Design, synthesis, and binding studies of bidentate Zn-chelating peptidic inhibitors of glyoxalase-I

Article abstract of DOI:10.1016/j.bmcl.2006.12.056

The known affinity of ethyl acetoacetate (ACC) toward divalent zinc prompted us to attempt its employment as a chelating moiety in the design of glyoxalase-I inhibitors. A practical synthetic route was developed to incorporate this pharmacophore into the side chain of glutamic acid, with flexibility to allow incorporation of additional functionality at the end-stage of the synthesis. Herein, the details of this synthetic approach as well as the evaluation of the resultant β-keto ester compounds are reported.

Full text of DOI:10.1016/j.bmcl.2006.12.056

Bioorganic & Medicinal Chemistry Letters 17 (2007) 3793–3797
Design, synthesis, and binding studies of bidentate Zn-chelating
peptidic inhibitors of glyoxalase-I
Swati S. More and Robert Vince*
Center for Drug Design, Academic Health Center, and Department of Medicinal Chemistry, College of Pharmacy,
University of Minnesota, 8-123A WDH, 308 Harvard Street SE, Minneapolis, MN 55455, USA
Received 11 November 2006; revised 12 December 2006; accepted 13 December 2006
Available online 21 December 2006
Abstract—The known affinity of ethyl acetoacetate (ACC) toward divalent zinc prompted us to attempt its employment as a che-
lating moiety in the design of glyoxalase-I inhibitors. A practical synthetic route was developed to incorporate this pharmacophore
into the side chain of glutamic acid, with flexibility to allow incorporation of additional functionality at the end-stage of the syn-
thesis. Herein, the details of this synthetic approach as well as the evaluation of the resultant b-keto ester compounds are reported.
ꢀ 2007 Elsevier Ltd. All rights reserved.
Substantial experimental evidence exists for the inhibito-
ry effect of endogenous a-ketoaldehydes on cell growth.1
Methylglyoxal, one of the simplest a-ketoaldehydes, is
O
H3C
H
S
Glyoxalase-I
formed by the non-enzymatic and enzymatic fragmenta-
tion of dihydroxyacetone phosphate and glyceraldehyde
3-phosphate2,3 and the catabolism of threonine.4 But the
therapeutic use of a-ketoaldehydes as potential antican-
cer agents is limited due to their detoxification to the
corresponding aldonic acids by the glyoxalase system.
This enzyme system requires reduced glutathione be-
cause the hemithioacetal formed non-enzymatically
from reduced glutathione and a-ketoaldehyde is the sub-
strate for glyoxalase-I and is further converted to D-lac-
tic acid by glyoxalase-II (Fig. 1).5
HO
G
Hemithioacetal
-
OH
O
O
H3C
H3C
-
O
H
OH
H3C
GS
GS
O
Enediol(ate) Intermediate
Methylglyoxal
H
OH
OH
H3C
H
OH
O
H3C
The major protective role played by the glyoxalase en-
zyme system prompted us to put forward a hypothesis
that glyoxalase-I (Glx-I) inhibitors could be potential
anticancer agents.6 This was confirmed later through
the development of numerous competitive and transi-
tion state analog inhibitors of Glx-I. The initial design
of competitive inhibitors was based on taking advantage
of the deep hydrophobic pocket in the active site of Glx-
I. A large number of S-substituted alkylglutathiones7
were synthesized; our S-p-bromobenzylglutathione (1)8
being the most active amongst them. Further, the devel-
D-Lactate
S
G
G-SH
O
Glyoxalase-II
Glutathione
S-D-Lactoylglutathione
Figure 1. Glyoxalase pathway.
opment of the enediol(ate) transition state mimics, S-(N-
aryl-N-hydroxycarbamoyl)glutathiones (3), by Creigh-
ton and Murthy9,10 started a new era of more potent
tight-binding Glx-I inhibitors. Enhanced potency was
due to possible chelation of the hydroxamate
function with the divalent zinc in the active site. But
the therapeutic use of these inhibitors as anticancer
agents was precluded because of their charged nature,
which limited their penetration inside cells and lability
Keywords: Methylglyoxal; Glyoxalase; Ethyl acetoacetate; Glutathi-
one; b-Keto ester; Transition-state inhibitor.
*
Corresponding author. Tel.: +1 612 6249911; fax: +1 612 6240139;
e-mail: vince001@umn.edu
0960-894X/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.12.056

3794
S. S. More, R. Vince / Bioorg. Med. Chem. Lett. 17 (2007) 3793–3797
Zn
Zn
O
Zn
Br
O
OR
O
N
O
N
O
O
S
S
Br
Br
O
O
O
O
O
O
O
O
N
N
N
H
N
X
X
X
H
H
H
HN
COOH
HN
COOH
HN
COOH
NH2
HN
COOH
NH2
NH2
NH2
COOH
7: R = CH3
8: R = p-Br-Bn
COOH
COOH
COOH
1: X = CH2
2: X = NH
3: X = CH2
4: X = NH
5: X = CH2
6: X = NH
Figure 2. Development in the design of glyoxalase-I inhibitors.
to c-glutamyltranspeptidase-mediated cleavage that led
to their inactivation. The utilization of diester prodrugs
of our S-p-bromobenzylglutathione inhibitor by Lo and
Thornalley11 overcame the cell-penetration problem. We
recently reported upon the metabolic stability and inhib-
itory potency of a urea-isostere containing analog (4)
(Fig. 2) of these tight-binding inhibitors.12
Glu172
O
O
O
O
O
O
O
Hydrophobic
pocket
O
R
O
O
Zn
O
N
H
Br
HN COOH
HOOC
O
Glu99
NH2
7: R = Me
8: R = p-Br-Bn
Figure 3. Rationale for the design of b-ketoester-based Glx-I
inhibitors.
Recently, we developed a potent, tight-binding carbo-
analog13 of hydroxamate based transition state inhibitor
3, compound 5 (Ki = 6.17 nM), and synthesized its met-
abolically stable analog 6 (Ki = 32.6 nM) (Fig. 2) which
retained the inhibitory potency. The bioactivity of this
series of compounds proved that the presence of sulfur
is inconsequential to the ability of these glutathione ana-
logs to inhibit Glx-I. Abolishing the Glx-II substrate
similarity of these molecules would be expected to im-
prove their in vivo stability against the hydrolysis of
the thioester function, which is mediated by Glx-II.
Removal of the sulfur atom also drastically reduced
the synthetic complexity, thus facilitating further struc-
tural explorations.
targets chosen were 7, the methyl ester analog, and 8,
with the p-bromobenzyl substituent.
Synthesis of 7 is outlined in Scheme 1. EDC/HOBt-med-
iated coupling of the commercially available Boc-L-
Glu(OBn)-OH with glycine t-butyl ester (as the HCl
salt) afforded the dipeptide 10. Selective removal of the
Boc-function of dipeptide 10 in presence of the t-butyl
ester was achieved by following the protocol of Hruby16
(4 N HCl/dioxane at 0 ꢁC). Amine 11 thus obtained was
coupled to Boc-L-Glu(OH)-OtBu (19) to obtain the tri-
peptide 12 in 84% yield. Compound 19 was prepared
in four steps from L-glutamic acid. Regioselective side-
chain methyl-esterification of glutamic acid 15 was
achieved using chlorotrimethylsilane in methanol.17
Amino acid 16 thus obtained was subjected to Boc
protection using di-tert-butyl dicarbonate and t-butyl
esterification of the carboxylic acid using t-BuOH/
DCC-DMAP to obtain the orthogonally protected
glutamic acid derivative 18. Finally, hydrolysis of the
side-chain methyl ester by lithium hydroxide gave the
acid Boc-L-Glu(OH)-OtBu (19).
The hydroxamate link, however, remained a foible that
could render futile the aforementioned improvements.
Although a strong ligand for zinc, this hydroxamide link
is quite susceptible to hydrolytic breakdown. Indeed, the
poor pharmacokinetics of hydroxamate-based enzyme
inhibitors have plagued drug-development efforts in
many an instance. We therefore deemed worthwhile
the search for an alternative zinc-chelating group that
would be more stable and also be amenable to substitu-
tions for carrying out SAR at the hydrophobic substitu-
ent. The full 4s orbital and an empty 4p orbital of zinc
allow for two covalent and two dative-coordinate
bonds. As a consequence, Zn(II) salts (zinc chloride
and zinc acetate) are known to form stoichiometric com-
plexes with bidentate ligands.14 In glyoxalase-I, Glu99
and Glu172 fulfill the covalent bonds, while two water
molecules co-ordinate to this ion.15 We sought to test
the possibility of displacing the two water molecules
by an ethyl acetoacetate (ACC)-like bidentate ligand
(Fig. 3). ACC has been known to form stable complexes
with divalent zinc, some of them being robust enough to
be utilized as stabilizers in vinyl halide polymers.14 Our
choice of ACC was also based on the potential ease in
variation of the terminal alkyl substituent. The initial
Hydrogenolysis of the tripeptide 12 caused its quantita-
tive conversion to the free acid 13. Assembly of the b-ke-
to ester18 functionality was achieved by in situ
conversion of acid 13 to its imidazolide with carbonyl
diimidazole (CDI), and subsequent treatment with the
Mg2+-enolate of potassium monomethyl malonate to re-
sult in the formation of b-keto methyl ester precursor
14. Global deprotection of protected b-keto ester 14
by treatment with trifluoroacetic acid lended us the title
compound 7.19
However, the same strategy was found to be inapplica-
ble to the synthesis of b-ketobenzyl ester 8 (Scheme 2).

S. S. More, R. Vince / Bioorg. Med. Chem. Lett. 17 (2007) 3793–3797
3795
COOBn
O
COOR
O
O
COOBn
e
a
b
c
BocHN
RHN
N
BocHN
COOH
H
HN
COOR1
HN
COOtBu
d
COOtBu
10 R = Boc, R1 = tBu
11 R = H, R1 = tBu
9
12 R = Bn
13 R = H
O
O
O
O
OCH3
OCH3
COOR1
COOR2
COOH
COOH
g
f
H
H
N
N
H2N
HN
O
RHN
HN
O
15
COOtBu
O
COOH
16 R = H, R1 = CH3, R2 = H
17 R = Boc, R1 = CH3, R2 = H
18 R = Boc, R1 = CH3, R2 = tBu
19 R = Boc, R1 = H, R2 = tBu
O
h
i
COOtBu
NHBoc
COOH
NH2
TFA
j
14
7
Scheme 1. Synthesis of 7. Reagents and conditions: (a) HClÆGlyOtBu, EDC, HOBt, NMM, CH2Cl2, 91%; (b) 4N HCl/dioxane, 0 ꢁC; (c) 19, EDC,
HOBt, NMM, CH2Cl2, 84%; (d) H2, 10% Pd/C, CH3OH, 86%; (e) CDl, CH3CN then MgCl2, potassium monomethyl malonate, CH3CN, 59%; (f)
TFA/CH2Cl2 (1:1), 79%; (g) TMSCl, rt, 97%; (h) Boc2O, Na2CO3, dioxane/H2O, 91%; (i) tBuOH, DCC, DMAP, CH2Cl2, 94%; (j) LiOH, THF/H2O,
84%.
OH
Br
Br
O
O
KOH, H2O
Br
O
OH
O
OK
O
O
PhMe, reflux
4 h, 83%
O
O
O
O
21
20
Meldrum'sAcid
O
COOtBu
a or b
N
BocHN
O
O
HN COOtBu
O
O
22
O
O
13
O
O
Br
O
Br
O
O
O
O
HO
O
c
H
N
H
H
d
N
N
HN
HN
HN
O
COOtBu
COOtBu
NHBoc
O
COOH
COOtBu
O
O
COOH
COOtBu
TFA
NH2
NHBoc
23
24
8
Scheme 2. Synthesis of 8. Reagents and conditions: (a) CDl, CH3CN then MgCl2, 21, CH3CN, 46%; (b) CDl, THF then 21, iPrMgCl, THF, 40%; (c)
Meldrum’s acid, DCC, DMAP, CH2Cl2 then 4-bromobenzyl alcohol, benzene, reflux, 3 h, 42%; (d) TFA/CH2Cl2 (1:1), 84%.
We were unable to achieve a mono-hydrolysis of di(4-
bromo)-benzyl malonate to the desired potassium
mono-(4-bromo)-benzyl malonate (21) in reasonable
reaction times and yields. A faster and higher yielding
(83%) method was the acylation of p-bromobenzyl alco-
hol by Meldrum’s acid, followed by neutralization of the
product with potassium hydroxide.20 In contrast to the
facile acylation observed during the synthesis of 14,
when the Mg2+-enolate of 21 was treated with the imi-
dazolide of 13, imide 22 was isolated as the major prod-
uct. This was thought to have happened because of the
slower attack of the enolate of 21 relative to the Lewis
acid (Mg2+), catalyzed cyclization. In order to sidestep
this facile ring-closure, we first acylated Meldrum’s acid
with the DCC-activated derivative of 13, presumably
forming 23 in situ, followed by addition of p-bromoben-
zyl alcohol and refluxing the reaction in benzene21 for 3
hours to give 24 in a gratifying 42% yield. Global depro-
tection of 24 by treatment with trifluoroacetic acid gave
the second target compound 8.22

3796
S. S. More, R. Vince / Bioorg. Med. Chem. Lett. 17 (2007) 3793–3797
Figure 4. Predicted binding conformations of b-keto esters 7 (left) and 8 (right) in the glyoxalase I active site.
The b-ketoesters 7 and 8 were examined for their ability
to inhibit yeast glyoxalase-I. Initial rates of the enzymat-
ic reaction were measured by following the increase in
absorption at 240 nm in 0.05 M phosphate buffer (pH
6.6). Methylglyoxal, GSH, inhibitor, and buffer were
then added to the cell and allowed to equilibrate at
30 ꢁC for 6 min (to allow formation of the hemimercap-
tal) before addition of the enzyme. The concentrations
of the hemimercaptal were calculated using the dissocia-
tion constant, 3.1 · 10À3 M, as previously determined
for this equilibrium reaction.8 Analysis of the enzyme
kinetics showed that b-ketoesters were competitive
inhibitors of Glx-I, with Ki values of 112.44
36.23 lM for 7 and 66.88 22.84 lM for 8. The Km
for the hemimercaptal, the Glx-I substrate, was
0.5 mM, which is in close agreement with the literature
value.23
the zinc metal. However, a hydrophobic substituent to
fill the large hydrophobic pocket in the glyoxalase active
site was lacking, which could explain the relatively high
Ki.
Docking of compound 8 showed similar interactions of
the b-keto ester functionality with the zinc ion (Fig. 4).
Though it had the advantage of having a p-bromobenzyl
substituent to fill the large hydrophobic pocket, the
positioning of this substituent was found to be in colli-
sion with the surface of the enzyme. A slight increase
in the number of atoms as compared to compounds 5
and 6 is probably responsible for the displacement of
the tripeptide backbone from its ideal position. Thus,
the inability of 8 to drastically improve upon 7 may stem
in part from the inaccurate spatial orientation of its
hydrophobic substituent. In addition, a reviewer cited
the lower ability of the b-ketoester function to coordi-
nate zinc as a reason for decreased activity.
The very fact that compounds 7 and 8 were inhibitors of
Glx-I signifies that the novel b-ketoester function does
hold potential in the design of inhibitors for this enzyme.
We then began investigating lowered potency of these
inhibitors when compared to the hydroxamates. One
way would be to predict in silico the binding conforma-
tion. X-ray crystallographic data on human glyoxalase-I
complexed with S-(N-hydroxy-N-p-iodophenylcarba-
moyl)glutathione have been reported by Cameron
et al.15 This was used as the starting point for the dock-
ing of our inhibitors. Glide, as implemented in the Mae-
stro software, was used for this docking. Validation of
this docking was obtained by docking the original li-
gand, S-(N-hydroxy-N-p-iodophenylcarbamoyl)gluta-
In summary, we have identified the b-ketoester function
as a possible zinc-chelating component of Glx-I inhibi-
tors. We have also overcome the key problems in the
construction of b-ketoester analogs of glutathione. Our
final synthetic approach should be amenable to the
introduction of a large number of substituents at the
hydrophobic portion. The results of docking indicate
that the b-ketoester function does coordinate to zinc,
but the placement of the hydrophobic substituent might
be inaccurate. Subsequent design and synthesis efforts
are currently in progress to improve upon the potency
of these inhibitors through correct placement of the
hydrophobic substituent.
˚
thione, in Glx-I active site. RMSD of 0.8810 A
between docked and crystallographically determined
conformation of this ligand validated the reliability of
our docking protocol. When compound 7 was docked
in the Glx-I active site (Fig. 4), the tripeptide backbone
overlaid perfectly with the original ligand in the X-ray
crystal structure. The glutamine terminal showed the
carboxylate forming a salt bridge interaction with
Arg37, Arg122 and hydrogen bonding with carboxam-
ide protons of Asn103. The amino terminus is also in-
volved in hydrogen bonding with carboxamide oxygen
of Asn103. The carboxylate oxygens of the glycine resi-
due are within hydrogen-bonding distance of the main-
chain nitrogens of Met157 and Lys156. The Oe1 atom
of Glu99 and the Ne2 atom of His126 as well as the b-ke-
to ester functionality form the coordination sphere of
Acknowledgment
This work was supported by the funding from the Cen-
ter for Drug Design, Academic Health Center, Universi-
ty of Minnesota.
References and notes
1. Fraval, H. N.; McBrien, D. C. J. Gen. Microbiol. 1980,
117, 127.
2. Phillips, S. A.; Thornalley, P. J. Eur. J. Biochem. 1993,
212, 101.

S. S. More, R. Vince / Bioorg. Med. Chem. Lett. 17 (2007) 3793–3797
3797
3. Casazza, J. P.; Felver, M. E.; Veech, R. L. J. Biol. Chem.
1984, 259, 231.
4. Lyles, G. A.; Chalmers, J. Biochem. Pharmacol. 1992, 43,
1409.
5. (a) Thornalley, P. J. Biochem. J. 1990, 269, 1; (b) Hertz, L.;
Gerald, A. D. J. Neurosci. Res. 2004, 79, 11.
6. Vince, R.; Wadd, W. B. Biochem. Biophys. Res. Commun.
1969, 35, 593.
7. Vince, R.; Daluge, S. J. Med. Chem. 1971, 14, 35.
8. Vince, R.; Daluge, S.; Wadd, W. B. J. Med. Chem. 1971,
14, 402.
202.3 (CH2C(@O)CH2), 174.2, 174.0, 172.5, 172.2, 170.3
(C(@O)), 54.0, 53.5 (a-C:Glu), 50.2 (C(@O)CH2C(@O)),
42.2 (CH2:Gly), 39.6 (CH2C(@O)CH2), 31.8 (c-C:Glu),
29.3, 27.2 (b-C:Glu); ESI HRMS 390.1531 (M+H)+,
C15H23N3O9+H+ requires 390.1512; Reverse phase HPLC
was run on Varian Microsorb column (C18, 5 l,
4.6 · 250 mm) using two solvent systems with 0.5 mL/
min flow rate and detected at 254 nm. Solvent system 1:
0.04 M TEAB (triethylammonium bicarbonate) in water/
70% acetonitrile in water = 1/1, tR = 4.83 min, puri-
ty = 99.6%. Solvent system 2: 0.04 M TEAB in water/
9. Murthy, N. S. R. K.; Bakeris, T.; Kavarana, M. J.;
Hamilton, D. S.; Lan, Y.; Creighton, D. J. J. Med. Chem.
1994, 37, 2161.
10. Hamilton, D. S.; Creighton, D. J. J. Biol. Chem. 1992, 267,
24933.
11. Lo, T. W. C.; Thornalley, P. Biochem. Pharmacol. 1992,
44, 2357.
12. More, S. S.; Vince, R. Bioorg. Med. Chem. Lett. 2006, 16,
6039.
13. More, S. S.; Vince, R. In preparation.
14. (a) Grummitt, O.; Perz, J.; Mehaffey, J. Org. Prep. Proced.
Int. 1972, 4, 299; (b) Backus, A. C.; Wood, L. L. Fr.
1546264. 1968.
15. Cameron, A. D.; Ridderstrom, M.; Olin, B.; Kavarana,
M. J.; Creighton, D. J.; Mannervik, B. Biochemistry 1999,
38, 13480.
16. Han, G.; Tamaki, M.; Hruby, V. J. J. Pept. Res. 2001, 58,
338.
17. Ager, D. J.; Babler, S.; Erickson, R. A.; Froen, D. E.;
Kittleson, J.; Pantaleone, D. P.; Prakash, I.; Zhi, B. Org.
Process Res. Dev. 2004, 8, 72.
70% acetonitrile in water = 20–100%
tR = 5.41 min, purity = 96.27%.
20. Felder, D.; Nava, M. G.; Del, P. C. M.; Eckert, J.;
Luccisano, M.; Schall, C.; Masson, P.; Gallani, J.;
Heinrich, B.; Guillon, D.; Nierengarten, J. Helv. Chim.
Acta 2002, 85, 288.
B
linear,
21. Sun, G.; Fecko, C. J.; Nicewonger, R. B.; Webb, W. W.;
Begley, T. P. Org. Lett. 2006, 8, 681.
22. Spectral data of compound 8: 1H NMR (300 MHz,
CD3OD/CF3COOD) d 7.59 (d, J = 9.0, 2H, Ar), 7.49 (d,
J = 8.7 Hz, 2H, Ar), 5.10 (s, 2H, CH2Ph), 4.18–4.10 (m, 1H,
a-CH:Glu), 3.98 (m, 1H, a-CH:Glu), 3.89 (m, 2H,
CH2:Gly), 3.61 (s, 2H, C(@O)CH2C(@O)), 2.74–2.69 (m,
2H, c-CH2C(@O)CH2), 2.36–1.79 (m, 6H, c-CH2:Glu, b-
CH2:Glu); 13C NMR (75 MHz, CD3OD/CF3COOD) d
202.6 (CH2C(=O)CH2), 175.3, 174.5, 174.1, 172.2, 171.6
(C(@O)), 136.1, 132.2, 130.1, 123.4 (CAr), 66.6 (CH2Ph),
54.1, 53.7 (a-C:Glu), 50.6 (C(@O)CH2C(@O)), 42.1
(CH2:Gly), 39.4 (CH2C(@O)CH2), 31.8 (c-C:Glu), 28.6,
27.3 (b-C:Glu); ESI HRMS 544.0955 (M+H)+,
C21H27BrN3O9+H+ requires 544.0931; Reverse phase
HPLC was run on Varian Microsorb column (C18, 5 l,
4.6 · 250 mm) using two solvent systems with 0.5 mL/min
flow rate and detected at 254 nm. Solvent system 1: 0.04 M
TEAB (triethylammonium bicarbonate) in water/70% ace-
tonitrile in water = 1/1, tR = 5.68 min, purity = 99.9%. Sol-
vent system 2: 0.04 M TEAB in water/70% acetonitrile in
water = 20–100% B linear, tR = 14.46 min, purity = 98.9%.
23. Vince, R.; Brownell, J.; Akella, L. B. Bioorg. Med. Chem.
Lett. 1999, 9, 853.
18. Bonnaud, B.; Funes, P.; Jubault, N.; Vacher, B. Eur.
J. Org. Chem. 2005, 15, 3360.
19. Spectral data of compound 7: 1H NMR (300 MHz,
CDCl3/CF3COOD) d 4.20–4.18 (m, 1H, a-CH:Glu), 3.92
(m, 1H, a-CH:Glu), 3.89 (d, J = 6.6 Hz, 2H, CH2:Gly),
3.59 (s, 3H, OCH3), 3.34 (s, 2H, C(=O)CH2C(@O)), 2.78–
2.54 (m, 2H, c-CH2C(@O)CH2), 2.25–2.02 (m, 3H, c-
CH2:Glu, b-CH2:Glu), 1.97–1.73 (m, 3H, c-CH2:Glu, b-
CH2:Glu); 13C NMR (75 MHz, CDCl3/CF3COOD) d