Examination and Enhancement of Enantioselective Autoinduction in Cyanohydrin Formation by Cyclo[ (R)-His- (R)-Phe]

Article abstract of DOI:10.1021/jo972238k

The cyclic dipeptide cyclo[(R)-His-(R)-Phe] (1) has been known since 1981 to catalyze the enantioselective formation of cyanohydrins from aldehydes and HCN. Although 1 has proved to be very effective in the production of optically active cyanohydrins, the precise structure of its catalytically active form remains unresolved. The reaction of 3-phenoxybenzaldehyde and HCN in the presence of 1 has also been shown to exhibit enantioselective autocatalysis: the product (S)-3-phenoxymandelonitrile reacts with 1 to form a new, more enantioselective catalytic species. It is now demonstrated that this autocatalytic phenomenon is general and that, furthermore, it can be used to improve the enantioselectivity of cyanohydrin formation for several problematic substrates. Upon addition of a small (8 mol %) quantity of (S)-mandelonitrile or (S)-3-phenoxymandelonitrile to these reactions, the enantioselectivity of cyanohydrin formation was improved by as much as 20% ee. This effect has been ascribed to the formation of a complex between the added (S)-cyanohydrin and 1 that exhibits superior enantioselectivity to 1, either alone or complexed to the cyanohydrins of problematic substrates. A mathematical model has been developed, on the basis of a two-state equilibrium between 1 and a complex of 1 and cyanohydrin and used to explain the enantioselective autoinduction phenomenon in terms of five parameters: rate constants for the production of (R)- and (S)-cyanohydrin by both 1 and its cyanohydrin complex and an association constant for the formation of a cyanohydrin complex by 1. Two versions of this model, based on monomeric and dimeric 1, have been evaluated in light of the available data. Examination of the results reveals that the complexes of 1 and many of the cyanohydrins studied are highly enantioselective catalysts but that the complexes of 1 and cyanohydrins are only weakly associated; moreover, the complexation of 1 with most cyanohydrins leaves the rate of cyanohydrin formation unchanged, though both autocatalysis and enantioselective poisoning have been observed as well.

Full text of DOI:10.1021/jo972238k

4604
J . Org. Chem. 1998, 63, 4604-4610
Exa m in a tion a n d En h a n cem en t of En a n tioselective Au toin d u ction
in Cya n oh yd r in F or m a tion by Cyclo[(R)-His-(R)-P h e]
Eugene F. Kogut, J ason C. Thoen, and Mark A. Lipton*
Department of Chemistry, Purdue University, West Lafayette, Indiana 47907
Received December 10, 1997
The cyclic dipeptide cyclo[(R)-His-(R)-Phe] (1) has been known since 1981 to catalyze the
enantioselective formation of cyanohydrins from aldehydes and HCN. Although 1 has proved to
be very effective in the production of optically active cyanohydrins, the precise structure of its
catalytically active form remains unresolved. The reaction of 3-phenoxybenzaldehyde and HCN
in the presence of 1 has also been shown to exhibit enantioselective autocatalysis: the product
(S)-3-phenoxymandelonitrile reacts with 1 to form a new, more enantioselective catalytic species.
It is now demonstrated that this autocatalytic phenomenon is general and that, furthermore, it
can be used to improve the enantioselectivity of cyanohydrin formation for several problematic
substrates. Upon addition of a small (8 mol %) quantity of (S)-mandelonitrile or (S)-3-phenoxy-
mandelonitrile to these reactions, the enantioselectivity of cyanohydrin formation was improved
by as much as 20% ee. This effect has been ascribed to the formation of a complex between the
added (S)-cyanohydrin and 1 that exhibits superior enantioselectivity to 1, either alone or complexed
to the cyanohydrins of problematic substrates. A mathematical model has been developed, on the
basis of a two-state equilibrium between 1 and a complex of 1 and cyanohydrin and used to explain
the enantioselective autoinduction phenomenon in terms of five parameters: rate constants for
the production of (R)- and (S)-cyanohydrin by both 1 and its cyanohydrin complex and an association
constant for the formation of a cyanohydrin complex by 1. Two versions of this model, based on
monomeric and dimeric 1, have been evaluated in light of the available data. Examination of the
results reveals that the complexes of 1 and many of the cyanohydrins studied are highly
enantioselective catalysts but that the complexes of 1 and cyanohydrins are only weakly associated;
moreover, the complexation of 1 with most cyanohydrins leaves the rate of cyanohydrin formation
unchanged, though both autocatalysis and enantioselective poisoning have been observed as well.
In tr od u ction
to the commercial use of 1 in the synthesis of pyrethroid
insecticides.5
The formation of carbon-carbon bonds by asymmetric,
catalytic processes is a central goal of modern organic
synthesis. The last 20 years have seen an explosion of
research in this area, resulting in the discovery of many
outstanding new catalytic species for the enantioselective
formation of carbon-carbon bonds.1 Typically, such
catalytic species fall into one of two categories: organ-
otransition metal catalysts and enzymes. One intriguing
exception to this categorization is the cyclic dipeptide
catalyst cyclo[(R)-His-(R)-Phe] (1), shown by Inoue and
co-workers in 1981 to effectively catalyze the enantiose-
lective addition of HCN to benzaldehyde in benzene.2
Subsequent investigations by Inoue3 and J ackson4 es-
tablished that 1 was most effective when the aldehyde
substrate was aromatic and the solvent was toluene, in
which 1 is a heterogeneous gel. These findings have led
The exceptionally high enantioselectivity exhibited by
a small molecule catalyst containing no transition metals
has prompted a number of researchers to examine the
structure and behavior of 1 in attempts to understand
its mode of catalysis. North6 and DeVries7 have both
undertaken detailed NMR and theoretical investigations
of the structure of 1 in DMSO solution and the solid state.
Though these studies have shed light on the intrinsic
conformational preferences of 1, their relevance to the
catalytic species has been limited by the lack of enanti-
oselectivity shown by 1 in those solvents in which it is
fully soluble. Thus, at present, no comprehensive model
of the catalytic complex exists.
(1) See, for example: Yanagisawa, A.; Nakashima, H.; Ishiba, A.;
Yamamoto, H. J . Am. Chem. Soc. 1996, 118, 4723-4724.
(2) Oku, J .; Inoue, S. J . Chem. Soc., Chem. Commun. 1981, 229-
230.
(3) (a) Oku, J .; Ito, N.; Inoue, S. Makromol. Chem. 1982, 183, 579-
580. (b) Asada, S.; Kobayashi, Y.; Inoue, S. Makromol. Chem. 1985,
186, 1755-1762. (c) Kobayashi, Y.; Asada, S.; Watanabe, I.; Hayashi,
H.; Motoo, Y.; Inoue, S. Bull. Chem. Soc. J pn. 1986, 59, 893-89. (d)
Kobayashi, Y.; Hayashi, H.; Miyaji, K.; Inoue, S. Chem. Lett. 1986,
931. (e) Tanaka, K.; Mori, A.; Inoue, S. J . Org. Chem. 1990, 55, 181-
185.
In recent years, our understanding of catalysis by 1
has been further complicated by several remarkable
discoveries. First, Inoue,3e J ackson,4b and workers at
(4) (a) Matthews, B. R.; J ackson, W. R.; J ayatilake, G.; Wilshire,
C.; J acobs, H. Aust. J . Chem. 1988, 41, 1697. (b) J ackson, W. R.;
J ayatilake, G. S.; Matthews, B. R.; Wilshire, C. Aust J . Chem. 1988,
41, 203-213.
(5) Stoutamire, D. W.; Tieman, C. H. U.S. Pat. 4,560,515, 1985.
(6) North, M. Tetrahedron 1992, 48, 5509-5522.
(7) Callant, D.; Coussens, B.; v. d. Maten, T.; de Vries, J . G.; de Vries,
N. K. Tetrahedron: Asymmetry 1992, 3, 401-414.
S0022-3263(97)02238-X CCC: $15.00 © 1998 American Chemical Society
Published on Web 06/13/1998

Enantioselective Autoinduction in Cyanohydrin Formation
J . Org. Chem., Vol. 63, No. 14, 1998 4605
Shell8 revealed that the method by which 1 was obtained
in the solid state affected its catalytic activity: more
amorphous solids were more effective and enantioselec-
tive catalysts. Second, Danda observed that the gellike
state of 1 in toluene solution exhibited thixotropic
behavior and that increased enantioselectivity was ob-
served upon increasing stirring rate (and concomitantly
decreased viscosity).9a In a separate publication, Danda
noted that the conversion of 3-phenoxybenzaldehyde (2b)
to its (S)-cyanohydrin (3b) catalyzed by 1 (eq 1) exhibited
enantioselective autoinduction,9b as defined by Alberts
and Wynberg.10 By monitoring the enantiomeric excess
noted that optimal preparations of 1 are amorphous
solids with low melting points.9e,4b In our hands, lyo-
philization of aqueous solutions of 112 was found to
produce an amorphous solid with optimal catalytic
properties. Moreover, unlike precipitation from solvent
mixtures, lyophilization resulted in complete mass re-
covery of 1. Hydrocyanation reactions catalyzed by
activated 1 were carried out in accord with the procedures
of Inoue.3e Yields and enantioselectivities were deter-
mined by analysis of the crude reaction products. Yields
1
were determined by integration of the H NMR spectra,
while enantioselectivity was determined by derivatization
of cyanohydrin products, either with (-)-menthyl chlo-
roformate13 or (-)-MTPA.14 During these experiments,
it was found that the data showed considerable variation.
To lessen the impact of such variation, all data were
collected in duplicate or triplicate.
Is Au toin d u ction A Gen er a l P h en om en on ?
One limitation of the studies by Danda and Shvo is
the number of substrates examined. All the examples
of enantioselective autoinduction used 3-phenoxybenzal-
dehyde as the substrate, leaving open the question of
whether such a phenomenon is restricted to this one case
or, rather, is a general property of 1. That question has
been tested by studying the conversion of five different
aldehydes (2a -e) to their corresponding cyanohydrins
(3a -e) as a function of time. It was found that much of
the variability from run to run could be attributed to
variable induction times; this problem was circumvented
by considering the enantioselectivity of reaction as a
function of conversion. The results of these studies are
presented in Figure 1.
In all five cases, the enantioselectivity of reaction was
shown to increase with increasing conversion. Thus, the
autoinduction observed in the hydrocyanation of 3-phe-
noxybenzaldehyde (2b) was found to be quite general,
though the degree of nonlinearity in the curves is highly
variable. Nonetheless, from these graphs it would appear
that autoinduction is a general property of 1, thereby
implying that most (if not all) cyanohydrins are capable
of interacting with 1sand in so doing, form more enan-
tioselective catalysts. The nature of the interaction
between cyanohydrin and 1, and the structure of the
complex, remain to be determined.
of the product as a function of time, Danda demonstrated
that the reaction grew more enantioselective with time,
thereby implying that the product cyanohydrin interacts
with 1 to form a more enantioselective catalyst. Danda
further elaborated this observation by demonstrating
that 1, when mixed with racemic cyanohydrin, prefer-
entially complexes the S isomer (the major product) in a
0.8:1 ratio. Addition of the (S)-cyanohydrin was shown
to eliminate autoinduction in the reaction. These obser-
vations have been largely corroborated by the recent
study of Shvo, in which autoinduction was also observed
in the hydrocyantion of 2b, and two other hydroxylic
species were shown to suppress autoinduction upon their
introduction to the reaction.11
While these findings have provided valuable informa-
tion about the nature of the catalytic behavior of 1, to
date no definitive explanation for enantioselectivity ex-
ists. Indeed, the recent studies on 1 have raised as many
questions as they have answered. For instance, it is
unknown whether the autocatalytic behavior of 1 is a
general phenomenon or limited only to 3-phenoxyben-
zaldehyde. Also, it is unknown whether autoinduction
results from autocatalysis or from another source. In
addition, no structural model currently exists for the
catalytic complex. In the present study, the reactions of
a number of different aldehydessboth alone and in the
presence of external additivesshave been examined to
shed further light on the complex behavior of 1.
Th e Kin etic Con sequ en ces of Au toin d u ction
Two different explanations for these data can be
proffered. The complexes formed between 1 and the
cyanohydrins (3) could be more effective catalysts than
uncomplexed 1 (autocatalysis) or the complexes could be
made more enantioselective catalysts by suppression of
a pathway leading to the undesired isomer, thus slowing
the turnover rate (enantioselective poisoning). This
latter possibility might be likened to product inhibition
in enzymatic systems, though the fact that complexation
with 3 reinforces the preference for the major isomer
argues against such a simple explanation. Though the
distinction between the two possibilities would seem to
Resu lts a n d Discu ssion
A number of methods for activating 1 as a catalyst have
been put forth in the literature, including precipitation
from methanol3e and methanol-ether,3e,9b spray drying,8
and supercritical CO2 drying.11 It has been empirically
(8) Dong, W.; Petty, W. U.S. Pat. 4,554,102, 1985.
(9) (a) Danda, H. Synlett 1991, 263-264. (b) Danda, H.; Nishikawa,
H.; Otaka, K. J . Org. Chem. 1991, 56, 6740-6741.
(10) Alberts, A. H.; Wynberg, H. J . Am. Chem. Soc. 1989, 111, 7265-
7266.
(11) Shvo, Y.; Gal, M.; Becker, Y.; Elgavi, A. Tetrahedron: Asym-
metry 1996, 7, 911-924.
(12) We are indebted to Dr. William Nugent of DuPont for
generous donation of 1. Activation of 1 was achieved by lyophilization
of its dilute aqueous solution.
(13) Westley, J . W.; Halpern, B. J . Org. Chem. 1968, 33, 3978.
(14) Dale, J . A.; Dull, D. L.; Mosher, H. S. J . Org. Chem. 1969, 34,
2543.
a

4606 J . Org. Chem., Vol. 63, No. 14, 1998
Kogut et al.
F igu r e 1. Graphs of enantioselective autoinduction in hydrocyanation reactions catalyzed by 1 using various aldehydes: (a)
benzaldehyde (2a ); (b) 3-phenoxybenzaldehyde (2b); (c) 3-naphthaldehyde (2c); (d) furfural (2d ); and (e) pivaldehyde (2e).
be a profound one, the net result is identical: increasing
enantioselectivity as product formation increases. There-
fore, distinguishing between the two cases requires a
thorough understanding of the kinetics of the system.
Such an understanding requires in turn the develop-
ment of a general kinetic model for such an autoinduction
process. Because there is no evidence for covalent
bonding between 1 and any of the other components, the
model must view the interaction of 1 and 3 as a reversible
equilibrium. The reaction of HCN and aldehyde in
toluene must be considered irreversible, as a catalyst is
unable to substantially perturb an equilibrium. Such a
model is presented in Scheme 1. In Scheme 1, complex-
ation of 1 and 3 is viewed as a simple two-state equilib-
rium between 1 and a noncovalent complex 1•3, both of
which catalyze an irreversible reaction of HCN and
aldehyde. Although, in principle, the two enantiomeric
cyanohydrins 3R and 3S could give rise to diastereomeric
complexes with 1, the finding of Danda9b that only 3S
complexes 1 leads to the simplifying assumption of only
one complex; hence, 1•3 is shown as a single species. No
background reaction is included in this model, in accord
with the findings of Inoue.3e Thus, our kinetic model is
completely described by five substrate-dependent
parameters: rate constants for the production of each
isomer of 3 by 1 and 1•3 (k1R, k1S, k13R, k13S) and an
association constant (KA) for the complexation of 1 by 3.
Using these parameters and the initial concentration of
aldehyde ([2]0), one can express the enantioselectivity of
product formation as a function of conversion (ø) (eq 2).15

Enantioselective Autoinduction in Cyanohydrin Formation
Sch em e 1
J . Org. Chem., Vol. 63, No. 14, 1998 4607
Ta ble 1. Ra te a n d Associa tion Con sta n ts fr om An a lysis
of En a n tioselectivity Da ta Usin g Eqs 2 a n d 3
the binding of the cyanohydrins to 1, though weak, is
quite variable from cyanohydrin to cyanohydrin. The
poorest complexation is seen with 3c and 3d , when the
benzene ring of 3a has been replaced by a 3-naphthyl or
2-furyl substituent, respectively; on the other hand, the
tightest complexation may occur with 3e, with a tert-
butyl side chain derived from pivaldehyde. Thus, a
simple steric explanation is unlikely to explain the degree
of association of cyanohydrins with 1.
a
a
aldehyde
eq
ee1,a
%
ee1•3
,
%
krel
KA (M-1
)
2a
2a
2
3
37
43
100
100
1.0
1.0
9.9
5.8
2b
2b
2
3
0.0
13
100
100
1.0
1.0
9.7
7.2
2c
2c
2
3
29
60
100
98
6.4
8.5
2.9
1.9
2d
2d
2
3
17
28
100
100
1.0
1.0
2.2
1.8
Examination of the autocatalytic rate acceleration krel
demonstrates that there is no consistent deviation in rate
attendant upon formation of the complex 1•3. Rather,
in three of the cases (2a ,b,d ) there is no significant rate
change; hydrocyanation of naphthaldehyde (2c) does
exhibit autocatalytic behavior, whereas eq 2 implies that
the cyanohydrin 3e actually poisons 1! Thus, it appears
that the enhancement of enantioselectivity attendant
upon complexation of 1 is independent of the rate of
catalysis and should therefore be viewed as an indepen-
dent phenomenon.
2e
2e
2
3
4.0
10
21
20
0.56
0.95
27
7.2
aDefined in ref 16.
An additional complication has been raised by Shvo.11
Examination of the consumption of HCN by potentio-
metric titration gave a second-order rate dependence on
1, leading to the conclusion that two molecules of 1 are
involved in the transition state. Although Shvo inter-
preted this finding in terms of catalysis by a polymer,
taking these findings along with Danda’s observation of
thixotropy leads to an alternate conclusion that the
catalytic species may be a dimer of 1. Such a hypothesis
can be incorporated into our kinetic model and results
in an alternate expression for enantioselectivity as a
function of conversion (eq 3).
Nonlinear curve-fitting of these two equations to the
data produces the curves shown in Figure 1. As can be
seen, both functional forms can be fit to the data within
error limits of the experimental values. Thus, no direct
conclusions can be drawn as to the aggregation state of
1 in the catalytic complex. Nonetheless, the parameters
derived from the fits provide useful insight into the
interactions responsible for effective catalysis. Since the
enantiomeric excess of a kinetic process can be defined
in terms of the rates of production of major and minor
isomers, the data obtained from the fitted curves can be
expressed in the form of four parameters: the enanti-
oselectivity of uncomplexed 1 (ee1), that of the complex
1•3 (ee1•3), the autocatalytic rate acceleration (krel),16 and
the association constant KA. These parameters, obtained
from the rate constants that are obtained in turn from
both eqs 2 and 3, are provided in Table 1.
The enantioselectivities of the two catalytic species, 1
and 1•3, also bear scrutiny. With all the aromatic
aldehydes (2a -d ), the model for autoinduction suggests
that the complexes 1•3a -d are wholly enantioselective,
regardless of the model used; in contrast, pivaldehyde
remains a poor substrate even after formation of the
complex 1•3e. Because uncomplexed 1 exhibits no worse
enantioselectivity for hydrocyanation of pivaldehyde
than, e.g., 3-phenoxybenzaldehyde (2b), one is forced to
conclude that the problem is more likely to lie with the
complex 1•3e than with the aldehyde itself.
Im p r ovin g En a n tioselectivity th r ou gh
Exp loita tion of Au toin d u ction
One implication of the previous conclusion is that
replacement of 1•3e with a more functional complex
might substantially improve the enantioselectivity of
forming the cyanohydrin 3e. More generally, the pos-
sibility exists that the reactions of more problematic
substrates can be made more enantioselective by addition
Several interesting implications for the behavior of 1
emerge from consideration of the data in Table 1. First,
k13S + k13R (k1S - k1R)(k13R + k13S) - (k1R + k1S)(k13S - k13R
)
k13R + k
k1R + k1S
%ee(ø) ) 100
+
ln 1 +
13S[2]0KAø
(2)
(3)
(k13S + k13R)2KA[2]0ø
{
}
k13R + k13S
[
]
k13S - k13R (k1S - k1R)(k13R + k13S) - (k1R + k1S)(k13S - k13R
)
k13R + k
%ee(ø) + 100
+
tan-1
13SKA[2]0ø
3/2
{
}
k13R + k13S
k
+ k1R
1S
[
x
]
(k13S + k13R
)
k
+ k1SKA[2]0ø
1R
x

4608 J . Org. Chem., Vol. 63, No. 14, 1998
Kogut et al.
Ta ble 2. Effect of Ad d ed Cya n oh yd r in s on th e
En a n tioselectivity of Ca ta lysis by 1
Ta ble 3. Effect of Ad d ed Hyd r oxylic Com p ou n d s on th e
En a n tioselective Hyd r ocya n a tion of F u r fu r a l (2d )
entry aldehyde additive time, h yield,a
%
ee,a,b
%
∆ee,c %
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
2b
2c
2d
2e
2f
none
3a
3b
none
3a
3b
none
3a
3b
none
3a
3b
none
3a
3b
none
3a
3b
none
3a
3b
none
3a
3b
4
4
4
1.5
1.5
1.5
7
7
7
4
4
4
2.5
2.5
2.5
5
5
5
92
85
94d
87d
67
87d
89d
73
+2
+4
95
88
+15
+15
entry additive amt,mol % yield,a
%
ee,a,b
%
∆ee,c, %
93
88
92e
95
53e
81
1
2
3
4
5
6
7
8
9
10
11
12
13
none
3a
(R)-3a
3b
3d
3d
3d
3e
4
MeOH
MeOH
(S)-5
(R)-5
92d
95
78
94
87
91
66
77
83
94
85
94
86
53d
81
50
80
75
70
28
55
73
57
58
72
58
+28
+27
8
8
8
+28
-3
94
100
80
21f
24f
32f
42g
61g
67g
23g
30g
28g
26f
46f
48f
40g
44g
55g
+27
+22
+17
-25
+2
100
100
96
100
100
95
+3
+11
8
20
80
8
8
8
20
8
8
+19
+25
+20
+4
2g
2h
2i
96
98
+6
+4
+5
+19
+5
5
5
5
8
8
8
100
100
100
100
100
100
+20
+22
aThe average of two trials unless otherwise noted. bDetermined
c
by 1H NMR of (-)-menthyl chloroformate derivatives. Defined as
the difference in ee between seeded and unseeded reactions. dThe
average of six trials.
+4
+15
aBased on the average of two trials unless otherwise noted.
dramatic improvement in enantioselectivity observed
when cyanohydrins were added to the reactions of 2d
(Table 2, entries 7-9).
bDetermined by 1H NMR of (-)-menthyl chloroformate derivatives
c
unless otherwise noted. Defined as the difference in % ee between
seeded and unseeded reactions. dThe average of three trials. e The
f
average of six trials. Determined by 1H NMR of (-)-MTPA
Our results largely bear out the findings of previous
researchers (Table 3). Whereas addition of (S)-mande-
lonitrile (3a ) produced a substantial improvement in
enantioselectivity (Table 3, entry 2), addition of its
enantiomer (R)-mandelonitrile (Table 3, entry 3) pro-
duced no observable change in enantioselectivity. These
results are in accord with Danda’s findings on the
hydrocyanation of 3-phenoxybenzaldehyde (vide supra).
A comparison of four different cyanohydrins, 3a ,b,d ,e,
as seeding agents demonstrated that while the three
aromatic cyanohydrins exhibited similar improvements
in enantioselectivity (Table 3, entries 2, 4, and 5), the
aliphatic cyanohydrin 3e afforded no significant increase
(Table 3, entry 8). In addition, while addition of the
furfural-derived cyanohydrin 3d produced an improve-
ment in enantioselectivity, the observed increase was not
as great as those observed with 3a ,b. Addition of
increasing amounts of 3d (Table 3, entries 6 and 7) did
not produce a concomitant increase in enantioselectivity.
Indeed, when sufficient cyanohydrin was added, a de-
crease in enantioselectivity was noted (Table 3, entry 7).
Such a finding reinforces the conclusions of DeVries,7 who
found that solubilization of 1 by addition of mandeloni-
trile resulted in loss of enantioselectivity, perhaps by
disruption of hydrogen bonding in the catalytic complex.
The dramatic difference in effect observed between (S)-
and (R)-mandelonitrile prompted us to examine the
addition of the achiral cyanohydrin of acetone (4) to the
reaction of furfural (Table 3, entry 9). In marked contrast
to (R)-mandelonitrile, 4 proved to effectively improve the
enantioselectivity of cyanohydrin formation. Thus, it
would appear that the chirality of, e.g., 3a is not an
essential factor in its ability to improve the enantiose-
lectivity of hydrocyanation, though differences between
the enantiomers of chiral cyanohydrins are pronounced.
Another test was conducted where methanol was exam-
ined for its ability to improve enantioselectivity. Shvo
had reported9 that addition of methanol eliminated
autoinduction in the hydrocyanation of 3-phenoxyben-
derivatives. gDetermined by 19F NMR of (-)-MTPA derivatives.
of a cyanohydrin known to form a highly enantioselective
complex with 1, in this case either 3a or 3b. Such a
possibility was explored with seven different substrates
(2c-i, Table 2). In all cases, an increase in enantiose-
lectivity was observed, in some cases by over 20% ee! This
finding not only reinforces the conclusions from the
kinetic analysis but also illustrates a method for improv-
ing the enantioselectivity of problematic substrates.
The results obtained using 3-phenoxybenzaldehyde as
a substrate (Table 2, entries 1-3) also served to validate
our kinetic model. Using the kinetic parameters and
binding constant from Table 1, the effect of addition of 8
mol % product to the reaction mixture can be predicted:
application of both eqs 2 and 3 predicts an increase of
1% in the enantiomeric excess of the product formed at
completion. In the event, an increase of 4% was observed,
within experimental error of the prediction.
Having established that addition of an exogenous
cyanohydrin can markedly improve the enantioselectivity
of catalysis by 1, we next sought to examine the general-
ity of the interaction by introducing other hydroxylic
compounds into a reaction catalyzed by 1. Such an
investigation was suggested by the work of Shvo11 and
Dong,17 in which alcohols other than cyanohydrins were
shown to eliminate enantioselective autoinduction.
Whereas those groups examined the reactions of 3-phe-
noxybenzaldehyde (2b), our study used the reaction of
furfural (2d ); this change was made in response to the
(15) Derivations of eqs 2 and 3 and the solution for parameters k1R
k1S, k13R, k13, and KA are given in the Supporting Information.
,
(16) The parameters ee1, ee13, and krel are defined as the ratios (k1S
-k1R)/(k1R + k1S), (k13S - k13R)/(k13R + k13S), and (k13R + k13S)/(k1R
+
k1S), respectively; the parameter ee1 reflects the enantioselectivity
exhibited by uncomplexed 1, ee13 reflects that of the complex 1•3, and
krel quantifies the catalytic advantage (or disadvantage) conveyed to 1
upon complexation by 3.
(17) Dong, W.; Friend, P. S. U.S. Pat. 4,611,076, 1986.

Enantioselective Autoinduction in Cyanohydrin Formation
J . Org. Chem., Vol. 63, No. 14, 1998 4609
kinetic models. Whereas eq 2 predicts that pivaldehyde
cyanohydrin (3e) binds 1 tighter than any other cyano-
hydrin studied, eq 3 predicts binding comparable to that
of the aromatic cyanohydrins 3a ,b. Thus, eq 2 implies
that 3e should be capable of poisoning the reaction of,
e.g., 2b with 1 while eq 3 implies the opposite; likewise,
eq 3 implies that addition of 3a or 3b should improve
the enantioselectivity of hydrocyanation of pivaldehyde
(2e) when catalyzed by 1 while eq 2 suggests the opposite.
These contrasting predictions were examined experimen-
tally (Table 2, entries 11 and 12; Table 3, entry 8), and
in both cases, the predictions of eq 3 were borne out.
Thus, on the basis of these data, one must conclude that
the dimeric model for 1 is in better accord with experi-
ment than a monomeric model.
F igu r e 2. Structure of chiral diketopiperazine polymers in
the solid state (from ref 17).
zaldehyde. In our experiments, addition of methanol
(Table 3, entries 10 and 11) resulted in only a modest
increase in enantioselectivity. In another experiment,
the chiral secondary alcohol 1-phenyl-1-ethanol (5) was
examined in both enantiomeric forms. When the S
isomer was added (Table 3, entry 12), an improvement
in enantioselectivity comparable to those produced by
cyanohydrins was observed; conversely, when the R
isomer was added (Table 3, entry 13), it was no more
effective than methanol at improving the enantioselec-
tivity. Thus, it appears that the nitrile functionality of
cyanohydrins is not essential to their interaction with 1,
but that the chirality of an added hydroxylic species is
the most critical factor leading to enhancement of enan-
tioselectivity.
Con clu sion s
It has been demonstrated that autoinduction in the
reactions of 1 is a general phenomenon, applicable to a
variety of different aldehydes. Furthermore, such be-
havior is explainable by reversible, noncovalent complex-
ation between 1 and the product cyanohydrins 3. Ex-
amination of the individual rate constants shows that
while the complexed catalyst 1•3 is more enantioselective
than 1 alone, it is rarely a more effective catalyst as
judged by rate acceleration. Additionally, examination
of the hydrocyanation of an aliphatic aldehydes
pivaldehydessuggests that poor substrates are those
aldehydes whose cyanohydrins cannot form a highly
enantioselective complex with 1; consequently, their
enantioselectivity can be substantially improved by the
addition of a chiral, aryl cyanohydrin or secondary
alcohol. Such an improvement in enantioselectivity was
also shown to result exclusively from addition of the S
isomers; when the enantiomeric R isomers were added,
no improvement in enantioselectivity was seen.
While both monomeric and dimeric complexes of 1 are
largely consistent with the data, the ability of cyanohy-
drins 3a ,b to improve the enantioselective hydrocyana-
tion of pivaldehyde strongly supports the dimeric model.
Equally important, one need not invoke a polymer of 1
to explain these results. Owing to the complexity of
developing a kinetic model for catalysis by a polymer of
1, it is difficult to conclusively exclude that possibility;
rather, one can say that a dimer of 1 is wholly concordant
with known experiments. Further experiments are
underway to investigate in detail the structure of the
complex 1•3.
P r obin g th e Aggr ega tion Sta te of th e Ca ta lytic
Com p lex
The most intractable problem surrounding the catalytic
behavior of 1 is the aggregation state of the true catalytic
species. Although Inoue’s original model of the catalytic
complex invoked a monomer of 1, multiple inconsistencies
of this model (vide supra) with experimental findings
force us to abandon it. Questions about the aggregation
state of 1 are brought to the fore by several recent
findings. First, the thixotropic behavior of 1 in toluene9a
is consistent with mechanical disruption of a polymer,
suggesting that 1 is polymeric in solution; it is reasonable
to believe that 1 forms a hydrogen-bonded polymer in
which the s-cis lactams of 1 form eight-membered,
hydrogen-bonded rings such as are seen in the solid state
(Figure 2).18 The improvement of enantioselectivity at
high stirring rates, however, suggests that either efficient
transport is needed for high enantioselectivity or that the
most enantioselective catalyst is not a polymer but a
lower oligomer.
This latter supposition is also supported by Shvo’s
observation that hydrocyanation of 3-phenoxybenzalde-
hyde is second-order with respect to the concentration
of 1.11 Although Shvo has argued that such kinetics are
consistent with catalysis by two histidine residues on a
polymer of 1, it can also be argued that they support a
dimeric structure for the catalytic form of 1. Such a view
has several attractive features, not least of which are
structural simplicity and consistency with Prelog’s mecha-
nistic work on catalysis of cyanohydrin formation.19
Indeed, our kinetic analysis of autoinduction in the
reactions of 1 (vide supra) demonstrates that a dimer is
entirely consistent with the available data, although a
monomeric catalyst cannot be easily excluded either.
Upon close examination, however, the reactions of
pivaldehyde (2e) provide distinction between the two
Exp er im en ta l Section
Gen er a l P r oced u r es. All reactions were performed in
glassware dried by flame or in an oven, and perfomed under
a positive pressure of nitrogen gas. All aldehydes were
distilled and stored under nitrogen prior to use. Diethyl ether
was distilled from sodium benzophenone ketyl and toluene
from calcium hydride. All other reagents were used as
received without further purification. 1H NMR spectra were
recorded on a Varian Gemini 200 MHz spectrometer and 19F
NMR were recorded on a General Electric QE-300 (282 MHz).
Activa tion of cyclo[(R)-His-(R)-P h e] (1). A solution of
cyclo[(R)-His-(R)-Phe] in deionized water (250 mg in 250 mL)
was stirred at room temperature for 72 h to completely dissolve
1. The solution was then frozen and lyophilized, yielding
>99% recovery of a flocculant white solid.
P r ep a r a tion of HCN. Concentrated H2SO4 (50 mL) was
added dropwise via additon funnel to an aqueous solution of
(18) MacDonald, J . C.; Whitesides, G. M. Chem. Rev. 1994, 94, 2383.
(19) Prelog, V.; Wilhelm, M. Helv. Chem. Acta 1954, 192, 1634.

4610 J . Org. Chem., Vol. 63, No. 14, 1998
Kogut et al.
NaCN (5 M, 200 mL) in a three-neck round-bottom flask
connected, by tubing, to a gas washing flask containing
anhydrous CaCl2 and a coldfinger condenser cooled to -78 °C.
Anhydrous HCN was collected as a solid on the surface of the
condenser and once HCN evolution was complete, allowed to
to the mixture via precooled syringe and the stirring rate
increased. At completion, the reaction was quenched by
addition of 0.1 N methanolic HCl (0.250 mL). The reaction
mixture was extracted with ether (7 mL) and water (7 mL).
The organic layer was dried over Na2SO4 and concentrated in
1
slowly warm to 5 °C and collected in
a
receiving flask
vacuo to leave a crude oil that was characterized by H NMR.
containing anhydrous CaCl2 at -78 °C. The HCN (20 mL, 0.5
mol) was stored as a solid at -78 °C until needed, at which
point it was warmed to 0° and transferred via chilled syringe.
CAUTION: HYDROGEN CYANIDE IS EXTREMELY TOXIC!
EITHER INHALATION OF HCN VAPOR OR SKIN CON-
TACT WITH THE LIQUID CAN CAUSE DEATH. EX-
TREME CAUTION SHOULD BE TAKEN WHEN PREPAR-
ING, STORING, AND USING HCN.20
Deter m in a tion of Op tica l P u r ity of Cya n oh yd r in s.
Meth od A (for 3a -d ). (1R,2S,5R)-(-)-Menthyl chloroformate
(0.025 mL, 0.012 mmol) was added to a solution of crude
cyanohydrin (3, 0.057 mmol) in toluene (0.5 mL). Pyridine
(0.015 mL, 0.019 mmol) was added and the reaction stirred at
room temperature for 12 h. The mixture was concentrated in
vacuo and analyzed by 1H NMR. The diastereomeric excess
1
was determined by H NMR analysis of methine signals near
Asym m etr ic Ad d ition of Hyd r ogen Cya n id e to Ald e-
h yd es. Au toin d u ction Stu d y. Dry toluene (3 mL) was
added to solid 1 (17.1 mg, 0.06 mmol) under nitrogen and
stirred at room temperature for 45 min. To this solution was
added 2 (3.0 mmol) and the resulting mixture cooled to -25
°C. HCN (0.300 mL, 7.5 mmol) was then added to the mixture
via precooled syringe and the stirring rate increased. Aliquots
of the reaction (0.250 mL) were withdrawn at periodic intervals
and quenched in 0.1 N methanolic HCl (0.100 mL). The
quenched aliquot was extracted with ether (4 mL) and water
(4 mL). The organic layer was dried over Na2SO4 and
concentrated in vacuo to leave a crude oil that was character-
δ 6, corresponding to the methine proton R to the cyano group
of each diastereomer of the cyanohydrin menthyl carbonate.
1H NMR (CDCl3): 3a δ 6.28 (major, s), 6.25 (minor, s); 3b δ
6.21 (major, s), 6.19 (minor, s); 3c δ 6.40 (major, s), 6.38 (minor,
s); 3d δ 6.35 (major, s), 6.34 (minor, s).
Meth od B (for 3e-i). To a small sample of 3 (35 µmol)
were added CDCl3 (0.300 mL) and 4-(dimethylamino)pyridine
(2 mg, 16 mmol) followed by (R)-(-)-R-methoxy-R-(trifluorom-
ethyl)phenylacetyl chloride [(-)-MTPA-Cl, (0.008 mL, 6.2
µmol)]. The reaction was continued for 6 h under nitrogen in
an NMR tube. The diastereomeric excess of the corresponding
MTPA ester was determined by 1H or 19F NMR. 1H NMR
(CDCl3): 3e δ 5.15 (major, s), 5.11 (minor, s); 3h δ 5.31 (major,
d, J ) 5.9 Hz), 5.28 (minor, d, J ) 5.7 Hz). 19F NMR (CDCl3):
3f δ -70.54 (major, s), -70.85 (minor, s); 3g δ -70.56 (major,
s), -70.84 (minor, s); 3i δ -70.62 (major, s), -70.86 (minor,
s).
1
ized by H NMR. 3a δ 7.65-7.40 (m, 5 H), 5.54 (s, 1 H), 3.25
(br s, 1 H); 3b δ 7.62-6.99 (m, 9 H), 5.50 (s, 1 H), 3.0 (br s, 1
H); 3c δ 8.03 (s, 1 H), 7.95-7.85 (m, 3 H), 7.63-7.52 (m, 3 H),
5.72 (s, 1 H), 2.75 (br s, 1 H); 3d δ 7.48 (d, J ) 1.8 Hz, 1 H),
6.64-6.59 (m, 1 H), 6.43 (dd, J ) 1.8, 1.5 Hz, 1 H), 5.56 (s, 1
H), 3.45 (br s, 1 H); 3e δ 2.36 (s, 1 H), 1.07 (s, 9 H); 3f δ 4.28
(d, J ) 6.2 Hz, 1 H), 2.46 (br s, 1 H), 1.91-1.67 (m, 6 H), 1.38-
1.03 (m, 5 H); 3g δ 4.50 (t, J ) 7.3 Hz, 1 H), 1.93-1.69 (m, 3
H), 0.98 (d, J ) 6.3 Hz, 6 H); 3h δ 4.29 (d, J ) 5.9 Hz, 1 H),
2.45 (br s, 1 H), 2.06 (dq, J ) 6.8, 6.0 Hz, 1 H), 1.11 (d, J ) 6.8
Hz, 3 H), 1.08 (d, J ) 6.8 Hz, 3 H); 3i δ 4.48 (t, J ) 6.7 Hz, 1
H), 2.85 (br s, 1 H), 1.91-1.79 (m, 2 H), 1.55-1.30 (m, 6 H),
0.91 (t, J ) 6.6 Hz, 3 H).
Ack n ow led gm en t. This research was funded by a
grant from the National Institutes of Health (GM53091).
We gratefully acknowledge Dr. William Nugent of
DuPont Corporation for a gift of cyclo[(R)-His-(R)-Phe]
and to DuPont for an Educational Aid Grant. We also
wish to thank Prof. J ohn Grutzner for helpful criticism
and advice.
Asym m etr ic Ad d ition of Hyd r ogen Cya n id e to Ald e-
h yd es. Seed in g Exp er im en ts. Dry toluene (1 mL) was
added to 1 (5.7 mg, 0.02 mmol) and 3 (0.08 mmol) under
nitrogen and stirred at room temperature for 45 min. To this
solution was added 2 (1.0 mmol) and the resulting mixture
cooled to -25 °C. HCN (0.100 mL, 2.5 mmol) was then added
Su p p or tin g In for m a tion Ava ila ble: Derivations of ki-
netic expressions and first derivative curves (7 pages). This
material is contained in libraries on microfiche, immediately
follows this article in the microfilm version of the journal, and
can be ordered from the ACS; see any current masthead page
for ordering information.
(20) Sittig, M. Handbook of Toxic and Hazardous Chemicals and
Carcinogens; Noyes: park City, 1991; pp 905-908.
J O972238K