Requirements for selective hydrophobic acceleration in the reduction of ketones

Article abstract of DOI:10.1021/ol0481481

(Chemical Equation Presented) Reductions of various quaternized hydrophobic β-keto amines were performed in water and in methanol using borohydride anions carrying hydrophobic groups. The most important requirement of the substrate to permit hydrophobical

Full text of DOI:10.1021/ol0481481

ORGANIC
LETTERS
2004
Vol. 6, No. 23
4331-4334
Requirements for Selective Hydrophobic
Acceleration in the Reduction of
Ketones
Mark R. Biscoe, Christopher Uyeda, and Ronald Breslow*
Department of Chemistry, Columbia UniVersity, New York, New York 10027
rb33@columbia.edu
Received September 13, 2004
ABSTRACT
Reductions of various quaternized hydrophobic
â-keto amines were performed in water and in methanol using borohydride anions carrying
hydrophobic groups. The most important requirement of the substrate to permit hydrophobically accelerated selective reductions is the ability
of the hydrophobic group of the substrate and its attached keto group to attain a coplanar relationship. Some derivatives of naturally occurring
steroid diones have also been employed as substrates to probe the mechanism and utility of these hydrophobically accelerated selective
reductions further.
The hydrophobic effect is the tendency of nonpolar surfaces
to aggregate in water in order to lower the free energy by
minimizing their interfacial hydrocarbon-water contacts.1
In our laboratory, the hydrophobic effect has been implicated
in the dramatic rate enhancements that are observed when
Diels-Alder reactions2 and benzoin condensations3 are
performed in an aqueous medium. In each of these coupling
reactions, we have proposed that hydrocarbon packing
intrinsic to the transition states results in a lower energy of
activation and thereby a rate acceleration. Building on these
observations, we have also developed a method to analyze
the geometry of transition states by determining the extent
that phenyl rings and other hydrophobic units pack in the
transition state.4
showed that in competition reactions enhanced reactivity can
be achieved for substrates bearing a hydrophobic group when
the reaction medium is changed from organic to aqueous
and a hydrophobic reagent is employed. Specifically, we
demonstrated that LiC6F5BH3 accelerates the selective reduc-
tion of a naphthyl ketone over a methyl ketone in water 40-
fold when compared with the same reaction using LiBH4 in
methanol. In that study, a general trend was established in
which increased hydrophobicity was directly related to
increased reactivity for reductions conducted in water.
To understand more completely the requirements and
mechanism of these hydrophobically accelerated reductions,
it is important to compare the relative rates of reduction
achieved when ketones and borohydrides with substituents
of different hydrophobicity and steric demand are employed.
Herein, we assess the propensity of a range of hydrophobic
groups to accelerate the reduction of ketones when the
Recently, we reported that the hydrophobic effect could
also be exploited to accelerate atom-transfer reactions.5 We
(1) (a) Breslow, R. Acc. Chem. Res. 1991, 24, 159-164. (b) Blokzijl,
W.; Engberts, J. Angew. Chem., Int. Ed. Engl. 1993, 32, 1545-1579. (c)
Otto, S.; Engberts, J. Org. Biomol. Chem. 2003, 1, 2809-2820.
(2) (a) Rideout, D.; Breslow, R. J. Am. Chem. Soc. 1980, 102, 7816-
7817. (b) Breslow, R.; Maitra, U.; Rideout, D. Tetrahedron Lett. 1983, 24,
1901-1904.
(4) (a) Breslow, R.; Groves, K.; Mayer, M. U. Org. Lett. 1999, 1, 117-
120. (b) Breslow, R.; Groves, K.; Mayer, M. U. J. Am. Chem. Soc. 2002,
124, 3622-3635. (c) Breslow, R. Acc. Chem. Res. 2004, 37, 471-478.
(5) Biscoe, M. R.; Breslow, R. J. Am. Chem. Soc. 2003, 125, 12718-
12719.
(3) Kool, E. T.; Breslow, R. J. Am. Chem. Soc. 1988, 110, 1596-1597.
10.1021/ol0481481 CCC: $27.50
© 2004 American Chemical Society
Published on Web 10/14/2004

normal aromatic ring.) The additional rate increases that
result from the “salting-out” of the hydrophobic groups using
lithium chloridesalong with the rate decrease that results
from a change to a methanolic mediumsalso strongly
implicate the hydrophobic effect.1a,3
Table 1. Ratios of Products 3/4 Formed in the Competition
Reactions of Quaternized â-Keto Amines 1 and 2 with
Substituted Borohydrides under Different Reaction
Conditionsa-d
The hydrophobic acceleration of reduction that was
observed using the 2-naphthyl derivative 1b was not observed
when the 1-naphthyl substrate 1c was employed. Instead,
the rate increase for 1c in competition reactions with 2 was
even smaller than that observed using the phenyl substrate
1a. Because the rate increase is directly related to the amount
of hydrocarbon packing permitted in the transition state, this
result indicates that there is more hydrocarbon packing with
1b than with 1c. The smaller salt effect on the reduction of
1c when compared with 1b supports this interpretation.
R2
D2O
4 M LiCl/D2O
47:53 (0.071)
CD3OD
1a
1a Ph
1a C6F5 74:26 (-0.620) 84:16 (-0.983) 40:60 (0.240)
H
44:56 (0.143)
56:44 (-0.143) 64:36 (-0.341) 35:65 (0.367)
31:69 (0.474)
1b
1b Ph
1b C6F5 91:9 (-1.371)
1c
1c
1c
1d
1d Ph
1d C6F5 20:80 (0.821)
1e
1e
1e
1f
H
53:47 (-0.071) 52:48 (-0.047) 35:65 (0.367)
67:33 (-0.420) 72:28 (-0.560) 38:62 (0.290)
Apparently the location of the carbonyl group of 1c at the
R-position on the naphthalene ring inhibits the formation of
a conjugated planar relationship between the naphthalene ring
and the carbonyl group, reflecting an unfavorable interaction
that would result with the peri hydrogen of the naphthalene
ring upon planarization. If the tendency of a hydrophobic
group to be in a coplanar conjugated conformation decreases,
the potential for removal of solvent-accessible nonpolar area
through hydrophobic overlap in the transition state will
decrease, while the steric demand that would be exerted by
the substrate hydrophobic unit on the incoming borohydride
will increase.
95:5(-1.745)
32:68 (0.447)
42:58 (0.191)
54:46 (-0.095)
19:81 (0.859)
30:70 (0.502)
H
Ph
32:68 (0.447)
40:60 (0.240)
C6F5 60:40 (-0.240) 64:36 (-0.341) 23:77 (0.716)
H
29:71 (0.531)
32:68 (0.447)
30:70 (0.502)
32:68 (0.447)
20:80 (0.821)
31:69 (0.474)
32:68 (0.447)
16:84 (0.982)
H
Ph
59:41 (-0.216) 56:44 (-0.143) 60:40 (-0.240)
56:44 (-0.143) 53:47 (-0.071) 54:46 (-0.095)
C6F5 54:46 (-0.095) 54:46 (-0.095) 53:47 (-0.071)
H
Ph
C6F5 50:50 (0.000)
H
57:43 (-0.167) 54:46 (-0.095) 50:50 (0.000)
47:53 (0.071) 49:51 (0.024) 50:50 (0.000)
51:49 (-0.024) 55:45 (-0.119)
1f
1f
No hydrophobic acceleration of reduction was observed
with the cyclohexyl substrate 1d and the benzyl substrate
1e. This lack of rate enhancement again reflects the inability
of the hydrophobic group on the borohydride to pack
efficiently on the cyclohexyl or benzyl group of the substrates
in the transition state for hydride transfer. In each case,
conformational flexibility of the hydrophobic groups
allowing the carbonyl group to lie out of the plane of the
phenyl or cyclohexyl groupswill inhibit hydrophobic pack-
ing during hydride transfer. When LiPhBH3 or LiC6F5BH3
was used as the hydride source instead of LiBH4, selectivity
for substrate 1e actually decreased. Apparently steric effects
dominate the reduction of 1e.
1g
1g Ph
1g C6F5 41:59 (0.216)
7:93 (1.533)
27:73 (0.589)
7:93 (1.533)
31:69 (0.474)
47:53 (0.071)
6:94 (1.630)
18:82 (0.898)
14:86 (1.076)
a All reactions were carried to ca. 5% conversion. b Experiments with
1a,d,f, g were conducted at a concentration of 20 mM. Experiments with
1b were conducted at a concentration of 6 mM. Experiments with 1c were
conducted at a concentration of 10 mM. c Reported ratios are within an
error of (1% in at least duplicate runs. The top six entries of this table
were previously published in ref 5. d ∆∆Gq (kcal/mol) for each reaction is
in parentheses.
reaction medium is changed from organic to aqueous. By
varying the hydrophobic part of the ketone and the borohy-
dride, we have determined the requirements for rate ac-
celeration in the hydrophobically directed reduction of
ketones.
From competition reactions performed between methyl
ketone 2 and aryl ketones 1a and 1b (Table 1), it is apparent
that the increased hydrophobicity in changing from a phenyl
(1a) to a 2-naphthyl group (1b) results in a significant rate
enhancement for carbonyl reduction when the reactions are
conducted in water using a hydrophobic reducing agent.5
When the more hydrophobic6,7 LiC6F5BH3 is employed
instead of LiPhBH3 as the reducing agent, selectivity for the
hydrophobic substrate likewise increases. (As we have
discussed,5 there may also be some assistance from quadru-
polar forces when a perfluorophenyl group packs onto a
To determine whether the propensity of an aryl ketone to
assume a conjugated coplanar conformation is an important
factor in the hydrophobic rate acceleration, we synthesized
compound 1g whose carbonyl group is held in coplanar
conjugation. Because the cyclohexanone component of 1g
could also be considered a potential source of hydrophobic
acceleration, compound 1f was also examined. Competition
reactions using 1f (Table 1) show that the carbocyclic
structure of cyclohexanone does not induce a hydrophobic
acceleration of reduction. When experiments were performed
using compound 1g, the rate increase for reduction using
(6) Sheppard, W. A.; Sheetz, C. M. Organic Fluorine Chemistry; W. A.
Benjamin, Inc.: New York, 1969.
(7) See the Supporting Information for a more efficient synthesis of
LiC6F5BH3 than previously reported in ref 5.
4332
Org. Lett., Vol. 6, No. 23, 2004

The change in ∆∆Gq by simply changing from LiBH4/
D2O to the salted-out LiC6F5BH3 is -2.154 kcal/mol, which
corresponds to a 38-fold relative rate increase. This rate
increase is comparable to that observed in the competition
reactions of 1b and 2 when the condition are changed from
LiBH4 in CD3OD to LiC6F5BH3 in 4 M LiCl/D2O.
Such a large dependence of selectivity on the reaction
medium and the borohydride substitutents was not seen when
sulfated 6-ketoepiandrosterone (5b) was tested under identi-
cal conditions of hydrophobic reduction. The perfluorophenyl
group on the borohydride is apparently not able to bind to
the rigid decalin ring system of 5b as well as it can bind to
the flat and rigid tetralone ring system of 5a. The similar
hydrophobic environments of the 6-keto and 17-keto groups
result in little change in selectivity when hydrophobic
conditions are employed. Of course both keto groups in 5b
could be experiencing hydrophobic acceleration, but to a
similar extent.
Table 2. Ratios of Products 6:7 Formed in the Partial
Reduction of Steroid 5 under Different Reaction Conditionsa-c
To determine if the rigid, saturated steroidal framework
is capable of hydrophobically accelerating the reduction of
a keto substituent, we synthesized the sulfate of 6-ketopreg-
nenolone (5c) in which the rigid 17-keto group of 5b is
replaced with a more flexible acetyl group. In 5c, the intrinsic
reactivity of the 6-keto group is much greater than that of
the 17-acetyl group, which is placed in a more sterically
demanding environment. When the reduction is performed
in CD3OD/D2O with LiBH4, the 6-keto group is reduced
preferentially over the 17-acetyl group in a ratio of 84:16.
When LiC6F5BH3 in 2 M LiCl/D2O is employed, the product
ratio becomes 95:5, which corresponds approximately to a
4-fold relative rate increase over the non-hydrophobic
conditions. This implies that the saturated steroidal frame-
work does indeed provide some hydrophobic acceleration,
although not as much as the unsaturated estrone framework
does.
R
D2O
4 M LiCl/D2O
14:86 (1.076)
1:1 CD3OD/D2O
5a
5a Ph
5a C6F5 78:22 (-0.750) 85:15 (-1.028)
H
13:87 (1.126)
60:40 (-0.240) 69:31 (-0.474)
10:90 (1.302)
32:68 (0.447)
46:54 (0.095)
39:61 (0.265)
38:62 (0.290)
39:61 (0.265)
5b
b
H
Ph
42:58 (0.191)
46:54 (0.095)
43:57 (0.167)
50:50 (0.000)
52:48 (-0.047)
5b C6F5 49:51 (0.024)
5c
5c Ph
5c C6F5 94:6 (-1.630)
H
85:15 (-1.028) 84:16d (-0.982) 84:16 (-0.982)
91:9 (-1.371)
93:7d (-1.533)
95:5d (-1.745)
85:15 (-1.028)
83:17 (-0.939)
a All reactions carried to ca. 5% conversion and conducted at a
concentration of 20 mM. b Reported ratios are within an error of (1% in
at least duplicate runs. c ∆∆Gq (kcal/mol) for each reaction is in parentheses.
d Reaction performed in 2 M LiCl/D2O.
We investigated the effect of greater structural variation
of the borohydride reagents on the observed rate increase in
reductions conducted under hydrophobic conditions (Table
3). We synthesized three borohydride reagents: the known9
sodium salt of a borohydride anion carrying both a cyano
group and a benzyl group (NaCNBnBH2), the novel lithium
salt of a borohydride anion carrying two phenyl groups
(LiPh2BH2), and our previously synthesized5 lithium salt of
a borohydride anion carrying a 2-naphthyl group (LiNaph-
BH3).
The use of NaCNBnBH2 in competitions of 2 with 1a and
1b indicates that a benzyl group has a very similar propensity
toward hydrophobic reduction as does the phenyl group
(Table 1). Thus, compared to the phenyl group of LiPhBH3,
the extra degree of rotational freedom in the benzyl group
of NaCNBnBH2 appears to have an almost negligible effect
on its ability to bind to substrates 1a and 1b. The use of
LiPh2BH2 shows that the availability of an additional phenyl
group for hydrophobic packing is advantageous despite the
additional intrinsic steric demand. It may contribute to
reagent/substrate binding, or exert a useful steric effect.
LiC6F5BH3 in D2O was significantly larger than that observed
in 1a, in which the phenyl group is not fully frozen into
coplanar conjugation. Since we have shown that the cyclo-
hexanone component of 1g cannot be responsible for the
acceleration, the accessibility of hydrophobic binding during
carbonyl attack, which results from the frozen coplanar
conformation, must be responsible for this rate enhancement.
From these results we expectedsand observedsa reversal
of chemoselectivity in the reduction of the sulfated naturally
occurring steroid 5a (Table 2) when the H2O/LiC6F5BH3
combination was employed instead of either the H2O/LiBH4
or CH3OH/LiBH4 combinations.8 When the reduction was
performed using H2O/LiBH4 or CH3OH/LiBH4, reduction of
the 6-keto group (6a) was disfavored compared to the
reduction of the 17-keto group (7a) (13:87 and 10:90,
respectively). However, when the reduction of 5a was
performed in 4 M LiCl/D2O with LiC6F5BH3, a dramatic
selectivity reversal was observed in which the reduction of
the 6-keto group (6a) was strongly favored over the reduction
of the 17-keto group (7a) in a ratio of 85:15.
(8) UV dilution experiments performed on substrates 1b and 5a indicate
that appreciable aggregation of the substrates does not occur under the
reaction conditions employed (see the Supporting Information).
(9) Charoy, L.; Valleix, A.; Toupet, L.; Le Gall, T.; van Chuong, P. P.;
Mioskowski, C. Chem. Commun. 2000, 2275-2276.
Org. Lett., Vol. 6, No. 23, 2004
4333

directly with the amount of hydrophobic surface on the
borohydride.
Table 3. Ratios of Products 3/4 Formed under Different
Reaction Conditions in the Competition Reactions of
Quaternized â-Keto Amines 1 and 2 with Borohydrides that
Bear Varying Hydrophobic Unitsa-d
The most ideal situation for hydrophobic acceleration is
one in which hydrophobic binding is intrinsic to the transition
state of the reaction (e.g., endo Diels-Alder attack2b). By
contrast, hydride transfer reactions can of course also occur
in geometries in which hydrophobic packing is not involved.
However, this study shows that by minimizing the steric
burden of attack from the side of the aryl substituent, and
by maximizing the hydrocarbon overlap when the attack does
occur, significant hydrophobic contributions to selective rate
accelerations in hydride transfer reactions can indeed be
achieved. Studies to use such hydrophobic packing in
stereoselective reductions are currently underway.
R1
1a Bn
R2
D2O
4 M LiCl/D2O
CD3OD
CN 59:41 (-0.216) 64:36 (-0.341) 37:63 (0.315)
CN 64:36 (-0.341) 67:33 (-0.420) 39:61 (0.265)
Ph 65:35 (-0.367) 73:27 (-0.589) 39:61 (0.265)
Ph 76:24 (-0.683) 81:19 (-0.859) 45:55 (0.119)
1b Bn
1a Ph
1b Ph
1a
1b
C
C
10H7
10H7
H
H
68:32 (-0.447) 74:26 (-0.620) 34:66 (0.393)
81:19 (-0.859) 85:15 (-1.028) 40:60 (0.240)
Acknowledgment. We thank the NIH and NSF for
financial support of this work. C.U. was the recipient of a
Pfizer Undergraduate Fellowship.
a All reactions carried to ca. 5% conversion. b Experiments with 1a were
conducted at a concentration of 20 mM. Experiments with 1b were
conducted at a concentration of 6 mM. c Reported ratios are within an error
of (1% in at least duplicate runs. d ∆∆Gq (kcal/mol) for each reaction is
in parentheses.
Supporting Information Available: NMR, MS, and
procedures for the synthesis of substituted borohydrides,
quaternary salts 1a-e,g, and steroids 5a-c; UV spectra for
dilution experiments with 1b and 5a; procedures for com-
Finally, upon changing from LiPhBH3 to LiC10H7BH3 in
salted water, a change in ∆∆Gq of -0.468 kcal/mol is
observed in the competition reaction of 1b and 2. Thus, the
use of LiC10H7BH3 again demonstrates that selectivity for
the hydrophobic substrate in competition reactions varies
1
petition reactions and samples of H NMR spectra from
competition experiments. This material is available free of
charge via the Internet at http://pubs.acs.org.
OL0481481
4334
Org. Lett., Vol. 6, No. 23, 2004