Glutamyl-γ-boronate inhibitors of bacterial Glu-tRNAGln amidotransferase

Article abstract of DOI:10.1016/S0960-894X(01)00499-1

Analogues of glutamyl-γ-boronate (1) were synthesized as mechanism-based inhibitors of bacterial Glu-tRNA^Gln amidotransferase (Glu-AdT) and were designed to engage a putative catalytic serine nucleophile required for the glutaminase activity of

Full text of DOI:10.1016/S0960-894X(01)00499-1

Bioorganic & Medicinal Chemistry Letters 11 (2001) 2561–2564
Glutamyl-ꢀ-boronate Inhibitors of Bacterial Glu-tRNAGln
Amidotransferase
Carl P. Decicco,a,* David J. Nelson,a Ying Luo,b Li Shen,b Kurumi Y. Horiuchi,b,*
Karen M. Amsler,c Lorie A. Foster,c Susan M. Spitz,b Jayson J. Merrill,c
Christine F. Sizemore,c Kelley C. Rogers,c Robert A. Copelandb and Mark R. Harpelb
aDepartment of Medicinal Chemistry, DuPont Pharmaceuticals Company, Experimental Station, PO Box 80400, Wilmington,
DE 19880, USA
bDepartment of Chemical Enzymology, DuPont Pharmaceuticals Company, Experimental Station, PO Box 80400, Wilmington,
DE 19880, USA
cDepartment of Antimicrobials, DuPont Pharmaceuticals Company, Experimental Station, PO Box 80400, Wilmington,
DE 19880, USA
Received 2 May 2001; accepted 16 July 2001
Abstract—Analogues of glutamyl-g-boronate (1) were synthesized as mechanism-based inhibitors of bacterial Glu-tRNAGln ami-
dotransferase (Glu-AdT) and were designed to engage a putative catalytic serine nucleophile required for the glutaminase activity of
the enzyme. Although 1 provides potent enzyme inhibition, structure–activity studies revealed a narrow range of tolerated chemical
changes that maintained activity. Nonetheless, growth inhibition of organisms that require Glu-AdT by the most potent enzyme
inhibitors appears to validate mechanism-based inhibitor design of Glu-AdT as an approach to antimicrobial development. # 2001
DuPont Pharmaceuticals Company. Published by Elsevier Science Ltd. All rights reserved.
Many bacteria lack a synthetase to directly form gluta-
mine (Gln)-loaded tRNA (Gln-tRNAGln). Instead, these
organisms require a Glu-tRNAGln amidotransferase to
transamidate mis-loaded Glu-tRNAGln and produce
this critical precursor of protein synthesis.1 Although a
cognate to Glu-AdT is present in eukaryotic mitochon-
dria,2 its absence in mammalian cytoplasm and its
essentiality for survival of many pathogenic bacteria
typical transamidases which employ an active site
cysteine (thiol nucleophile) as the displacing reactant.1,4
In the absence of any structural information about the
active site of Glu-AdT, we targeted substrate-based
(Gln) derivatives that take advantage of the critical
active site serine as inhibitors of the transferase activity.
A number of serine inactivators or ‘serine traps’ have
been developed and successfully applied in the design of
serine protease inhibitors. In general the serine trap
contains an electrophilic center that is reactive toward a
hydroxyl nucleophile. Ketones with alpha activating
groups (electron withdrawing) such as esters, amides,
halogen, phosphonates, b-lactams, and more sophisti-
cated activated heterocycles have all been studied.5
Specificity for a particular serine protease can usually be
achieved by covalent attachment of a peptide sequence
which is complementary to active site recognition ele-
ments in the protease. Boronic acids have been shown
to be effective serine traps for a number of serine pro-
teases. In general they display reversible inhibition
kinetics and proceed through a serine-boronate acetal.6
In light of our limited knowledge of the active site of
Glu-AdT, we initially chose to study the boronic acid
implicate the enzyme as
chemotherapy.
a target for antibiotic
Glu-AdT catalyzes the amidolysis of l-Gln and transfer
of NH3 to transamidate Glu-tRNAGln. Amino acid
sequence alignment with related amidases and site-
directed mutagenesis studies3 suggest that an active site
serine is essential to the mechanism of Gln-dependent
transamidation and implicate the hydroxyl group of this
residue as the nucleophilic species in the amidolysis
reaction step. This observation contrasts with the more
*Corresponding authors. Tel.:+1-302-695-1821; fax:+1-302-695-
1821; e-mail: carl.p.decicco@dupontpharma.com. Tel.:+1-302-695-
2632; fax:+1-302-695-8313; e-mail: mark.r.harpel@dupontpharma.
com
0960-894X/01/$ - see front matter # 2001 DuPont Pharmaceuticals Company. Published by Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00499-1

2562
C. P. Decicco et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2561–2564
group because it is isosteric with the carboxamide
present in the Gln substrate.
analogues of interest (Table 1). Scheme 2 outlines the
preparation of the one carbon homologues starting
from the methyl or tert-butyl esters of l-allyl glycine
(Cbz). The b-alanine derivative 6 was obtained from the
Boc-b-lactam E as shown in Scheme 3.11
To test our design concept, Glu-g-B(O2R) was synthe-
sized from (S)-N-Cbz-vinylglycine methyl ester as
described previously.7 This compound displayed IC50
values of 1.6 mM and 100 nM, respectively, against the
glutaminase and transferase activities of Glu-AdT
(Table 1). Despite its zwitterionic character, 1 displayed
Table 2. Bacterial MIC determinationsa
antibacterial activity against
organisms (Table 2).
a host of relevant
Compd S. pyo.b S. pneu.b S. aur.b E. faec.b H. pyl.b E. colic
1d
2d
3d
8d
10
11
12
14
15
16
17
8
6
8
8
32
>64
>64
>64
>64
64
3
2
2
16
64
64
8
3
4
16
>64
>64
>64
>64
8
32
32
64
16
>64
>64
>64
>64
ND
32
>64
>64
>64
8
64
>64
8
We next examined analogues of the parent in search of
more active compounds. Scheme 1 describes the syn-
thetic approach used to make analogues of the lead
compound. Most compounds were synthesized and
assayed directly as pinanediol boronate esters.8 The
synthesis of the protected vinyl glycine derivative B was
accomplished in four steps from commercially available
A.9 Boron addition to the olefin of B was effected using
BH3–Me2S followed by treatment with a-pinene/pin-
anediol to yield the key protected boronate C.10 Selec-
tive manipulation of either the carboxy or amino
protecting groups enabled the synthesis of a variety of
>64
>64
>64
>64
>64
>64
>64
>64
>64
32
NDe
ND
ND
>64
ND
ND
3
>64
>64
8
>64
>64
>64
>64
aValues are given in mg/mL. Highest concentration tested was 64 mg/mL;
values >64 mg/mL indicate that no growth was observed at this level.
Compounds 4, 5, 7, 9, 13, and 18–20 exhibited MIC values >64 mg/mL
for all test organisms. Determinations were not performed for 6.
bStrains used: Streptococcus pyogenes ATCC 13709; Streptococcus
pneumoniae ATCC 49619; Staphylococcus aureus ATCC 29213, Enter-
obacter faecalis ATCC 29212; Helicobacter pylori ATCC 43629.
cIdentical values were obtained for Escherichia coli strains tolC and
SM101. Both strains harbor functional Gln-tRNAGln synthetase and
do not express Glu-AdT.
Table 1. In vitro glutaminase (GLA) and transferase (TRA) activity
dMIC >64 mg/mL for S. aureus ATTC 29213, 25923, and 43300, as
well as S. carnosus ATCC 51365, and S. epidermidis ATCC strains
12228 and 51625.
eNot determined.
Compd
R1
R2
IC50 (mM)
GLA
TRA
1ab
2a
3b
4
–OH
–OCH3
–OCH3
–OH
–NHSO2CF3
—
–H
–H
–H
–CH3
–H
—
–H
–H
–H
–H
–H
–H
1.6
1.3
0.10
0.05
0.07
1.1
4.7
6.5
10
20
20
1.5
NDc
55
5
6d
7
ND
>80e
40
–NHSO2BzNH2
–NHCH2CONHCH3
–NHSO2Bz
–NHCH3
—
8
9
40
10
11
12
13
14
15
16
17
18
19
20
21
>80
>80
>80
>80
>80
>80
>80
>80
>80
>80
>80
>80
35
>20f
ND
>20
ND
>20
>20
>20
>20
>20
>20
>20
–OtBu
–OH
–OH
–OH
–OCH3
—
–OH
–OMe
–OH
–OMe
–COCH3
–Cbz
–COCH2NH2
–Cbz
–CO2tBu
Cbz
-H
–H
Cbz
aAll compounds except 1 and 2 were prepared as pinanediol esters.
For 1 and 2, pd=(OH)2.
bBased on multiple determinations for 1 and 2, standard error of
amidase-derived IC50 is ꢀ25%; standard error in the transferase assay
is ꢀ20%.
Scheme 1. (a) CH2C(CH3)2, H2SO4, CH2Cl2, 78%; (b) LiOH, 1:1
THF/H2O 95%; (c) NMM, CH3CH2OCOCl, NaBH4 45%; (d) 1-
SeCN-2-NO2C6H4, Bu3P 71%; (e) BH3 Me2S, a-pinene, CH3COH,
(+)-pinanediol; (f) TFA 98%; (g) H2, Pd-C, AcOH, MeOH 98%; (h)
CH3COCl, CH2Cl2/DMF66% or HO 2CCH2NHBOC, NaHCO3,
HOAt, EDC 64%; (i) TFA; (j) NaHCO3, HN-R, HOAt, EDC, 30–
83%; (k) H2, Pd–C, AcOH/MeOH. pd=pinanediol. *See Table 1 for
structure of 4.
cNot determined.
.
dSee Scheme 2 for structure.
eCorresponds to <20% inhibition at single inhibitor concentrations of
ꢁ20 mM for samples run in parallel with assays for 1 or 2.
fCorresponds to <20% inhibition at a single inhibitor concentration
of 5 mM for samples run in parallel with assays for 1.

C. P. Decicco et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2561–2564
2563
values between glutaminase and transferase assays
reflect both the use of a lower relative Gln concentration
in the latter assay (0.5ÂKm vs 2ÂKm) and a mechanistic
shift from time-dependent, tight-binding inhibition
(glutaminase-only activity) to fast-binding, reversible
inhibition (Gln-dependent transferase activity), as we
have observed for 1.3
As demonstrated in Table 1, Glu-AdT is generally
intolerant of alterations to 1. With the exception of the
a-carboxymethyl derivatives (2 and 3), even modest
changes to Ca substituents cause at least a 10-fold
reduction in potency. Although a free a-carboxyl group
may not be absolutely required for inhibition, proper
orientation and electronic environment of the carbonyl
group is essential: displacement by a methylene (6),
alteration of pKa (5, 7, and 8), alkylation with a larger
group (12), or loss of the carboxyl (11) is not well tol-
erated. All but the sterically most simple modifications
of the a-amino (–NHCH3; 4) were inactive as inhibitors.
Thus, specific interactions between Glu-AdT and both
Ca substituents of Gln-like analogues contribute to
binding energy and specificity. This demonstrated bind-
ing specificity is consistent with other properties of Glu-
AdT and related Gln-utilizing hydrolases. Although the
three-dimensional structure of Glu-AdT has not been
solved, the structures of several other Gln-utilizing
hydrolases13 reveal multiple active-site interactions with
both a-amino and a-carboxyl groups of substrate. In
Scheme 2. (a) CH2C(CH3)2, H2SO4, CH2Cl2, 98% or CH3I, K2CO3,
.
acetone, 83%; (b) BH3 Me2S, a-pinene, CH3COH; (+)-pinanediol,
65%; (c) (R=Me) H2/Pd–C, HCl/MeOH, 74%; (d) (R=Ot-Bu),
TFA, 90%; (e) H2, Pd–C, HCl/MeOH, 91%.
.
Scheme 3. (a) KCN, C6H5CH2OH, DMF, 82%; (b) BH3 Me2S, a-
pinene, CH3COH, (+)-pinanediol, 59%; (c) H2, Pd–C; AcOH/MeOH;
(d) TFA, 98%.
Holding the amine group constant, we first examined
preliminary SAR around the carboxylate moiety of the
lead. Methyl ester 2 proved to be equipotent with 1, but
we could not rule out ester hydrolysis under the assay
conditions. tert-Butyl ester 12 was found to be more
than 200-fold less potent than 1. Interestingly, mono-
methylamine 4 was found to be an order of magnitude
less potent than 1. Removal of the carboxyl group
entirely to give the proteo derivative 11, and homo-
logation of one methylene unit to give the b-amino acid
6 resulted in loss of activity. Similarly, carboxamide 10
was also inactive. We surmised that the placement of the
negative charge must be important for activity and
investigated an isoelectronic organic acid equivalent.
We thus prepared acyl sulfonamides 5, 7, and 9, and
found that the most acidic derivative 5 did inhibit
transferase activity with an IC50 of 4.7 mM; the other
two in the series were active in the 10–20 mM range. The
glycine derivative 8 was also moderately active with an
IC50 of 20 mM.
addition, l-asparagine is utilized at 6-fold reduced kcat
/
Km by Glu-AdT (not shown). Extension of the back-
bone of 1 by a single carbon unit (18–21) may therefore
disrupt interactions of the boronic acid with the catalytic
serine, in addition to those of the Ca substituents.
The minimal growth-inhibition concentrations (MIC)
for a representative panel of bacteria provides an index
of whole cell antibacterial activity of these compounds
(Table 2).14 Included in this set is S. pyogenes, the nat-
ural source of the Glu-AdT utilized for in vitro enzyme
assays. All of the test organisms except E. coli require
Glu-AdT, as based on a functional or genomically-
inferred lack of Glu-tRNAGln synthetase. Hence, a spe-
cific in vivo inhibitor of Glu-AdT should impact all test
organisms except E. coli.
The most potent Glu-AdT inhibitors (1–3) inhibited
growth of most test bacteria that require Glu-AdT for
growth (except S. aureus, vide infra). Notably, these
compounds did not inhibit growth of E. coli tolC, a
pump-defective strain,15 or E. coli SM101 (lpxA2), an
outer membrane permeability mutant,16 consistent with
the absence of Glu-AdT in this organism. Compounds
with lesser in vitro efficacies showed little cellular activ-
ity except against E. coli. Likewise, 8, which exhibited
higher IC50 values than 1–3 and low MIC values for most
organisms, also inhibited E. coli, indicating the activity
may not be related to Glu-AdT inhibition. Thus, enzyme
inhibition and bacterial growth inhibition correlate in
this series, but only for inhibitors with IC50 ꢂ100 nM.
All of the amine derivatives in which the carboxyl group
was held constant resulted in inactive compounds (13–
15). One-methylene homo-Gln derivatives of 1, 3, 14,
and 16 (18–21) were also made and found to be inactive
against the enzyme.
Compounds were evaluated as in vitro inhibitors of
both the glutaminase-only (absence of ATP and tRNA
substrates) and Gln-dependent transferase activities of
Glu-AdT (Table 1).12 The rank order of inhibitor
potencies are similar between glutaminase-only and
transferase assays, as expected if the compounds inhibit
transferase activity by blocking formation of NH3 in the
glutaminase reaction. The lower comparative IC50
A notable exception is the recalcitrance of multiple
strains of S. aureus and other Staphylococcus species

2564
C. P. Decicco et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2561–2564
towards even the best inhibitors. The basis of this effect
is unclear. However, preliminary results show that 1 is a
potent inhibitor of S. aureus Glu-AdT in vitro.17
Therefore, the compounds must uniquely be exported,
lack cell penetrance, and/or be metabolically processed
by this genus. H. pylori is also less affected than antici-
pated, but qualitatively aligns with trends observed for
S. pyogenes.
K. FEBS Lett. 1970, 7, 23. (b) Lienhard, G. E.; Koehler, K. A.
Biochemistry 1971, 10, 2477. (c) Matthews, D. A.; Alder, R. A.;
Birktoft, J. J.; Freer, S. T.; Kraut, J. J. Biol. Chem. 1975, 250,
7120. (d) Palumaa, P.; Jarv, J. Biochim. Biophys. Acta 1984,
35. (e) Kettner, C. A.; Shenvi, A. B. J. Biol. Chem. 1984, 259,
15106. (f) Crompton, I. E.; Cuthbert, B. K.; Lowe, G.; Waley,
S. G. Biochem. J. 1988, 251, 453. (g) Kettner, C. A.; Bone, R.;
Agard, D. A.; Bachovchin, W. W. Biochemistry 1988, 27,
7682. (h) Weber, P. C.; Lee, S.-L.; Lewandowski, F. A.;
Schadt, M. C.; Chang, C.-H.; Kettner, C. A. Biochemistry
1995, 34, 33750.
7. (a) Preparation: Denniel, V.; Bauchat, P.; Danion, D.;
Renee, D.-B. Tetrahedron Lett. 1996, 37, 5111. (b) (S)-Amino-
6-boronohexanoic acid (ABH) was prepared as an arginase
inhibitor: Cox, J. D.; Kim, N. N.; Traish, A. M.; Christianson,
D. W. Nat. Struct. Biol. 1999, 6, 1043.
Compounds with (1–3) and without (10–12, 14, and 16)
antibacterial properties were also surveyed for mamma-
lian cellular toxicity.18 IC50 values were >100 mM for
all compounds in these assays.
In summary, replacement of the g-carboxamide of Gln
with boronic acid provides a route to potent inhibition
of both glutaminase and (due to its reliance on Gln
hydrolysis) net transferase activities of Glu-AdT.
Indeed, 1 is the most potent inhibitor of Glu-AdT yet
reported. Our results demonstrate a chemically narrow
SAR with respect to the spacing of the boronic acid and
Ca as well as simple modifications to the carboxyl and
amino groups. In the absence of information on the
three-dimensional active-site structure, further work is
restricted on this class of compounds as antibacterials.
However, we are encouraged by the apparent correla-
tion between analyses for the best in vitro enzyme inhi-
bitors, supportive of on-target in vivo effects, and the
fact that mammalian cellular toxicity is not a general-
ized limitation of this compound class. Our results
therefore validate Gln-AdT as a target and lend cre-
dence to a mechanism-based approach to the design of
in vivo inhibitors of this enzyme.
8. Pre-incubation of esterified compounds for 2 h at 23 ꢃC in
reaction buffer (50 mM HEPES, 25 mM MgCl2, 15 mM KCl,
1% DMSO, pH 7) and direct comparison of compounds 2
(unprotected) and 3 (pinanediol protected) in inhibition assays
demonstrated that deprotection is rapid and complete under
standard assay conditions.
9. Berkowitz, D.; McFadden, J.; Smith, M.; Pedersen, M. In
Peptidomimetics Protocols, 1st ed.; Kazmierski, W., Ed.;
Humana: New Jersey, 1999; Chapter 27.
10. Brown, H.; Jadhav, P.; Desai, M. J. Am. Chem. Soc. 1982,
104, 4303.
11. Hauser, F.; Ellenberger, R. Synthesis 1995, 9027.
12. Enzymatic assays were carried out with purified recombi-
nant Streptococcus pyogenes Glu-AdT overexpressed in
Escherichia coli (K. C. Rogers, et al., manuscript in prepara-
tion). Glutaminase-only assays [50 mM Gln (2ÂKm), 10 mg/mL
AdT and 0.1–20 mM inhibitor] and transferase assays [5 mM
Glu-tRNAGln (>10ÂKm), 300 mM ATP (2ÂKm), 10 mM Gln
(0.5ÂKm), 1 mg/mL Glu-AdT and 0.01–20 mM inhibitor], will
be presented elsewhere (ref 3; Horiuchi, K. Y.; Harpel, M. R.;
Shen, L.; Luo, Y.; Rogers, K. C.; Copeland, R. A. Biochem-
istry 2001, 40, 6450). In all cases, enzyme was pre-incubated
with inhibitor (including Glu-tRNAGln and ATP for transfer-
ase reactions) for 15 min prior to initiation with Gln. All
assays were carried out in duplicate.
Acknowledgements
We thank Ms. Kathy Wang and Dr. M. John Rogers
for cloning and initial preparation of purified Glu-AdT
from S. pyogenes, Dr. Percy Carter for thoughtful sug-
gestions and Dr. Andrew M. Stern for support
throughout these studies.
13. (a) Thoden, J. B.; Huang, X.; Raushel, F. M.; Holden, H.
Biochemistry 1999, 38, 16158. (b) Muchmore, C. R.; Krahn,
J. M.; Kim, J. H.; Zalkin, H.; Smith, J. L. Protein Sci. 1998, 7,
39. (c) Kim, J. H.; Krahn, J. M.; Tomchick, D. R.; Smith, J. L.;
Zalkin, H. J. Biol. Chem. 1996, 271, 15549.
14. MIC values are the lowest inhibitor concentrations that
elicit no visible bacterial growth after incubation according to
the National Committee of Clinical Laboratory Standards
guidelines [NCCLS Document M7-A4 (Vol. 17:N2), Jan.
1997]. H. pylori MIC values were determined by dilution in
Brain–Heart Infusion Broth (containing 0.25% yeast extract
and 10% horse serum) at 37 ꢃC and 10% CO2 atmosphere for
72 h from an inoculum density of 6Â105 colony forming units
per well. Strain performance was monitored with NCCLS
recommended control antibiotics; rifampin was used for H.
pylori.
References and Notes
1. (a) Curnow, A. W.; Hong, K.-W.; Yuan, R.; Kime, S.-I.;
Martins, O.; Winkler, W.; Henkin, T. M.; Soll, D. Proc. Natl.
Acad. Sci. U.S.A. 1997, 94, 11819. (b) Zalkin, H. Adv. Enzy-
mol. Rel. Areas Mol. Biol. 1993, 66, 203. (c) Gagnon, Y.;
Lacoste, L.; Champagne, N.; Lapointe, J. J. Biol. Chem. 1996,
271, 14856.
2. (a) Martin, N. C.; Rabinowitz, M. Biochemistry 1978, 17,
1628. (b) Schon, A.; Kannangara, C. G.; Gough, S.; Soll, D.
Nature 1988, 331, 187.
15. Lomovskaya, O.; Lewis, K. Proc. Natl. Acad. Sci. U.S.A.
1992, 89, 8938.
3. Harpel, M. R.; Horiuchi, K. Y.; Luo, Y.; Shen, L.; Jiang,
W.; Nelson, D. J.; Rogers, K. C.; Decicco, C. P.; Copeland,
R. A. manuscript in preparation.
16. Galloway, S.; Raetz, C. R. H. J. Biol. Chem. 1991, 265,
6394.
17. Berbaum, J. C.; Rogers, K. C. unpublished.
4. Massiere, F.; Badet–Densot, M. A. Cell Mol. Life Sci.
1998, 54, 205.
5. Edwards, P. D.; Bernstein, P. R. Med. Res. Rev. 1994, 14,
127.
6. (a) Anatov, V. K.; Ivanina, T. V.; Berezin, I. V.; Martinek,
18. (a) Simon, P.; Townsend, R. M.; Harris, R. R.; Jones,
E. A.; Jaffee, B. D. Transplant. Proc. 1993, 25, 19. (b) Bartlop,
J. A.; Owen, T. C. Bioorg. Med. Chem. Lett. 1991, 1, 611. (c)
CellTiter 96 Non-Radioactive Cell Proliferation Assay Tech-
nical Bulletin #TB112, Promega Corporation.