Chiral polyamines from reduction of polypeptides: Asymmetric pyridoxamine-mediated transaminations

Article abstract of DOI:10.1016/j.bmcl.2005.01.021

BH₃·THF can reduce polypeptides to polyamines with retention of chirality. The resulting polyamines are intriguing general platforms for asymmetric catalysis, given the diverse structures available and their relative ease of synthesis. We have constructed a number of chiral pyridoxamine catalysts based on reduced peptides. These compounds transaminate α-ketoacids with moderate to good enantioselectivity, while their peptidyl counterparts show almost no chiral induction.

Full text of DOI:10.1016/j.bmcl.2005.01.021

Bioorganic & Medicinal Chemistry Letters 15 (2005) 1351–1355
Chiral polyamines from reduction of polypeptides:
asymmetric pyridoxamine-mediated transaminations
Wenjun Zhou, Nancy Yerkes, Jason J. Chruma, Lei Liu and Ronald Breslow*
Columbia University, Department of Chemistry, 3000 Broadway, New York, NY 10027, USA
Received 2 December 2004; revised 6 January 2005; accepted 10 January 2005
Abstract—BH3ÆTHF can reduce polypeptides to polyamines with retention of chirality. The resulting polyamines are intriguing gen-
eral platforms for asymmetric catalysis, given the diverse structures available and their relative ease of synthesis. We have con-
structed a number of chiral pyridoxamine catalysts based on reduced peptides. These compounds transaminate a-ketoacids with
moderate to good enantioselectivity, while their peptidyl counterparts show almost no chiral induction.
ꢀ 2005 Elsevier Ltd. All rights reserved.
We have reported the use of polyethylenimines as cata-
lysts in transaminations by pyridoxamine derivatives.1
The ethylenediamine units have attractive properties as
acid/base catalysts, but the polymers are of course achi-
ral. If polypeptides could be reduced to polyamines
without chain cleavage and with retention of the amino
acid chirality, this would be an attractive approach to
chiral catalysts.
to synthesize and modify, and (3) their optimization
can be accomplished via library screening.4
We tried LiAlH4, Red-Al, BH3ÆMe2S, BH3/TiCl4, and
BH3ÆTHF in the reduction5 of the tripeptide 1 (Scheme
1), and found that use of BH3ÆTHF provides the best
yield in the reduction.6 The workup of the BH3ÆTHF
reduction is also the least tedious; methanol completely
1
removes boron as its volatile ester, B(OCH3)3. The H
In particular, the amine and ammonium catalytic com-
ponents of the polymer would be in a chiral environ-
ment. They could promote asymmetric proton
transfers, and also serve as metal ion ligands and as
nucleophiles. In contrast with the polymers made by
polymerization of ethylenimine, they should also have
well defined structures and lengths, reflecting those of
their parent polypeptides. Furthermore, when made
from polypeptides, the polyamines should be easy to
modify and optimize.
NMR7,8 of 2 and 20 (derived from the dimethylamino
amides of L- and D-alanine via the BH3ÆTHF reduction
and Mosher derivatization,9 Scheme 2) confirmed that
stereochemistry was retained with BH3ÆTHF.
Next, we applied the BH3ÆTHF reduction method to
peptides of various lengths. We found that the yields
of the polyamines are 80% (n = 3), 75% (n = 6), 60%
(n = 9), and 41% (n = 12), respectively (Fig. 1). Combin-
ing these observations, it is conceivable that a large
Discovery of low molecular weight chiral catalysts has
been a focus of research during the past several dec-
ades.2 This search spawned the development of many
interesting compounds that function as asymmetric cat-
alysts.3 Among these compounds, the peptide-based chi-
ral catalysts are of special interest because (1) they offer
intriguing analogies to enzymatic systems where the chi-
ral induction relies heavily on the conformational deter-
minants and functional group arrays, (2) they are easy
*
Corresponding author. Tel.: +1 212 854 2170; fax: +1 212 854
2755; e-mail: rb33@columbia.edu
Scheme 1.
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.01.021

1352
W. Zhou et al. / Bioorg. Med. Chem. Lett. 15 (2005) 1351–1355
Scheme 2.
deprotected with TBAF and activated with mesyl chlo-
ride. The resulting mesylate was displaced by the thiol
group of L-cysteine methyl ester to afford a fully pro-
tected pyridoxamine-ligated cysteine (Cys*) in 48%
overall yield from pyridoxamine. Then this pyridox-
amine-ligated cysteine unit was coupled with an N-pro-
tected amino acid (e.g., Ac-L-Phe) at the N-terminus,
followed by ester hydrolysis and amidation with a C-
protected amino acid (e.g., L-Ala-OMe) at the C-termi-
nus. Using this procedure we can synthesize various
peptides that incorporate one pyridoxamine unit.
Figure 1. Polyglycine pre-cursors for BH3ÆTHF reduction.
library of potential asymmetric catalysts can be con-
structed from chiral polyamines with diverse lengths uti-
lizing the BH3ÆTHF reduction method. The longest
reduced peptide reported previously comprised only
two amino acid units.10
An alternative and more efficient route toward the pyr-
idoxamine-ligated peptides is outlined in Scheme 5. The
disulfide-linked peptide scaffold is first constructed from
either methyl cystinateÆ2HCl for dipeptides or N-Boc
protected cystine for tripeptides. Reduction of the disul-
fide bond with NaBH4, followed by alkylation of the
resulting free thiol with 5-(bromomethyl)pyridox-
amine12 afforded the desired pyridoxamine-ligated di-
and tripeptides in 55–88% yield over two steps.
With a successful synthetic method for reduced peptides
in hand, our next target was to construct reduced pep-
tide based catalysts. In our initial efforts, we synthesized
the first pyridoxamine catalysts linked to reduced pep-
tides. Imperialiꢀs group had recently studied some
peptide-based pyridoxamine catalysts.11 Thus, we were
very interested to compare the peptide based pyridox-
amine catalysts with our reduced counterparts (Scheme
3).
The pyridoxamine-containing peptides were purified by
preparative reverse-phase HPLC, and then reduced by
the aforementioned BH3ÆTHF method (Scheme 4). MS
and NMR studies of the reduction products revealed
that the pyridoxamine moiety was not damaged by the
reduction. After another round of reverse-phase HPLC
purification, we successfully obtained reduced peptides
incorporating pyridoxamine.
Our synthesis (Scheme 4) starts with commercially avail-
able pyridoxamine. After Boc protection of the amino
group, the benzylic OH group was selectively protected
by TBDMS. The phenolic OH group was protected by
the MOM group. The benzylic silyl ether was then
While rather large enantiopure oligoamines are attain-
able via our newly developed procedure, our initial stud-
ies focused on learning the properties of relatively small
systems. We examined a tripeptide framework, Ac-Phe-
Cys*-Ala, and its reduced form, Red-Ac-Phe-Cys*-Ala.
The central Cys* indicates the location of the pyridox-
amine unit; ꢁRedꢀ means reduced. In total, five different
combinations of stereochemistry were considered:
LLL, LLD, DLL, DLD, and DDL.13,14
We found that before reduction, all the peptide-based
pyridoxamines exhibited very low enantioselectivity
(ee < 5%) in the transamination reactions with either
phenylpyruvic or pyruvic acid as substrates in water
(Table 1). On the other hand, the reduced peptide-based
pyridoxamines showed moderate ee values ranging from
5% to 54% in aqueous media. The validity of the
Scheme 3. Proposed pyridoxamine–mediated transamination of a-
ketoacids to a-amino acids.

W. Zhou et al. / Bioorg. Med. Chem. Lett. 15 (2005) 1351–1355
1353
Scheme 4. Reagents and conditions: (a) Boc2O, Et3N, 90%; (b) TBDMSCl, Im, 95%; (c) MOMCl, NaH, 75%; (d) TBAF, 86%; (e) MsCl, Et3N,
quantitative; (f) L-Cys-OEt, NaH, 87%; (g) Ac-L-Phe, HATU, i-Pr2NEt, 95%; (h) LiOH, THF, 70%; (i) L-Ala-OMe, HATU, i-Pr2NEt, 82%; (j) TFA,
95%; (k) BH3ÆTHF, 80%.
Scheme 5. Alternative route toward pyridoxamine-ligated cysteine containing di- and tripeptides.
observed enantioselectivity was confirmed by comparing
Red-Ac-Phe-Cys*-D-Ala with the corresponding mirror
image, Red-Ac-D-Phe-D-Cys*-Ala. The LLD catalyst
afforded L-phenylalanine and L-alanine in 39% and
54% ee, respectively. The enantiomeric DDL catalyst
generated the corresponding D-enantiomers of phenylal-
anine and alanine in 38% and 50% ee, respectively. We
have shown that such transaminations can occur with
many catalytic turnovers, using sacrificial amino acids,1b
but in this work we use the pyridoxamine as a reagent.
As in other such transaminations, we see generally high
chemical yields of the product amino acids.
We found that LLL provided ee values similar to those
given by LLD. Meanwhile, DLL provided ee values
close to those given by DLD, so the C-terminal amino
acid unit in the tripeptide does not have much influence
on the enantioselectivity of the whole system. Thus we
synthesized Red-Ac-Phe-Cys*, a substrate that did not
have the C-terminal amino acid unit. The ee values for

1354
W. Zhou et al. / Bioorg. Med. Chem. Lett. 15 (2005) 1351–1355
Table 1. The % ee values in the transamination reactions between the
peptide or reduced peptide based pyridoxamine and phenylpyruvic or
pyruvic acid in aqueous mediaa,b,c
Table 2. The % ee values in the transamination reactions between Red-
Ac-Phe-Cys* and various a-ketoacids in water and in methanola,b
Ketoacid
H2O (%)
CH3OH (%)
Catalyst
Phenylpyruvic
acid (%)
Pyruvic
acid (%)
Pyruvic acid
64 (L)
38 (L)
50 (L)
43 (L)
45 (L)
47 (L)
53 (L)
18 (L)
24 (L)
85 (L)
39 (L)
24 (L)
10 (L)
22 (L)
Phenylpyruvic acid
3-Methyl-2-oxobutanoic acid
4-Methyl-2-oxopentanoic acid
2-Oxopentanoic acid
a-Ketoglutaric acid
Oxolacetic acid
Red-Me2-Cys*
9 (L)
33 (L)
39 (L)
14 (L)
5 (L)
38 (D)
37 (L)
31 (L)
40 (L)
24 (L)
37 (L)
31 (L)
15 (L)
48 (L)
<5
6 (D)
52 (L)
54 (L)
19 (L)
14 (L)
50 (D)
64 (L)
18 (L)
52 (L)
37 (L)
39 (L)
45 (L)
35 (L)
52 (L)
<5
Red-Ac-Phe-Cys*-Ala
Red-Ac-Phe-Cys*-D-Ala
Red-Ac-D-Phe-Cys*-Ala
Red-Ac-D-Phe-Cys*-D-Ala
Red-Ac-D-Phe-D-Cys*-Ala
Red-Ac-Phe-Cys*
a Values are means of two or three experiments. Standard deviation is
about 1–2%.
Red-Ac-Ala-Cys*
Red-Ac-Trp-Cys*
b Transamination reaction condition: catalyst (ca. 2.5 · 10ꢀ3 M),
ketoacid (3.8 · 10ꢀ2 M), EDTA (7.7 · 10ꢀ3 M). pH = 7.5, t = 20 ꢁC.
Red-Ac-Tyr-Cys*
Red-Bz-Phe-Cys*
turally-stabilizing intra-molecular hydrophobic interac-
tions in the oligoamine reagent. Similarly, when the
transamination reactions were conducted in methanol,
the observed ee values of the corresponding amino acids
were significantly lower than those obtained under aque-
ous conditions. An exception to this phenomenon was
observed with b-branched-a-ketoacids; L-valine was ob-
tained in 85% ee from transamination of 3-methyl-2-
oxobutanoic acid with Red-Ac-Phe-Cys* in methanol,
representing an increase of 35% ee over the aqueous
conditions. Similar results were obtained from 3-
methyl-2-oxopentanoic acid (data not shown) (Table 2).
Red-i-PrCO-Phe-Cys*
Red-t-BuCO-Phe-Cys*
Red-HOCH2CO-Phe-Cys*
Ac-Phe-Cys*-Ala
Ac-Phe-Cys*-Ala
<5
<5
<5
<5
Ac-Phe-Cys*-D-Ala
Ac-D-Phe-Cys*-Ala
Ac-D-Phe-Cys*-D-Ala
Ac-D-Phe-D-Cys*-Ala
Ac-Phe-Cys*
<5
<5
<5
<5
<5
<5
<5
<5
a Values are means of two or three experiments. Standard deviation is
about 1–2%.
b Transamination reaction condition: catalyst (ca. 2.5 · 10ꢀ3 M),
ketoacid (3.8 · 10ꢀ2 M), EDTA (7.7 · 10ꢀ3 M). pH = 7.5, t = 20 ꢁC.
c The amino acid product was derivatized with o-phthalaldehyde and
N-Boc-cysteine before analysis by reverse-phase HPLC. For more
details see Ref. 1e.
In summary, we have found that BH3ÆTHF can be uti-
lized to reduce polypeptides to polyamines with reten-
tion of chirality. We propose that the resulting
polyamines are highly interesting platforms for asym-
metric catalysis, since a large diversity of chiral struc-
tures can be inexpensively synthesized in parallel.
this LL catalyst are indeed very close to those provided
by LLL and LLD. The dipeptidyl catalyst that was not
reduced, that is Ac-Phe-Cys*, showed nearly zero
enantioselectivity in the transamination.
In our first study, we successfully utilized the reduced
polypeptides to construct a number of chiral pyridox-
amine derivatives. These transaminate a-ketoacids to
the corresponding a-amino acids with moderate enantio-
selectivity, whereas their peptidyl counterparts show
almost no chiral induction.
By contrast, the N-terminal amino acid unit has a cru-
cial role in chiral induction. Change of LLL or LLD
to DLL or DLD in the Ac-Phe-Cys*-Ala series de-
creases the ee values from 39% to 5% for phenylalanine
and from 54% to 14% for alanine. Removal of both the
N- and C-terminal residues (i.e., Red-Me2-Cys*) led to
very low and poorly defined enantioselectivity.15 We
therefore synthesized a selection of reduced dipeptidyl
catalysts that differ in the N-terminal amino acid
N- and a-C-substituents. Red-Ac-Phe-Cys* achieves
better enantiomeric excesses than Red-Ac-Ala-Cys*,
Red-Ac-Trp-Cys*, or Red-Ac-Tyr-Cys*. Red-Ac-Phe-
Cys* is also better than Red-t-BuCO-Phe-Cys*, Red-
i-PrCO-Phe-Cys*, or Red-HOCH2CO-Phe-Cys*. The
highest ee value (64%) in water came from Red-Ac-
Phe-Cys* for the transamination of pyruvic acid to
L-alanine.
In subsequent publications we will report other appli-
cations of these unique polyamines, with their well de-
fined structures and chiralities. Not only are they
chirally defined, they also can carry many useful side-
chains related to those of proteins. Thus they have
wide potential applications in synthesis and catalysis,
and in biology.
References and notes
1. (a) Liu, L.; Breslow, R. Bioorg. Med. Chem. 2004, 12,
3277–3287; (b) Liu, L.; Zhou, W.; Chruma, J.; Breslow, R.
R. J. Am. Chem. Soc. 2004, 126, 8136–8137; (c) Zhou, W.;
Liu, L.; Breslow, R. Helv. Chim. Acta 2003, 86, 3560–
3567; (d) Liu, L.; Rozenman, M.; Breslow, R. J. Am.
Chem. Soc. 2002, 124, 12660–12661; (e) Liu, L.; Rozen-
man, M.; Breslow, R. Bioorg. Med. Chem. 2002, 10, 3973–
3979.
We next studied the transamination between Red-Ac-L-
Phe-L-Cys* with other a-ketoacids and in different sol-
vents. Under aqueous conditions, the highest ee values
were obtained from pyruvic acid, providing L-alanine
in 64% ee. Larger a-ketoacid substituents, especially
those containing aromatic systems (e.g., phenylpyruvic
acid) negatively influenced the enantioselectivity of the
transamination reaction, possibly by disrupting struc-
2. Noyori, R. Angew. Chem., Int. Ed. 2002, 41, 2008–2022.
3. Catalytic Asymmetric Synthesis; Ojima, I., Ed.; Wiley-
VCH: New York, 2000.

W. Zhou et al. / Bioorg. Med. Chem. Lett. 15 (2005) 1351–1355
1355
1
4. (a) For examples see: Hoveyda, A. H.; Hird, A. W.;
Kacprzynski, M. A. Chem. Commun. 2004, 1779–1785; (b)
Miller, S. J. Acc. Chem. Res. 2004, 37, 601–610; (c)
Berkessel, A. Curr. Opin. Chem. Biol. 2003, 7, 409–419; (d)
Gilbertson, S. R.; Collibee, S. E.; Agarkov, A. J. Am.
Chem. Soc. 2000, 122, 6522–6523.
5. Some reduced peptides were studied before as potential
peptidase inhibitors. See: Vanderesse, R.; Grand, V.;
Limal, D.; Vicherat, A.; Marraud, M.; Didierjean, C.;
Aubry, A. J. Am. Chem. Soc. 1998, 120, 9444–9451,
However, there is no report of such reduced peptides as
chiral catalysts.
8. H NMR (D2O, 300 MHz) for 20: 8.10, d, 1H, J = 8.1 Hz;
7.60, m, 5H; 4.12, m, 1H; 3.60, s, 3H; 2.51, m, 2H; 2.28, s,
6H; 1.10, d, 3H, J = 6.6 Hz.
9. Dale, J. A.; Dull, D. L.; Mosher, H. S. J. Org. Chem. 1969,
34, 2543–2549.
10. Hallꢀs group also demonstrated that the BH3 reduction of
peptides proceeds without any appreciable racemization.
See: Manku, S.; Laplante, C.; Kopac, D.; Chan, T.; Hall,
D. G. J. Org. Chem. 2001, 66, 874–885.
11. (a) Roy, R. S.; Imperiali, B. Tetrahedron Lett. 1996, 37,
2129–2132; (b) Shogren-Knaak, M. A.; Imperiali, B.
Bioorg. Med. Chem. 1999, 7, 1993–2002.
6. Typical procedure: 0.30 g of 1 was dissolved in 10 mL of
THF, and 20 mL of BH3 solution in THF (1 M) was added.
The mixture was heated under reflux for 2 days. The
solvent was removed and then 30 mL of MeOH was added.
The mixture was heated under reflux again for 1 day. After
the removal of solvent, the residue was dissolved in 2 mL of
CH2Cl2. After 2 mL of HBr (33% in AcOH) was added
dropwise, the product tetrahydrobromide (0.39 g,
yield = 80%) precipitated as a white solid. 1H NMR
(D2O, 300 MHz): 8.10, m, 3H; 7.60, m, 9H; 4.90, s, 2H;
4.35, s, 2H; 3.61, m, 12H. CI MS: 377.4 (tetramine M+1).
7. 1H NMR (D2O, 300 MHz) for 2: 8.05, d, 1H, J = 8.1 Hz;
7.60, m, 5H; 4.12, m, 1H; 3.50, s, 3H; 2.39, m, 2H; 2.19, s,
6H; 1.19, d, 3H, J = 6.6 Hz.
12. Breslow, R.; Hammond, M.; Lauer, M. J. Am. Chem. Soc.
1980, 102, 421–422.
13. Ac-Phe-Cys*-Ala: 1H NMR (CD3OD, 300 MHz): 8.57,
s, 1H; 7.35, m, 5H; 4.68, m, 2H; 4.55, s, 2H; 4.45, m,
1H; 4.20, m, 2H; 3.80, s, 3H; 3.30–2.90, m, 4H; 2.85, s, 3H;
2.05, s, 3H; 1.52, d, 3H, J = 7.2 Hz. CI MS: 546.1
(M+1).
14. Red-Ac-Phe-Cys*-Ala: 1H NMR (D2O, 300 MHz): 7.99,
s, 1H; 7.25, m, 5H; 4.31, s, 2H; 3.80, s, 3H; 3.65–2.60, 15H;
2.59, s, 3H; 1.31, m, 6H. CI MS: 476.3 (M+1).
15. When Red-Me2N-Cys* was reacted with 2-oxopentanoic
acid in MeOH with Zn(OAc)2, D-norvaline was obtained
in 39% ee: Zimmerman, S. C.; Czarnik, A. W.; Breslow, R.
J. Am. Chem. Soc. 1983, 105, 1694–1695.