mail: ulf.tilstam@cmcsol.com
Received 25 April 2007
Abstract: This letter describes the Kumada coupling of bromo-
O
O
benzoic acid derivatives with the acetaldehyde enolate synthon
(1,3-dioxolan-2-ylmethyl)magnesium bromide. The lithium carbo-
xylate salt shows the highest reactivity in the coupling reaction
while addition of t-BuOLi affords optimal selectivity. During this
work, we have observed a strong directing effect of the sodium
carboxylate function which can be useful in differentiating electron-
ically similar diaryl bromides.
NC
NC
O
OH
1
2
CO2Et
OH
Key words: directed Kumada coupling, carboxylate-directed Pd
cross-coupling, isochromane synthesis
O
O
O
O
NC
NC
O
The efficient synthesis of functionalized isochromane de-
rivatives remains a challenging matter in modern organic
synthesis, as a number of natural products and drug candi-
dates contain such a derivative either in the final com-
pound or as a key precursor.2 We encountered just such a
challenge in our synthesis of the isochromane skeleton 1
(Scheme 1). Our retrosynthetic strategy of the target iso-
chromane core was guided by two goals: (a) to avoid the
use of cyanide on large scale and minimize protecting
group manipulations, and (b) to introduce the stereogenic
center with a reliable asymmetric reduction of a ketoester
precursor (metal- or enzyme-catalyzed). Thus the key
chiral alcohol 2 could be derived via 3, which in turn
could come from the trisubstituted intermediate 4,
containing three orthogonal oxidation functional groups
(protected aldehyde, nitrile and carboxylic acid).
CO2H
CO2Et
3
4
Scheme 1
of 2-bromobenzoic acid derivatives with a Grignard
reagent. Indeed in our case, just as in the Yu work, ligands
(either phosphine or N-heterocyclic carbene) were detri-
mental to the reaction although Pd2dba3 (containing the
weak dibenzylydine acetone ligand) was needed as the
catalyst.
Not only can we demonstrate the directing effect of the
carboxylate (in our work both Li and Na were effective
with different substrates) but we can also show significant
regioselectivity in the presence of other aryl C–Br bonds
(e.g. 2,4-dibromobenzoic acid).
On the way to 4, we observed a powerful directing effect
of the carboxylate group on the Pd-catalyzed Kumada
coupling reaction, which depends on the solvent and addi-
tives and in some cases the counterion.
Resuming our synthetic plan, in order to accomplish our
goals, we started our work with substrates already
containing the nitrile function such as the commercially
During preparation of this manuscript an excellent report
appeared by Yu and co-workers detailing a C–H activa-
tion–Suzuki cross-coupling methodology using a sodium
carboxylate as the directing group.2 They have demon-
strated elegantly that this carboxylate functions as a pow-
erful chelating/activating functionality to accomplish the
insertion of Pd(0) in the absence of ligands into the adja-
cent aryl C–H bond followed by a Suzuki cross-coupling
reaction of the resulting Pd-aryl species (Scheme 2).
Ar
CO2H
CO2Na
R
10% Pd(OAc)2
R
50% BQ
O
Ag2CO3
+
B
K2HPO4
t-BuOH
100 °C
Ar
O
As detailed below we have observed the same activating
effect, also in the absence of ligand, in the cross-coupling
H
O
Ar
O
OH
OH
R
R
SYNLETT 2007, No. 14, pp 2179–2184
0
3
.0
9
.2
0
0
7
BQ = 1,4-benzochinone
Advanced online publication: 13.08.2007
DOI: 10.1055/s-2007-985572; Art ID: G13607ST
© Georg Thieme Verlag Stuttgart · New York
Scheme 2 Prof. Yu’s carboxylate-directed C–H and cross-coupling

2180
I. N. Houpis et al.
LETTER
available toluene derivative 5 (Scheme 3) or the corre-
sponding benzoic acid and subsequently introduce an
acetaldehyde enolate synthon at the ortho position. We
envisioned introducing the latter functionality via a
Kumada coupling between 7 and commercially available
Grignard reagent 8, an acetaldehyde enolate synthon. An
additional goal of this approach was to avoid protection/
deprotection of the acid and utilize it directly in the
Kumada coupling.
catalyst
solvent
O
O
NC
Br
+
additives
ligand
CO2H
BrMg
7
8
O
O
NC
NC
+
CO2H
CO2H
Br
O
N
9
4
O
NC
N
NC
Br
Scheme 4
Br
CF3CO2H
Me
Me
the reaction afforded significant amounts of 9. This was
the first time we observed complete conversion into
product.7 Addition of phosphine ligands or Cu additives
neither improved the selectivity of the reaction nor had
a negative effect on the conversion (Table 1, entries 7
and 8).
5
6
NC
Br
Cl2Ni(bpy)
4% NaOCl
CO2H
MeCN–H2O
7
Scheme 3
Intrigued by the formation of 9 we first confirmed that the
reagent 8 was not contaminated with vinyl magnesium
bromide by NMR examination of the commercial reagent.
Likewise, monitoring the reaction by NMR did not indi-
cate the in situ formation of vinyl Grignard. Furthermore,
control experiments also showed that reaction of 7 with
vinyl magnesium bromide did not give 9 but rather
indicated an attack on the nitrile function and gave
decomposition products.
Direct bromination of the commercially available para-
cyanobenzoic acid was not successful under a variety of
conditions. Consequently a two-step process was devel-
oped (Scheme 3). Bromination of the cyanotoluene 5 was
accomplished by employing Br2-hydantoin in TFA to give
6 in 87% yield. Oxidation to the carboxylic acid was con-
sistently achieved with a Ni catalyst, (bpy)NiCl2, in
MeCN–H2O with bleach as the stoichiometric oxidant in
95% yield.
Having thus convinced ourselves that the undesired by-
product did not come from a contamination or decom-
position of our reagent we postulated the following
mechanism (Scheme 5, path A). Thus the initial oxidative
addition product, trans-Pd(II) species 10, undergoes a
Initial attempts to couple 7 with the stable and commer-
cially available dioxolane Grignard 8 (Scheme 4),2 under
varying Kumada conditions failed to produce any of the
desired product 4. Thus mixing 7 and 8 (1:2 ratio) in the
presence of Ni(II) or Ni(0) precursors and ligands of dif-
ferent steric and electronic properties, under a variety of
solvents and temperatures, gave no reaction as determined
by HPLC monitoring.3 Similarly, using Pd-phosphine cat-
alysts also resulted in recovery of the starting materials.4
This was in contrast to the reaction of other Grignard re-
agents with 7 in the presence of Ni or Pd catalysts. For ex-
ample PhMgBr (2 equiv) quantitatively reacted with the
nitrile function to give the corresponding benzophenone
compound with no trace of the cross-coupling product.
Thus it became clear that the reactivity of 8 was unique
and a special solution would have to be developed to solve
this problem.5
Table 1 Initial Coupling of 7 with 8a
Entry
Catalyst
Additives
Conversionb 4:9
1
2
3
4
5
6
7
8
Pd(OAc)2
PdCl2(PhCN)2
Pd2(dba)3
Pd2(dba)3
Pd2(dba)3
Pd2(dba)3
Pd2(dba)3
Pd2(dba)3
–
45%
65%
98%
53%
77%
60%
80%
25%
0:100
–
0:100
52:48
60:40
31:69
55:45
38:62
18:82
–
P(c-Hex)3
dcppc
dpppc
Cu(acac)2
CuI
Attempts at using Pd(II) catalysts in the absence of
phosphine ligands resulted in the surprising formation of
the styrene derivative 9 instead of the desired coupling
product 4 (Table 1, entries 1 and 2).
Finally, when 7 and 8 were heated to 65 °C in THF6 for
20 hours with tris(dibenzylideneacetone)dipalladium(0)
[Pd2(dba)3] in the absence of ligand (Table 1, entry 3), the
desired product 4 was observed albeit in modest yield as
a The reactions were run in THF at 65 °C with 10% catalyst load and
8 (2.1 equiv).
b Conversion represented in mol ratio determined by quantitative
HPLC analysis.
c dcpp: dichlorophenylphosphine; dppp: 1,3-bis(diphenylphos-
phino)propane.
Synlett 2007, No. 14, 2179–2184 © Thieme Stuttgart · New York

LETTER
Kumada Coupling of an Acetaldehyde Synthon with 2-Bromobenzoates
2181
b-alkoxide elimination to the enol ether complex 11, in-
stead of reductive elimination to 4. An intramolecular
migratory insertion, placing the Pd at the distal carbon,
would generate 12, which can undergo a second b-alk-
oxide elimination to give 9. An alternative possibility
(Scheme 5, path B) is conceivable where 13 is generated
via decomposition of 8 (perhaps induced by Pd) followed
by migratory insertion onto 14 to form the common inter-
mediate 12. This possibility is less likely, as we have
shown that reaction of the commercially available vinyl
ether HO(CH2)2OCHCH2 and 7 in the presence of two
equivalents of MeMgBr, Pd2(dba)3 in THF at 65 °C failed
to produce even traces of 4 or 9. In both scenarios, the re-
sulting Pd(OCH2CH2O)L2 complex is reduced to Pd(0) by
8.
Table 2 Optimized Coupling of 7 with 8 with Pd2(dba)3
Entry
Solvent
THF
Additivesa
LiHMDS
NaHMDS
LiHMDS
KHMDS
BuLi
Conversion 4:9
1
2
65%
69%
76%
23%
98%
97%
94%
98%
96%
96%
72:28
56:44
THF
3
THF–toluene
THF–toluene
THF
73:27
64:36
79:21
84:16
82:18
92:8
4
5
6
THF
t-BuOLi
t-BuOLi
t-BuOLi
t-BuONa
t-BuOK
7
THF–MTBE
THF–dioxane
THF
8
To overcome this side reaction, we reasoned that the
reductive elimination pathway to 4 could be promoted
from a cis-Pd(II) generated by stronger coordination of
the carboxylate function to the Pd.8
9
81:19
80:20
10
THF
a The starting material was dissolved in THF and deprotonated with
the base at 0 °C. The volatiles were evaporated in vacuo and the sol-
vent mixture was adjusted. Pd2(dba)3 was added and the mixture was
heated to 65 °C.
To accomplish this goal, we examined more nucleophilic
salts, than the Mg salts, generated in situ by addition of
bases to 7, in a selection of solvent and solvent mixtures
(Table 2). Indeed formation of the Li-carboxylate of 7
with LiHMDS followed by reaction with 9 afforded an
improved ratio of 4:9 (Table 2, entry 1) albeit in lower
conversion. Generating the carboxylate with n-BuLi
(Table 2, entry 4) improved the conversion while t-BuOLi
offered significant advantage both in terms of conversion
and selectivity (Table 2, entry 6). Finally the combination
of THF–dioxane (7:3) afforded the desired 4 as the major
product of the reaction in 67% isolated yield (Table 2,
entry 8).
Gratified with this result, we proceeded with the scale up
of this reaction only to find that our results varied consid-
erably from reaction to reaction and different batches of t-
BuOLi.9 After examining the usual variations such as or-
der of addition, rate of temperature increase and catalyst
load, we considered the possibility that the titer of the base
might vary thus causing the observed irreproducibility in
the reaction.
Path A
O
O
O
S
S
β-alkoxide
elimination
intramolecular
NC
NC
O
Pd
Pd
CO2MgBr
Heck
S
CO2MgBr
10
11
S
S
O
Pd
NC
8
+
NC
Pd(II)
Pd(0)
O
CO2H
CO2MgBr
12
9
Path B
NC
PdBrL2
BrMgO
CO2MgBr
O
O
"Pd"
14
O
9
12
MgBr
8
13
Scheme 5
Synlett 2007, No. 14, 2179–2184 © Thieme Stuttgart · New York

2182
I. N. Houpis et al.
LETTER
Consequently we decided to preform and isolate the Li-7
carboxylate before subjecting it to the coupling reaction
conditions. This was easily accomplished by reacting 7
with t-BuOLi in THF at ambient temperature to isolate the
salt as a non-hygroscopic, crystalline, THF solvate, in
>90% yield. The salt Li-7 (Table 3) was suspended in
THF–dioxane (7:3) and treated with 8, to produce a homo-
geneous solution which was treated with Pd2(dba)3 (5%)
and heated to 65 °C. A reproducible reaction ensued (up
to 150-g scale) that gave 4 in ca 75% in situ yield by quan-
titative HPLC analysis (4:9 = 90:10). We next examined
the possibility that the remaining trace amounts of alkox-
ide base during the formation of Li-7 might had an effect
on the reaction selectivity and we were pleased to find that
it was indeed the case (Table 3). So, while addition of 5%
of t-BuOLi had little effect on the outcome of the reaction,
when the amount was increased to 20% (Table 3, entry 2),
we observed a relatively fast (5 h) and clean reaction to
give the desired 4 in 83% assay yield with only trace
amounts of the styrene derivate 9.10 Examination of other
additives having different steric and electronic properties
did not improve the ratio of 4:9 or the yield of the
reaction.11 Counterion effects were also minimal with Li
giving the best results. It is, however, worth noting that
TMSONa (Table 3, entry 8) afforded results similar to
t-BuOLi.
Br
O
O
Pd2(dba)3 (3%)
+
THF-Dioxane (7:3)
65 °C
Br
CO2Li
BrMg
21
t-BuONa (1 equiv)
t-BuOH (1 equiv)
O
O
+
Br
CO2H
Br
CO2H
23
22
78%
<3%
Scheme 6
afforded any coupling product under our reaction condi-
tions. Electron-withdrawing groups (7, 18) afforded
excellent yields of the coupling product but electron-
donating groups have a negative impact on the reaction.
So the yield of the coupling product dropped to 45% and
30% with 15 and 16, respectively, while no coupling
reaction occurred with 17.
It is worth noting that 2-chlorobenzoic acid did not react
with 8 under our reaction conditions while the reaction
with 2-iodobenzoic acid gave significant amounts of
decomposition products.
In order to understand the role of the carboxylate as well
as the scope of the reaction, several substrates were
examined under our optimized reaction conditions
(Scheme 2).12 As seen in Scheme 2, the carboxylate func-
tion is crucial to ensure success of the coupling reaction.
Thus neither the ester 19 nor the toluene derivative 20
The ortho selectivity for the cross-coupling reaction of the
2-bromobenzoic acid derivatives was best exemplified by
reaction of the dibromo derivative 21. Under the standard
reaction conditions (Table 3), the product 22 was obtained
Table 3 Effect of Alkoxides on the Coupling of Li-7 and 8
O
O
NC
Br
O
NC
O
Pd2(dba)3 (5%)
NC
+
+
THF–dioxane (7:3)
65 °C
CO2Li·THF
CO2H
BrMg
CO2H
Additives
Li-7
8
4
9
Entry
Additives (mol%)
t-BuOLi (5%)
Conversion (yield)
4:10
1
2
3
4
5
6
7
8
9
94%
91:9
t-BuOLi (20%)
t-BuOLi (50%)
t-BuOK (20%)
NaOMe (20%)
98% (83%)13
88%
95:5
87:13
90:10
89:11
92:8
95%
95%
CF3CH2ONa (20%)
CF3CH2OLi (20%)
TMSONa (20%)
TMSOK (20%)
94%
94%
90:10
93:7
96% (80%)
96%
89:11
Synlett 2007, No. 14, 2179–2184 © Thieme Stuttgart · New York

LETTER
Kumada Coupling of an Acetaldehyde Synthon with 2-Bromobenzoates
2183
in low yield. However, further optimization revealed that In conclusion, we have demonstrated the coupling of the
t-BuONa was most effective for this substrate in the acetaldehyde enolate synthon 8 with bromobenzoic acid
presence of one equivalent of t-BuOH (Scheme 6). Thus derivatives under Pd catalysis with alkoxide additives and
22 could be obtained in 78% isolated yield bearing an have demonstrated the ortho-directing effect of the
additional aromatic halide functionality, which can be carboxylate functionality.
further elaborated to prepare diversely functionalized
trisubstituted benzoic acid derivatives and isochromanes.
References and Notes
Encouragingly, under these new reaction conditions 15
also reacted more effectively to afford 21 in an improved
yield of 62% (Figure 1).
(1) (a) Current address: Johnson and Johnson Pharmaceutical
Research and Development, Chemical Development,
Turnhoutseweg 30, 2340, Beerse, Belgium. (b) Current
address: CMC Solutions, Overhemstraat 3, 3320 Hoegarden,
Belgium.
Br
Br
Me
Br
(2) Giri, R.; Maugel, N.; Li, J.-J.; Wang, D.-H.; Breazzano, S.
P.; Saunders, L. B.; Yu, J.-Q. J. Am. Chem. Soc. 2007, 129,
3510.
MeO
CO2Li
CO2Li
CO2Li
15
16
17
Yield of adduct: 0%
(3) (a) Giles, R.; Green, I. R.; van Eeden, N. Eur. J. Org. Chem.
2004, 4416. (b) Guiso, M.; Bianco, A.; Marra, C.;
Cavarischia, C. Eur. J. Org. Chem. 2003, 3407. (c) Agejas-
Chicharro, J.; Bueno Melendo, A. B.; Camp, N. P.; Gilmore,
J.; Lamas-Peteira, C.; Timms, G. H.; Williams, A. C. PCT
Int., WO 03053948, 2003. (d) Michaelidis, M. R.;
Schoenleber, R.; Thomas, S.; Yamamoto, D. M.; Britton, D.
R.; MacKenzie, R.; Kebabian, J. W. J. Med. Chem. 1991, 34,
2946.
(4) (a) For a historical account of the development of the
Kumada coupling, see: Tamao, K. J. Organomet. Chem.
2002, 653, 23. For publications with comprehensive
references on Pd- and Ni-catalyzed Kumada coupling, see:
(b) Tasler, S.; Lipshutz, B. H. J. Org. Chem. 2003, 68, 1190.
(c) Boelm, V. P. W.; Weskamp, T.; Gstoettmayr, C. W. K.;
Herrmann, W. A. Angew. Chem. Int. Ed. 2000, 39, 1602.
(d) Huang, J.; Nolan, S. P. J. Am. Chem. Soc. 1999, 121,
9889.
Yield of adduct: 45%
Yield of adduct: 30%
Br
62% using t-BuONa
NO2
Br
NC
Br
NC
CH3
CO2Et
CO2Li
19
20
Yield of adduct: 0%
18
Yield of adduct: 0%
Yield of adduct: 80%
Figure 1
Finally, we were able to demonstrate proof-of-principle
for our synthetic strategy by synthesizing racemic 1 as
shown below (Scheme 7). Experimental details for this
synthesis and the asymmetric reduction will be communi-
cated in due course.
1) CDI, THF, r.t.
O
2) MgCl2, (2 equiv), 50 °C
O
O
COOEt
O
MgBr
NC
(2 equiv)
NC
Br
O
COOK
78%
Pd (cat.)
OH
OH
O
72–80%
O
O
1. NaBH4, THF
r.t., 18 h
NC
NC
O
O
2. HCl (workup)
75%
OMe
O
O
NC
Ph3SiH
TiCl4
O
CH2Cl2
80%
OH
Scheme 7
Synlett 2007, No. 14, 2179–2184 © Thieme Stuttgart · New York

2184
I. N. Houpis et al.
LETTER
(8) In the absence of the nitrile Pd-catalyzed coupling of
(5) The Grignard, 8, is stable despite the b-alkoxy function but
shows remarkably low reactivity. For example, reaction with
aldehydes takes place only at 60 °C: Zhou, W.; Gumina, G.;
Chong, Y.; Wang, J.; Schinazi, R. F.; Chu, C. K. J. Med.
Chem. 2004, 47, 3399.
(6) An extensive screening protocol was conducted. Some
examples of the catalysts that were evaluated include:
Cl2Ni(PPh3)2, Cl2Ni(dppp), Cl2Ni(dppe) [dppe: 1,2-
bis(diphenylphosphino)ethane], Ni(acac)2, Ni(acac)2–
Imes+Cl– (Imes: 1,3-dimesitylimidazol-2-ylidene), Ni/C, in
THF, Bu2O, toluene, NMP, at 60–80 °C.
unprotected halobenzoic acid have been reported:
Bugamagin, N. A.; Luzikova, E. V. J. Organomet. Chem.
1997, 532, 271.
(9) Solvent screens with NMP, DMSO, toluene, trifluoro-
toluene, MTBE, gave no reaction under these reaction
conditions.
(10) Attempts to convert 8 into the corresponding Zn, boronic
acid or trifluoroborate derivatives, to evaluate its coupling
under Pd catalysis, were not successful.
(11) Variations on the ratio of 4:9 from 80:20 to 92:8 were
observed, reaction time varied between five hours and 24
hours, while the conversion varied more than 20%.
(12) Coordination of carboxylates to Pd(II) oxidative addition
intermediates has been described in Heck coupling
reactions: Dounay, A. B.; Overman, L. E. Chem. Rev. 2003,
103, 2945.
(7) The catalysts screened include: Pd(PPh)4, Pd(OAc)2–P(cy)3,
Pd(OAc)2–Pt-Bu3, PdCl2–dppf [1,1-bis(diphenylphos-
phino)ferrocene], Pd2(dba)3–P(tol)3, Pd2(dba)3–P(furyl)3.
(13) The product 4 can be isolated in ca 98% purity in 92% yield
by crystallization from EtOAc–heptane. The styrene
derivative 9 can be thus completely removed.
Synlett 2007, No. 14, 2179–2184 © Thieme Stuttgart · New York

Copyright of Synlett is the property of Georg Thieme Verlag Stuttgart and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.