Observations concerning the existence and reactivity of free α-amino aldehydes as chemical intermediates: Evidence for epimerization-free adduct formation with various nucleophiles [11]

Full text of DOI:10.1021/ja000136x

3236
J. Am. Chem. Soc. 2000, 122, 3236-3237
Observations Concerning the Existence and
Reactivity of Free r-Amino Aldehydes as Chemical
Intermediates: Evidence for Epimerization-Free
Adduct Formation with Various Nucleophiles
Andrew G. Myers,* Daniel W. Kung, and Boyu Zhong
Department of Chemistry and Chemical Biology
HarVard UniVersity
Cambridge, Massachusetts 02138
ReceiVed January 11, 2000
The idea that chiral R-amino aldehydes are viable chemical
intermediates that can be generated and sustained in solution
without protective groups has not been widely considered, if at
all. In the course of the development of synthetic routes to a series
of antitumor alkaloids1 we were led to investigate fundamental
properties of chiral R-amino aldehydes. Our research has shown
that R-amino aldehydes can be prepared without protective groups,
under conditions defined herein, and that they undergo a series
of largely epimerization-free transformations, a finding of po-
tentially great practical consequence.
The literature concerning free R-amino aldehydes dates at least
to 1908 with Fischer’s description of the preparation of glycine
aldehyde in solution, by the semi-reduction of glycine ethyl ester
with sodium amalgam. His success in the procedure was
ascertained by derivatization experiments; the enhanced stability
of the proposed intermediate amino aldehyde at acidic pH and
the possibility that it might exist as the hydrate were noted.2 The
literature on the topic following this work is not large, and we
are unaware of any discussions of unprotected chiral R-amino
aldehydes as viable intermediates in solution nor of their use in
asymmetric transformations.3 The notion that R-amino aldehydes
are subject to spontaneous dimerization followed by air oxidation
to form pyrazines also dates to the beginning of the last century
and has been incorporated in at least one organic chemistry
textbook.4 Results from our own experiments led us to question
whether unprotected R-amino aldehydes might not be viable
species in solution and, moreover, whether such intermediates
might be held and trapped without epimerization of the stereogenic
“R” center. The following exploratory studies address these
questions.
Figure 1.
aqueous solution of aldehyde hydrate was found to produce
phenylalaninol in quantitative yield and >99% ee. After 90 d,
reduction produced phenylalaninol of 99.2% ee. These experi-
ments demonstrate what, in retrospect, may seem obvious:
R-amino aldehydes are autoprotective at acidic pH; the R-amino
group is protonated6 and, by virtue of the strongly electron-
withdrawing ammonium ion, the aldehyde exists completely as
its tetrahedral solvent adduct. The solvent adduct is not formed
irreversibly, however. This was shown in the case of the methanol
adducts 3, formed in quantitative yield by hydrogenolysis of 1 in
methanol (H2, Pd/C, 1.05 equiv CF3CO2H, 23 °C). The diaster-
eomeric mixture of methyl hemiacetals (ratio 1.1:1) underwent
dynamic exchange of the methoxyl groups with the solvent upon
dissolution in methanol-d3. Analysis by 1H NMR established that
both the major and minor diastereomeric hemiacetals exchanged
with the solvent, with half-lives of 15 and 4 h, respectively.
Although slow, the exchange reaction is still much faster than
enolization of the aldehyde, for after 5 d at 23 °C reduction of
the mixture of hemiacetals 3 with sodium borohydride afforded
phenylalaninol of 99.1% ee (96% yield). These experiments show
that the rates of nucleophilic addition of water, methanol, and
borohydride to the transient R-amino or R-ammonio aldehyde
intermediate are much faster than enolization under the conditions
specified. When the pH of the aqueous or methanolic solutions
of the R-amino aldehyde intermediate was raised to above ∼5,
self-condensation of the R-amino aldehyde did occur, as evidenced
by clouding of the reaction solution and the formation of 2,5-
dibenzylpyrazine, among other products. However, even under
more basic conditions it is possible to generate and trap free
R-amino aldehydes more rapidly than they epimerize, a finding
suggested by the borohydride addition experiments above and
further supported by experiments described below.
N-Carbobenzyloxy phenylalaninal (1, >99% ee), prepared in
98% yield by oxidation of the corresponding alcohol with the
Dess-Martin periodinane in dichloromethane at 23 °C,5 was
subjected to hydrogenolysis in a 1:1 mixture of D2O and dioxane-
d8 containing 1.05 equiv of trifluoroacetic acid (final pH ∼3).
The “free” R-amino aldehyde (2) was formed in nearly quantita-
1
tive yield, as determined by H NMR analysis (Figure 1). From
1
the H NMR spectrum it was evident that 2 existed entirely as
Addition of 1 equiv of potassium cyanide to the diastereomeric
mixture of methyl hemiacetals 3, derived from 2 as described
above, led to complete and clean conversion (99% yield, 1H NMR
analysis) to a 1.2:1 mixture of diastereomeric cyanohydrins 4
(final pH 8).7 The product R-amino cyanohydrins (4) did not
survive chromatography on silica gel. Their optical purity was
assessed by reduction with sodium borohydride to form phenyl-
alaninol of 99% ee. In addition to establishing the optical purity
of the cyanohydrins 4, the latter experiment demonstrated that
expulsion of cyanide and reduction of the intermediate R-amino
aldehyde proceeded without competing epimerization under the
the aldehyde hydrate. In this state compound 2 displays remark-
able stability. Even after standing in solution for 90 d at 23 °C in
the air, 2 showed no evidence of decomposition, nor any evidence
of deuterium incorporation into the R-position (1H NMR analysis).
From the latter finding it was concluded that enolization did not
occur at a detectable rate under these conditions. This conclusion
was confirmed when the addition of sodium borohydride to the
(1) Myers, A. G.; Kung, D. W. J. Am. Chem. Soc. 1999, 121, 10828-
10829.
(2) Fischer, E. Chem. Ber. 1908, 41, 1019-1023.
(3) Adams successfully prepared optically active histidinal by Fischer’s
method: Adams, E. J. Biol. Chem. 1955, 217, 317-324. See also: Bullerwell,
R. A. F.; Lawson, A. J. Chem. Soc. 1951, 3030-3032. 2-Deoxy-2-amino
sugars are formally derivatives of R-amino aldehydes but are not included as
part of this discussion.
(4) Nenitzescu, C. D. Chimie Organica; Editura Didactica: Bucharest,
Romania, 1968; p 716.
(5) Myers, A. G.; Zhong, B.; Movassaghi, M.; Kung, D. W.; Lanman, B.
A.; Kwon, S. Tetrahedron Lett. 2000, 41, 1359-1362.
(6) The pKa of the protonated amino group is estimated to be 8, using
2-deoxy-2-amino D-glucose as a reference: Blasko, A.; Bunton, C. A.; Bunel,
S.; Ibarra, C.; Moraga, E. Carbohydr. Res. 1997, 298, 163-172.
(7) This method of preparing R-amino cyanohydrin intermediates, such as
4, involving sequential hydrogenolysis, filtration to remove catalyst, and
addition of potassium cyanide, requires no purification and is generally superior
to the two-step procedure we previously employed9 using N-Fmoc-protective
groups.
10.1021/ja000136x CCC: $19.00 © 2000 American Chemical Society
Published on Web 03/21/2000

Communications to the Editor
J. Am. Chem. Soc., Vol. 122, No. 13, 2000 3237
Figure 3.
event was traced to the second step of the sequence and almost
certainly involved epimerization of the intermediate N-Fmoc phen-
ylglycinal (6) (Figure 3).10 When modification of reaction condi-
tions failed to provide any improvement in optical activity, the
possibility of accessing 8 via the free R-amino aldehyde (in effect,
inverting the ordering of the last two steps of the sequence) was
considered. Remarkably, this proved highly successful. Thus,
treatment of the cyanohydrins derived from N-Fmoc phenylgly-
cinal (7) with morpholine in N,N-dimethylformamide efficiently
cleaved the Fmoc group8 to furnish the corresponding R-amino
cyanohydrins 9. Further subjection of this mixture to morpholine
in trifluoroethanol as solvent then furnished the diastereomeric
morpholino nitriles 8 in 79% combined yield and 89-93% ee.
There is little question that this process involved the intermediacy
of free phenylglycinal (10). That this intermediate is trapped by
morpholine addition, further progressing to morpholinium ion for-
mation and trapping by cyanide, faster than any racemizing event,
is noteworthy. Further studies have shown this protocol to be opti-
mal for producing optically active morpholino nitrile adducts such
as 8 with a wide range of chiral R-amino aldehydes as substrates.
In summary, we have shown that the propensity for R-amino
aldehydes to undergo nucleophilic addition to the aldehyde
carbonyl group generally outpaces any racemization event.
Nucleophiles examined thus far include water, methanol, boro-
hydride, cyanide, morpholine, and the R-amino group of another
R-amino aldehyde (or its equivalent). We have shown that chiral
R-amino aldehydes form stable solvent adducts in mildly acidic,
polar media such as water or methanol and that adduct formation
occurs reversibly and without epimerization. In addition, we have
prepared cyclic dimers of primary R-amino cyanohydrins, com-
pounds which have not been previously prepared to our knowl-
edge and which certainly warrant further study. Most importantly,
we have demonstrated that R-amino aldehydes are viable inter-
mediates for consideration in proposed transformations (biosyn-
thetic, laboratory) involving the transfer of molecular asymmetry.
Figure 2.
basic conditions of the borohydride addition. It is perhaps
surprising that compounds 4 could be formed at all, given their
potential for self-condensation by the Strecker reaction. Somewhat
fortuitously, it was learned that methanolic solutions of cyano-
hydrins 4 (0.2 M) dimerized spontaneously upon standing. The
products of this dimerization reaction are diastereomeric 2,5-
dicyanopiperazines (5), resulting from consecutive inter- and
intramolecular Strecker condensation reactions, respectively. The
yield of cyclic dimers is typically ∼55%, following their isolation
by chromatography on Sephadex LH-20, and the proportion of
the three diastereomeric products is 8:1:1. The cyclic dimers also
cannot be chromatographed on silica gel without material loss.
Nevertheless, flash column chromatography on silica gel (12%
EtOAc-hexanes eluent) did provide a pure sample of the major
diastereomeric dimer 5a (28% yield), a cystalline solid whose
three-dimensional structure was determined by X-ray analysis
(structure shown, Figure 2). Both nitrile groups are axial and
antiperiplanar to a nitrogen lone pair, a common feature among
amino nitrile structures recorded in the Cambridge Crystal-
lographic Data Base. The “R” centers are homochiral, suggesting
that epimerization did not occur during the dimerization process.
This was confirmed by chiral HPLC analysis (Chiralcel OD
column, 2-propanol-hexanes graded eluent) using authentic
standards of dimers derived from antipodal and racemic phenyl-
alanine. In this manner, the major diastereomer (5a) was shown
to be of 96% ee. The minor diastereomeric products are tentatively
assigned on the basis of 1H NMR analysis as the two alternative
dimeric structures that are possible without epimerization of the
R-center; in each, one cyano group is epimeric with respect to
structure 5a. To our knowledge, N-unsubstituted cyclic dimers
such as 5 have not been described before and represent intriguing
compounds for further study in terms of their reactivity, biological
activity, and potential occurrence in nature.
Acknowledgment. Financial support from the NIH is gratefully
acknowledged. D.W.K. acknowledges an NSF predoctoral fellowship.
We gratefully acknowledge Dr. Ludmila Birladeanu for her considerable
assistance with the German literature and for providing ref 4.
Supporting Information Available: Reproductions of 1H NMR
spectra of compounds 2, 3, 4, 5a, 8, and crystal structure data for
compound 5a (PDF). This material is available free of charge via the
Internet at http://pubs.acs.org.
The idea that free R-amino aldehydes can serve as viable, and
perhaps even advantageous, intermediates in asymmetric synthesis
has already impacted our research in a practical way. In an effort
to broaden the scope of our published methodology to prepare
“C-protected” R-amino aldehydes we found that treatment of
N-Fmoc phenylglycinal (6, 99% ee),5 a substrate exceedingly
prone to base-catalyzed epimerization, under the reported condi-
tions [cyanohydrin formation (7) with HCN in methanol-dichlo-
romethane, amino nitrile formation with morpholine in trifluoro-
ethanol,8 and Fmoc cleavage with DBU in dichloromethane]9
produced the diastereomeric morpholino nitriles 8 in 97%
combined yield, but with complete racemization. The racemizing
JA000136X
(9) Myers, A. G.; Kung, D. W.; Zhong, B.; Movassaghi, M.; Kwon, S. J.
Am. Chem. Soc. 1999, 121, 8401-8402.
(10) The facility with which N-protected R-amino aldehydes epimerize is
well documented. For reviews of N-protected R-amino aldehydes as synthetic
intermediates, see: (a) Jurczak, J.; Golebiowski, A. Chem. ReV. 1989, 89,
149-164. (b) Fisher, L. E.; Muchowski, J. M. Org. Prep. Proced. Int. 1990,
22, 399-484. (c) Reetz, M. T. Angew. Chem., Int. Ed. Engl. 1991, 30, 1531-
1546. (d) Sardina, F. J.; Rapoport, H. Chem. ReV. 1996, 96, 1825-1872. (e)
Reetz, M. T. Chem. ReV. 1999, 99, 1121-1162. Strategies to minimize the
epimerization of N-protected R-amino aldehydes include Rapoport’s 9-(9-
phenylfluorenyl) substitution of the R-amino group, (f) Lubell, W. D.;
Rapoport, H. J. Am. Chem. Soc. 1987, 109, 236-239, and Reetz’ N,N-dibenzyl
derivatization: (g) Reetz, M. T.; Drewes, M. W.; Schmitz, A. Angew. Chem.,
Int. Ed. Engl. 1987, 26, 1141-1143.
(8) There is a marked solvent effect in the reaction of 7 with morpholine.
In the protic solvent trifluoroethanol, Strecker reaction occcurs without Fmoc
cleavage. In the dipolar aprotic solvent DMF, cleavage of the Fmoc group
occurs without competing Strecker condensation.