High-affinity inhibitors of tRNA-guanine transglycosylase replacing the function of a structural water cluster

Article abstract of DOI:10.1002/chem.200901270

The tRNA-modifying enzyme tRNA-guanine transglycosylase (TGT) is essential for the pathogenic mechanism of Shigella flexneri, the causing agent of the bacterial diarrheal disease shigellosis. Herein, the synthesis of a new class of rationally designed 6-amino-imidazo[4,5-g]quinazolin-8(7H)-one- (lim-benzoguanine) based inhibitors of TGT are reported. In order to accommodate a small hydrophobic crevice opening near the binding site of ribose-34, 2-aminoethyl substituents were introduced in position 4 of the heterocyclic scaffold. For this purpose, a synthetic sequence consisting of iodination, Suzuki cross-coupling, hydroboration, Mitsunobu reaction, and Gabriel synthesis was employed, furnishing a primary amine that served as a common intermediate for the preparation of a series of derivatives. The resulting ligands displayed very low inhibition constants, down to K_i = 2nM. Substantial additional inhibitory potency is gained by interaction of terminal lipophilic groups attached to the substituent at position 4 with the hydrophobic crevice shaped by Val45 and Leu68. At the same time, the secondary ammonium center in the substituent displaces a cluster of water molecules, solvating the catalytic residues Asp102 and Asp280, without loss in binding affinity. In addition, a synthetic intermediate with an unusual 3,6,7,8,9,10-hexahydroimidazo[4,5-g][1,3] benzodiazepine core, as confirmed by X-ray analysis, is reported.

Full text of DOI:10.1002/chem.200901270

FULL PAPER
DOI: 10.1002/chem.200901270
High-Affinity Inhibitors of tRNA-Guanine Transglycosylase Replacing the
Function of a Structural Water Cluster
Philipp C. Kohler,[a] Tina Ritschel,[b] W. Bernd Schweizer,[a] Gerhard Klebe,[b] and
FranÅois Diederich*[a]
Abstract:
The
tRNA-modifying
this purpose, a synthetic sequence con-
sisting of iodination, Suzuki cross-cou-
pling, hydroboration, Mitsunobu reac-
tion, and Gabriel synthesis was em-
terminal lipophilic groups attached to
the substituent at position 4 with the
hydrophobic crevice shaped by Val45
and Leu68. At the same time, the sec-
ondary ammonium center in the sub-
stituent displaces a cluster of water
molecules, solvating the catalytic resi-
dues Asp102 and Asp280, without loss
in binding affinity. In addition, a syn-
thetic intermediate with an unusual
enzyme tRNA–guanine transglycosy-
lase (TGT) is essential for the patho-
genic mechanism of Shigella flexneri,
the causing agent of the bacterial diar-
rheal disease shigellosis. Herein, the
synthesis of a new class of rationally
ployed, furnishing
a primary amine
that served as a common intermediate
for the preparation of a series of deriv-
atives. The resulting ligands displayed
very low inhibition constants, down to
Ki =2 nm. Substantial additional inhibi-
tory potency is gained by interaction of
designed 6-amino-imidazo
zolin-8(7H)-one- (lin-benzoguanine)
ACHTUNGTREN[NUNG 4,5-g]quina-
based inhibitors of TGT are reported.
In order to accommodate a small hy-
drophobic crevice opening near the
binding site of ribose-34, 2-aminoethyl
substituents were introduced in posi-
tion 4 of the heterocyclic scaffold. For
3,6,7,8,9,10-hexahydroimidazo
[1,3]benzodiazepine core, as confirmed
by X-ray analysis, is reported.
ACHTUNGTRENN[UNG 4,5-g]-
AHCTUNGTRENNUNG
Keywords: antibiotics · drug design ·
glycosylases · inhibitors · shigellosis
Introduction
view of the distinctively different substrates, selectivity for
the causing agent Shigella flexneri should be achievable. We
The tRNA modifying enzyme tRNA–guanine transglycosy-
lase (TGT, EC 2.4.2.29) has been recognized as a potential
have previously introduced 6-amino-imidazoACTHNGUTERN[UNG 4,5-g]quina-
zolin-8(7H)-one (lin-benzoguanine)[4] as a promising scaffold
for the design of TGT inhibitors.[5,6] We subsequently dem-
onstrated the large potential of this heterotricyclic scaffold
with a series of inhibitors featuring amino substituents in po-
sition 2, which reach into the spacious ribose-33 pocket (for
the numbering see Table 1).[7,8] Some of these ligands afford-
ed inhibitory constants Ki down to the single-digit nanomo-
lar range.[7,8] Here, we report a new series of highly active li-
gands featuring a substituent in position 4 to address the hy-
drophobic surface depression shaped by Val45 and Leu68
near the ribose-34 site without paying, as with previous li-
gands,[5,6] a large energetic penalty for replacing a consensus
water cluster solvating the two catalytic Asp side chains
(Asp102 and Asp280) of the enzyme.
target for the design of new drugs against shigellosis,[1]
a
bacterial diarrheal disease causing approximately 1.1 million
fatalities each year.[2] Eukaryotic TGT catalyzes the ex-
change of guanine-34 by queuine (Q, 7-{[(4,5-cis-dihydroxy-
2-cyclopenten-1-yl)amino]methyl}-7-deazaguanine),
while
prokaryotic TGT mediates the exchange of guanine by the
Q-precursor preQ1 (7-aminomethyl-7-deazaguanine).[3] In
[a] P. C. Kohler, Dr. W. B. Schweizer, Prof. Dr. F. Diederich
Laboratorium fꢀr Organische Chemie, ETH Zꢀrich
Hçnggerberg, HCI, 8093 Zꢀrich (Switzerland)
Fax : (+41)44-632-1109
E-mail: diederich@org.chem.ethz.ch
[b] T. Ritschel, Prof. Dr. G. Klebe
Institut fꢀr Pharmazeutische Chemie
Philipps-Universitꢁt Marburg, Marbacher Weg 6
35032 Marburg (Germany)
Results and Discussion
Fax : (+49)6421-282-8994
Replacement of structural water: The replacement of struc-
tural water near polar enzyme functionalities by ligands re-
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200901270.
Chem. Eur. J. 2009, 15, 10809 – 10817
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
10809

Table 1. Previously synthesized lin-benzoguanine inhibitors substituted in
position 4 and their respective inhibition constants for Z. mobilis TGT.[5]
highly dynamic disorder of donor and acceptor functionality
as present in an extensive water network, which is possible
in the bulk solution; therefore their release has a negative
enthalpy term. Entropically, the gain in degrees of freedom
associated with this release is favorable. In the case of a
polar binding pocket, the release of structural water is still
associated with a favorable entropy term, as the orientation
of the water molecules will be constrained by the environ-
ment, but here the desolvation of polar protein functionali-
ties has an enthalpic cost. The desolvation free energy is
close to zero if the enthalpic cost is ꢀ5 kcalmolꢁ1, about the
enthalpy of a hydrogen bond.[16,17] If the bound water pro-
vides even more stabilization to the polar pocket or contrib-
utes to a residual solvation of polar functional groups, the
overall release can become strongly unfavorable. In ligand
design, this must of course be counterbalanced by a free en-
thalpy gain provided by additional ligand-protein interac-
tions - in other words, replacement of the lost protein–water
hydrogen bond(s) by protein–ligand interactions.
Compound
R
H
Ki/nm
4100ꢂ1000[a]
1
2
3
4
1000ꢂ300
6900ꢂ1700
3700ꢂ900
[a] Competitive inhibition constant Kic (uncompetitive inhibition constant
Kiu =7900 nm). For all other compounds presented in this work, uncom-
petitive inhibition is negligible.
mains a major challenge in rational drug design. Until re-
cently, such water present in binding sites was usually ne-
glected in the design of ligands, as it was conceptually diffi-
cult to handle. Generally, the entropic benefit of releasing
trapped water into the bulk phase was deemed sufficient for
justifying displacement. When structural water molecules
are required for stabilizing and solvating hydrophilic
enzyme functionalities, however, this approach becomes
problematic.[9] Desolvation of such residues can lead to
severe enthalpic penalties that are hard to identify and non-
trivial to overcome. An early example where the importance
of fixed water molecules was recognized is that of the l-ara-
binose binding protein, which is selective for l-arabinose
over d-fucose because of unfavorable interactions of bound
water molecules with the d-fucose methyl group. The water
can be replaced by a CH2OH group of d-galactose, resulting
in strong binding.[10] In the case of HIV-1 protease, fixed
water was successfully replaced by an inhibitor in which the
water molecule was mimicked by a cyclic urea moiety.[11] For
the oligopeptide-binding protein OppA, it has been shown
that the water content of the binding site can change accord-
ing to the nature of the bound tripeptide Lys-X-Lys, where
X is variable, and that OppA is thus able to accommodate
different side chains of X of varying size and polarity.[12] Re-
cently, several attempts were made to classify crystal water
in protein structures as “replaceable” or “non-replaceable”
by computational means.[13–15]
Preliminary work: The TGT active site exhibits a hydropho-
bic patch formed by Val45 and Leu68, where the phosphate,
linking ribose-34 to ribose-35, in bound tRNA is located. In
previous work, lin-benzoguanine- (1) derived inhibitors
bearing substituents in position 4 (2–4) were synthesized in
order to gain hydrophobic interactions with this patch. The
synthesis relied on Sonogashira coupling of phenylacety-
lenes with a 4-iodobenzimidazole as precursor to the lin-
benzoguanine scaffold and subsequent reduction of the
triple bond.[5] When the inhibition constants were compared
with that of the unsubstituted scaffold 1, no significant im-
provement in binding affinity was seen, although crystal
structures of enzyme–ligand complexes showed binding of
the substituents in the ribose-34 pocket (Table 1).
In another study, we carefully analyzed the enzyme’s
active site.[19] This computation and structure-based ap-
proach revealed a well-defined water cluster located be-
tween the two catalytic aspartates Asp102 and Asp280. A
ReliBase+[18] search of all known TGT crystal structures
was performed to derive a set of consensus water posi-
tions.[19] A superposition of the crystal structure of inhibitor
2 in the active site with this consensus water cluster
(Figure 1) reveals an explanation for the relatively poor af-
finities of inhibitors 2–4: The water cluster is essential for
the solvation of the negatively charged residues of Asp102
and Asp280.[6,19] Upon binding of an appropriately substitut-
ed inhibitor, the substituent attached to address the hydro-
phobic surface depression to accommodate ribose-34 displa-
ces the water molecules to varying extent, leaving the aspar-
tates “naked”. This results in Coulomb repulsion of the two
most likely negatively charged residues as well as unfavora-
From a thermodynamic point of view, the binding free en-
thalpy associated with binding of a ligand into a water-con-
taining pocket can be split into an enthalpic and an entropic
part, both of which have contributions originating from the
ligand, the protein, and from the released water.[9] We can
distinguish the cases of apolar and polar pockets. In the
former case, the contribution from the water release is both
enthalpically and entropically favorable. Inside the pocket,
some water molecules are unable to form four virtually un-
perturbed hydrogen bonds, experiencing at the same time a
ꢁ
ble CO2 ···CH2 interactions with the apolar substituent,
which cause the side chain to distribute over two conforma-
tions. The concomitant enthalpic cost is only partially com-
pensated by the favorable interactions with the ribose-34 hy-
drophobic surface and the entropic benefit[16] resulting from
the release of water molecules into the bulk solvent. Thus,
10810
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 10809 – 10817

Inhibitors of tRNA-Guanine Transglycosylase
FULL PAPER
Figure 2. Predicted binding mode of the designed inhibitor 5 (green) in
the TGT active site (black). The protonated amino linker undergoes a hy-
drogen bond to Asp280, while the hydrophobic cyclohexyl ring occupies
the hydrophobic surface formed by Val45 and Leu68.
Figure 1. Crystal structure of inhibitor 2 (yellow: “up” conformer; green:
“down” conformer) in the TGT active site (grey), overlaid with the con-
sensus water cluster (red) located between Asp102 and Asp280. Dashed
lines represent hydrogen bonds to the “down” conformer. The residues
Val45 and Leu68 form a hydrophobic surface (shown) that is occupied by
the phenyl substituent of 2. The phenethyl moiety of 2 crosses the water
cluster, thereby expelling the water molecules from the active site. Atom
distances in ꢃ. PDB code 1Y5V.[5]
ethanediyl linker.[5] N-Alkynylamides (ynamides) can be
prepared from the corresponding secondary amides and bro-
moacetylenes using a copper(II) catalyzed cross-coupling re-
action.[21] Attempts were made to couple such alkynes to an
iodinated benzimidazole and subsequently hydrogenate the
triple bond, but this strategy was found to be impractical
and only seemed to work with alkynylsulfonamides, which
are hard to hydrolyze to the unprotected amines. To avoid
these problems, a different reaction sequence was chosen.
The synthesis started from commercially available benzi-
midazole-5-carboxylic acid (6), which was esterified, nitrat-
ed, and N-protected to furnish a mixture of isomers 7a and
7b. Bromination of isomer 7a in position 2, furnishing inter-
mediate 8,[7] followed by SNAr substitution with methyla-
mine, delivered 2-(methylamino)benzimidazole (9). After
reduction of the nitro group, iodination of 5-aminobenzimi-
dazole 10 occurred exclusively in position 4 to give benzimi-
dazole 11 in excellent yield. Suzuki cross-coupling with
4,4,5,5-tetramethyl-2-vinyl-1,3,2-dioxaborolane furnished 4-
vinylbenzimidazole 12. This styrene derivative could be con-
verted to alcohol 13 by hydroboration with 9-BBN, followed
by oxidative workup (Scheme 1).
In a first attempt to synthesize a secondary amine bearing
a hydrophobic substituent, alcohol 13 was tosylated and sub-
sequently substituted with (cyclohexylmethyl)amine. Yields
in this procedure were unsatisfactory because of side prod-
ucts resulting from N-tosylation of the exocyclic NH2 group
of the lin-benzoguanine or elimination instead of substitu-
tion. Still, it was possible to cyclize benzimidazole 14 with
chloroformamidinium chloride under very harsh reaction
conditions with concomitant N-deprotection to yield the de-
sired inhibitor 5 (Scheme 2).
the overall binding affinity is left unchanged or somewhat
reduced when compared to the unsubstituted parent scaf-
fold, which leaves the cluster of water molecules unpertur-
bed.
Ligand design: The rational design of the new ligands, de-
scribed in detail elsewhere,[19] was based on a DrugScoreCSD
HotSpot analysis, docking studies using GOLD,[13] and mod-
eling with the program MOLOC;[20] and has been rigorously
validated by extensive crystal structure determinations.[19]
We reasoned that side chains must be able to take over the
function of the expelled water molecules and pick up the H-
bond network with remaining water molecules and the as-
partate side chains. At the same time, they should bear a hy-
drophobic moiety capable of undergoing favorable hydro-
phobic interactions with the ribose-34 surface. Secondary
amines were deemed especially suitable as polar linkers be-
cause they were assumed to be protonated in the TGT
active site, thus forming a strong ionic hydrogen bond to
Asp280. As apolar substituents, small hydrocarbon rings,
such as cyclohexane, were chosen. Previous work had shown
that substitution of the lin-benzoguanine scaffold with a 2-
methylamino group is highly favorable because it raises the
pKa of the fused imidazole ring by about one unit. The pro-
tonation of the aminoimidazole moiety enables the forma-
tion of a charge-assisted hydrogen bond to the Leu231 back-
bone carbonyl.[7,8] Therefore, the 2-methylamino group was
also incorporated in the designed ligands, resulting in lin-
benzoguanine 5 (Figure 2).[19]
Because the tosylation–substitution sequence (Scheme 2)
was thought unsatisfactory, a more reliable synthetic path-
way was searched for. Cyclization of alcohol 13 with chloro-
formamidinium chloride gave access to lin-benzoguanine 15.
Alternatively, a Mitsunobu reaction of alcohol 13, followed
by cleavage of phthalimide 16 with hydrazine, furnished the
primary amine 17. Cyclization with chloroformamidinium
Synthesis: The previously reported synthesis of 4-substituted
lin-benzoguanines employed Sonogashira cross-couplings of
phenylacetylenes with a 4-iodobenzimidazole to introduce
the side chain. The triple bond was then reduced to the
Chem. Eur. J. 2009, 15, 10809 – 10817
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10811

F. Diederich et al.
first reported by DeStevens and
Dughi.[26] X-ray structure analy-
sis confirmed the presence of
the seven-membered cyclic
aminal (Figure 3). Subsequent
reduction of the crude inter-
mediate with NaBHACHTUNRGTNEUNG(OAc)3 fur-
nished the desired tertiary
amine 29 as a single product.
Hydrogenolysis provided meth-
ylamine 30, and subsequent cyc-
lization gave lin-benzoguanine
27 (Scheme 4).
Control compound 31 was
synthesized by acylation of
amine 17, followed by cycliza-
tion of the resulting amide 32.
The 4-phenylbutyl substituent
of lin-benzoguanine 33 was in-
troduced by Sonogashira cross-
coupling of aryl iodide 11 with
4-phenyl-1-butyne to furnish
alkyne 34. Hydrogenation of
the triple bond to the alkyl de-
rivative 35 required repeated
Scheme 1. Synthesis of alcohol 13. a) SOCl2, MeOH, 508C; 97%. b) HNO3, H2SO4, 508C; 76%. c)
Me2NSO2Cl, Et3N, toluene, reflux; 35% 4a, 29% 4b. d) 1) LHMDS, THF, ꢁ788C; 2) CBr4, THF, ꢁ78 !
228C; 56%. e) MeNH2, EtOH, 08C; 91%. f) Zn, AcOH, H2O, 22 8C; 94%. g) I2, NaHCO3, CH2Cl2, H2O,
228C; 98%. h) [PdCl
228C; 70%. LHMDS=lithium hexamethyldisilazide; DME=1,2-dimethoxyethane; 9-BBN=9-
borabicyclo[3.3.1]nonane.
2ACHTUNGTRENNUNG(PPh3)2], Et3N, DME, H2O, 80 8C; 87%. i) 1) 9-BBN, THF, 228C; 2) H2O2, NaOH, H2O,
0
!
ACHTUNGTRENNUNG
Scheme 2. First-approach to the synthesis of inhibitor 5. a) 1) pTsCl,
Et3N, DMAP, CH2Cl2, 08C; 2) (cyclohexyl)methylamine, CH2Cl2, 0 !
228C; 38%. b) Chloroformamidinium chloride, dimethyl sulfone, 1508C;
54%. pTsCl=para-toluenesulfonyl chloride; DMAP=4-(dimethylami-
no)pyridine.
chloride yielded lin-benzoguanine 18, while reductive ami-
nation with appropriate aldehydes gave access to the secon-
dary amines 14 and 19–22, and, after cyclization, to lin-ben-
zoguanines 5 and 23–26, respectively. The cyclization yields
were somewhat lower for the benzylic amines 25 and 26 due
to partial debenzylation under the reaction conditions. This
synthetic sequence is less prone to side reactions than the
first approach. Rather, it seems to be applicable to a large
variety of aldehydes RCHO, giving easy access to a broad
range of substituents (Scheme 3).
For the synthesis of the control compound 27 with an N-
methylaminoethyl substituent, a benzyl protecting group
had to be introduced to prevent double methylation of the
primary amine 17. When benzylated intermediate 21 was
subjected to mild Eschweiler–Clarke-type conditions, tetra-
hydro-1,3-benzodiazepine 28 was obtained exclusively. Few
examples[22–25] are known of this unusual structure that was
Figure 3. Molecular structure of 28 (ORTEP plot), arbitrary numbering.
Atomic displacement parameters obtained at 203 K are drawn at the
50% probability level. The two NH groups form intramolecular H-bonds
(green dotted; N10-H···O26: 1.97 ꢃ, N15-H···O18 2.10 ꢃ).
addition of catalyst, presumably due to poisoning by the
substrate (total catalyst load of 28 mol%). Cyclization with
chloroformamidinium chloride afforded the tricycle 33
(Scheme 5).
Biological activity: The synthesized inhibitors were subject-
ed to kinetic measurements with the enzyme, as described in
the literature.[5,7,8] Simple substitution of the parent scaffold
2-methylamino-lin-benzoguanine (36)[7,8] with a 4-phenylbu-
10812
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 10809 – 10817

Inhibitors of tRNA-Guanine Transglycosylase
FULL PAPER
elsewhere.[19] Apparently, the
uncharged
hydroxy
group
cannot fully replace the expel-
led water molecules in terms of
hydrogen bonding to the Asp
side chains and residual water
molecules. This goal is appa-
rently achieved by the amino-
and
methylamino-substituted
compounds 18 and 27, which
show the same or a slightly im-
proved affinity, respectively. An
amide linker (31) leads again to
a strong loss in binding potency
because the amide bond lacks a
positive charge and seems to be
too rigid and unable to undergo
the necessary hydrogen bonds.
Furthermore, inadequate desol-
vation of the amide bond upon
transfer from the bulk water
environment to the protein
binding site is detrimental to
high affinity binding. Although
benzylamine 25 bears a hydro-
phobic phenyl substituent on
the amino group, it displays the
same Ki value as control com-
pound 27. In the case of naph-
thalene 26, there is even a loss
in affinity. It seems that aromat-
ic substituents are too rigid to
align optimally to the hydro-
phobic surface crevice. The lin-
benzoguanines 23, 5, and 24
with alicyclic substituents on
the secondary amine linker, on
the other hand, display strong
inhibitors of TGT in the low
nanomolar range. The strongest
inhibitor 23 (2 nm) exhibits a
~30-fold increase in binding
strength compared to the un-
substituted scaffold 36, confirm-
ing the formation of favorable
hydrophobic interactions with
the apolar surface near ribose-
34 (Table 2).[19]
Scheme 3. Synthesis of inhibitors 5, 15, 18, and 23–26. a) Chloroformamidinium chloride, dimethyl sulfone,
1508C; 65%. b) Phthalimide, PPh3, DIAD, CH2Cl2, 228C; 85%. c) H2NNH2·H2O, CH2Cl2/MeOH, 228C; 96%.
d) Chloroformamidinium chloride, dimethyl sulfone, 1508C; 68%. e) RCHO, NaBHACTHNUTRGNE(UNG OAc)3, CH2Cl2, 228C;
60% 19, 61% 14, 55% 20, 71% 21, 63% 22. f) Chloroformamidinium chloride, dimethyl sulfone, 1508C; 67%
23, 54% 5, 76% 24, 26% 25, 41% 26. DIAD=diisopropyl azodicarboxylate.
Scheme 4. Synthesis of inhibitor 27. a) CH2O, HCO2H, CH2Cl2/H2O, 0 ! 228C. b) NaBH
ACHTUNGRTEN(NUGN OAc)3, 1,2-dichloro-
ethane, reflux; 83% over two steps. c) 1 bar H2, Pd/C, THF, 228C; 87%. d) Chloroformamidinium chloride, di-
methyl sulfone, 1508C; 60%.
tyl residue (33) results in a loss in affinity due to repulsive
interactions between the alkyl chain and the negatively
charged Asp102 and Asp280 side chains and due to the ex-
pulsion of the solvating water molecules. In this case, the re-
pelled water cluster is obviously not properly replaced. In-
hibitor 15 bearing a 2-hydroxyethyl group in position 4 loses
a factor of 1.7 in binding affinity compared to the parent
scaffold 36. The conserved binding mode of compound 15
could be confirmed by crystal structure analysis described
These results, together with the previously reported ones
(Table 1),[5] are strong evidence for the validity of our design
hypothesis. The water cluster located between Asp102 and
Asp280 can be successfully replaced by substituents bearing
a polar, H-bond-forming functionality in the appropriate po-
sition. The protonated secondary ammonium ions 18 and 27
succeed in replacing the function of the expelled water mol-
ecules by forming strong ionic H-bonds to Asp280 and par-
ticipating in the H-bonding network of the remaining water
Chem. Eur. J. 2009, 15, 10809 – 10817
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10813

F. Diederich et al.
to fit the aromatic rings deeply into the hydrophobic
patch.[19]
Protein–ligand co-crystal structure: The predicted binding
mode of lin-benzoguanine 5 is further supported by a crystal
structure determination, described in detail in another
study.[19] The tricyclic lin-benzoguanine scaffold is held in
position by an array of hydrogen bonds to Asp102, Asp156,
Gln203, Gly230, Leu231, and Ala232, as well as by stacking
interactions with the Tyr106 and Met260 sandwiching from
above and below (not shown). The protonated secondary
ammonium group forms two strong ionic H-bonds (d-
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
the lin-benzoguanine scaffold. The CH2-CH2-NH2+-CH2 unit
in the linker chain shows a slightly strained gauche confor-
mation in order to pick up these interactions. The cyclohex-
ane moiety adopts a well-defined chair conformation and
lies on the hydrophobic surface formed by Val45 and Leu68
(Figure 4).[19]
Scheme 5. Synthesis of inhibitors 31 and 33. a) Cyclohexanecarbonyl
chloride, Et3N, CH2Cl2, 228C; 61%. b) Chloroformamidinium chloride,
dimethyl sulfone, 1508C; 22%. c) 4-Phenyl-1-butyne, [PdCl2ACTHNUTRGEN(NUG PPh3)2],
CuI, iPr2NH, 808C; 62%. d) 1 bar H2, Pd/C, CH2Cl2/MeOH, 228C; 77%.
e) Chloroformamidinium chloride, dimethyl sulfone, 1508C; 54%.
Conclusions
Table 2. lin-Benzoguanine inhibitors substituted in position 4 and their inhibition constants.
In this study, the possibility of
displacing fixed water mole-
cules between Asp102 and
Asp280 by polar, H-bond-form-
ing linkers is demonstrated.
While a simple alcohol, such as
15, is unable to fully replace the
function of the expelled water
molecules, the amines 18 and
27 display the same or a slightly
improved binding compared to
the unsubstituted scaffold 36.
Accordingly, the protonated
ammonium centers are able to
form H-bonds strong enough to
replace the ones lost by the re-
lease of water. The costs in po-
tency associated with the desol-
vation of the two Asp side
chains are therefore fully com-
pensated by the formation of
Compound
R
H
Ki/nm
Compound
R
Ki/nm
25ꢂ2
36
58ꢂ36[7]
25
33
235ꢂ50
26
105ꢂ10
15
18
97ꢂ5
55ꢂ3
23
5
2ꢂ1
4ꢂ2
27
31
26ꢂ6
24
2.5ꢂ1
1400ꢂ100
molecules, in full accordance with our hypothesis. Attaching
a terminal alicyclic ring to the ammonium substituent, as in-
troduced in inhibitors 23, 5, and 24, leads to a large increase
in binding potency that must be due to hydrophobic interac-
tions with the ribose-34 hydrophobic patch. Terminal aro-
matic rings attached to the ammonium center, as exempli-
fied by compounds 25 and 26, seem to be less beneficial,
which is probably due to the reduced flexibility of the ring
that prevents an optimal adaptation to the apolar surface
crevice. Furthermore, according to the modeling, unfavora-
ble torsional angles NH2+-CH2-Csp2-Csp2 seem to be required
ionic H-bonds to Asp280 and involving a remaining water
molecule in the interactions. Further substitution of the ter-
minal ammonium group with aliphatic substituents (23, 5,
24) experiences additional hydrophobic interactions with the
apolar surface of a small crevice formed by Val45 and
Leu68 near the ribose-34 site, causing another favorable
contribution to the binding free energy. Thus, the inhibitory
constants of 23, 5, and 24 are further lowered and the most
potent inhibitor 23, bearing a terminal cyclopentyl ring, ex-
hibits a Ki of 2ꢂ1 nm. The aromatic substituents in inhibi-
tors 25 and 26 seem to be too rigid for optimal nesting on
10814
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 10809 – 10817

Inhibitors of tRNA-Guanine Transglycosylase
FULL PAPER
with a Golden-Gate ATR unit. The bands are indicated in cmꢁ1 and cor-
responding strength (s: strong; m: medium; w: weak). Broad bands are
indicated as br. NMR spectra were recorded at 228C unless otherwise
stated. The signal positions are given in ppm relative to tetramethylsi-
lane, the coupling constants (J values) in Hz. The signal multiplicities are
given as s (singulet), d (doublet), t (triplet), q (quadruplet), quint. (quin-
tet), sext. (sextet), sept. (septet), or combinations thereof. Broad signals
are indicated as br. 300 MHz: Varian Gemini 300, Varian Mercury 300,
or Bruker ARX 300. 400 MHz: Bruker DRX 400 or Bruker AV 400.
500 MHz: Bruker DRX 500. Mass spectra were recorded and evaluated
by the MS service of the Laboratorium fꢀr Organische Chemie at ETH
Zꢀrich. Line positions are given in m/z units. EI-MS: Waters Micromass
AutoSpec-Ultima spectrometer. ESI- and MALDI-MS: Varian IonSpec
FT-ICR-MS spectrometer. 3-Hydroxypicolinic acid (3-HPA) was used as
matrix for MALDI-MS. Elemental analyses were performed by the Mik-
rolabor at the Laboratorium fꢀr Organische Chemie at ETH Zꢀrich with
a LECO CHN/900 instrument. Nomenclature mainly follows the comput-
er program ACD Name. Crystallographic pictures were prepared using
PyMol.[27]
Figure 4. X-ray crystal structure of inhibitor 5 (green) in the active site of
TGT from Z. mobilis (grey) described in ref. [19]. The residues Val45
and Leu68 form a hydrophobic surface (shown) that hosts the cyclohex-
ane ring. Atom distances in ꢃ. Resolution 1.78 ꢃ. PDB code 3EOS.
General procedure 1 (GP 1) for the cyclization with chloroformamidini-
um chloride:[5] The benzimidazole (1.0 equiv), chloroformamidinium
chloride (2.0–2.8 equiv), and dimethylsulfone (50–67 equiv) were mixed
and heated to 1508C for 3–4 h.
General procedure 2 (GP 2) for the reductive amination of a primary
the hydrophobic patch; therefore no favorable overall con-
tribution to the binding affinity is observed.
amine: To the amine (1.00 equiv), a solution of the aldehyde (0.99–
1.08 equiv) in CH2Cl2 (0.075–0.107m) and NaBHACHTNUTRGNE(NUG OAc)3 (3.0–6.2 equiv)
were added and the mixture was stirred at 228C for 30 min–17 h. Subse-
quently, 1m aqueous NaOH was added and the mixture extracted with
CH2Cl2. The combined organic phases were dried over Na2SO4, filtered,
and evaporated. Purification was accomplished by CC (SiO2; CH2Cl2/
MeOH/25% aqueous NH3 97:2:1).
All of these ligands, bearing different amino substituents
in position 4 of the lin-benzoguanine scaffold, became avail-
able by a versatile synthetic route starting from the 2-meth-
ylamino-4-iodo-benzimidazole 11. A sequence of Suzuki
cross-coupling, hydroboration, Mitsunobu reaction, and Ga-
briel synthesis furnishes the primary amine 17 as a late
common intermediate, which is further substituted via re-
ductive amination. This synthetic route allows for easy fur-
ther optimization of the substituent in position 4, in order to
tune binding and physicochemical properties of the ligands.
Methyl 1-[(dimethylamino)sulfonyl]-2-(methylamino)-5-nitro-1H-benz-
AHCTUNGTREiGNUNN midazol-6-carboxylate (9): Ethanolic MeNH2 solution (20 mL,
161 mmol, ~8m) was added to 8 (3.19 g, 7.84 mmol) at 08C, and the mix-
ture was stirred at the same temperature for 10 min. After evaporation,
purification by CC (SiO2; hexane/AcOEt 50:50! 0:100) afforded 9
(2.55 g, 91%) as a yellow solid. M.p. 169–1708C; 1H NMR (CDCl3,
300 MHz): d=7.89, 7.81 (2 s, 2H), 6.39 (brq, J=4.8 Hz, 1H), 3.89 (s,
3H), 3.17 (d, J=4.8 Hz, 3H), 2.95 ppm (s, 6H); 13C NMR (CDCl3,
75 MHz): d=29.95, 38.67, 53.06, 112.19, 112.94, 119.54, 133.66, 144.39,
146.33, 156.08, 165.89 ppm; IR (neat): n˜ =3436 (m), 2923 (w), 1722 (s),
1632 (s), 1567 (m), 1524 (s), 1444 (s), 1418 (m), 1371 (s), 1343 (s), 1293
(s), 1254 (s), 1219 (s), 1149 (s), 1114 (m), 1051 (m), 1012 (m), 982 (s), 891
(m), 875 (m), 838 (m), 822 (m), 788 (m), 728 (s), 713 (m), 631 cmꢁ1 (m);
HR-MALDI-MS (3-HPA): m/z: calcd for C12H16N5O6S+ [M+H]+:
358.0816; found: 358.0822; elemental analysis calcd (%) for C12H15N5O6S
(357.35): C 40.33, H 4.23, N 19.60; found: C 40.62, H 4.28, N 19.42.
Experimental Section
Materials and general methods: Compounds 7a/b and 8 were prepared as
described in the literature.[7] Details for the biological assay, the X-ray
crystal structure of 28, and the synthetic procedures for compounds 15,
18–27, and 30–35 can be found in the Supporting Information. Chemicals
were reagent-grade, purchased from commercial suppliers, and used with-
out further purification unless otherwise stated. AcOEt, hexane, and
CH2Cl2 for extraction and chromatography were used in technical quality
and distilled before use. Absolute THF, CH2Cl2, and toluene used in re-
actions were prepared by distillation over a drying agent (THF: Na/ben-
zophenone; CH2Cl2: CaH2; toluene: Na). Reaction monitoring was ach-
ieved by thin-layer chromatography on silica-coated glass plates with
fluorescence marker (60 F254, Merck). Visualization occurred by UV light
(254 nm) or coloring with potassium permanganate solution (KMnO4
(3 g), K2CO3 (20 g), NaOH (0.25 g), H2O (300 mL)). Alternatively, ana-
lytical LC/MS using an Ultimate 3000 series LC instrument combined
with an MSQ Plus mass spectrometer from Dionex, using a Zorbax
Eclipse Plus C18 column (30ꢄ3 mm; 3.5 mm pore size) from Agilent or a
Reprospher C18-Aqua column (30ꢄ3 mm; 3 mm pore size) from Dr.
Maisch was employed. Column chromatography (CC) was done on Kie-
selgel 60 silica gel from Fluka or SilicaFlash F60 from SiliCycle (normal-
phase), or MCI gel CHP20P from Supelco (reverse-phase) with 0–0.5 bar
overpressure. The corresponding solvent mixtures are given in brackets.
Drying in high vacuum (HV) was done at 10ꢁ2 Torr. Melting points were
measured on a Bꢀchi Meltingpoint B-540 apparatus and are uncorrected.
IR spectra were recorded on a Varian 800 FT-IR ATR device equipped
Methyl 5-amino-1-[(dimethylamino)sulfonyl]-2-(methylamino)-1H-benz-
AHCTUNGERTGiNNUN midazol-6-carboxylate (10): Compound 9 (2.55 g, 7.14 mmol) was dis-
solved in a mixture of AcOH/H2O 5:1 (200 mL). Zn powder (4.72 g,
72.2 mmol) was added and the mixture stirred at 228C for 20 min and fil-
tered through Celite (rinsed with MeOH). After evaporation and drying
in HV, the residue was re-dissolved in CH2Cl2 and again filtered through
Celite (rinsed with CH2Cl2). Purification by CC (SiO2; AcOEt) gave 10
(2.20 g, 94%) as a pale yellow solid. M.p. 178–1798C; 1H NMR (CDCl3,
300 MHz): d=8.00 (s, 1H), 6.62 (s, 1H), 6.33 (brq, J=4.8 Hz, 1H), 5.71
(brs, 2H), 3.86 (s, 3H), 3.13 (d, J=4.8 Hz, 3H), 2.91 ppm (s, 6H);
13C NMR (CDCl3, 75 MHz): d=29.75, 38.67, 51.38, 102.53, 104.28, 113.83,
123.61, 148.05, 149.24, 156.00, 168.47 ppm; IR (neat): n˜ =3405 (s), 3306
(m), 2951 (w), 1686 (m), 1593 (s), 1452 (s), 1418 (s), 1375 (s), 1263 (s),
1236 (s), 1196 (s), 1157 (s), 1025 (s), 955 (s), 710 cmꢁ1 (s); HR-MALDI-
MS (3-HPA): m/z: calcd for C12H17N5O4S+ [M]+: 327.0996; found:
327.0989.
Methyl
5-amino-1-[(dimethylamino)sulfonyl]-4-iodo-2-(methylamino)-
solution of 10 (2.20 g,
1H-benzimidazol-6-carboxylate (11): To
a
6.71 mmol) in CH2Cl2 (120 mL), saturated aqueous NaHCO3 solution
(60 mL) and iodine (2.00 g, 7.89 mmol) were added and the mixture was
stirred at 228C for 3 d. After addition of saturated aqueous Na2S2O3
Chem. Eur. J. 2009, 15, 10809 – 10817
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10815

F. Diederich et al.
(50 mL), the aqueous phase was extracted with CH2Cl2 (2ꢄ100 mL) and
the combined organic phases were dried over Na2SO4, filtered, and
evaporated. Purification by CC (SiO2; hexane/AcOEt 50:50) yielded 11
(2.99 g, 98%) as a yellow solid. M.p. 177–1788C; 1H NMR (CDCl3,
300 MHz): d=8.02 (s, 1H), 6.43 (brs, 2H), 6.38 (brq, J=5.1 Hz, 1H),
3.87 (s, 3H), 3.21 (d, J=5.1 Hz, 3H), 2.91 ppm (s, 6H); 13C NMR
(CDCl3, 75 MHz): d=29.91, 38.71, 51.73, 73.06, 103.85, 113.96, 121.59,
148.47, 149.88, 155.51, 167.93 ppm; IR (neat): n˜ =3313 (m), 3277 (w),
2947 (w), 1683 (m), 1574 (s), 1406 (m), 1383 (s), 1261 (s), 1195 (s), 1144
(s), 1022 (s), 953 (s), 781 (m), 709 cmꢁ1 (s); HR-ESI-MS: m/z: calcd for
C12H17IN5O4S+ [M+H]+: 454.0040; found: 454.0043; elemental analysis
calcd (%) for C12H16IN5O4S (453.26): C 31.80, H 3.56, N 15.45; found: C
31.82, H 3.58, N 15.16.
26.68, 29.84, 31.47, 38.13, 38.76, 49.12, 51.41, 56.82, 104.35, 111.98, 113.48,
122.91, 146.98, 148.18, 155.37, 169.16 ppm; IR (neat): n˜ =3437 (w), 3413
(w), 3338 (w), 2926 (w), 2842 (w), 1676 (w), 1583 (s), 1505 (w), 1415 (s),
1389 (m), 1312 (w), 1262 (m), 1202 (s), 1161 (s), 1074 (w), 1021 (w), 962
(m), 891 (w), 807 (w), 715 (s), 662 cmꢁ1 (w); HR-MALDI-MS (3-HPA):
m/z: calcd for C21H35N6O4S+ [M+H]+: 467.2435; found: 467.2439.
6-Amino-4-{2-[(cyclohexylmethyl)amino]ethyl}-2-(methylamino)-1,7-di-
hydro-8H-imidazoACHTNUTGRNEUNG[4,5-g]quinazolin-8-one (5): GP 1 was followed using
14 (38 mg, 0.081 mmol). Purification by CC (MCI gel; H2O[0.1% HCl]/
MeCN 100:0!95:5), followed by lyophilization, provided 5 (21 mg, 54%,
HCl triple salt) as a fawn solid. M.p.> 2508C; 1H NMR (D2O + 1 drop
trifluoroacetic acid (TFA), 400 MHz): d=7.91 (s, 1H), 3.40–3.36 (m,
2H), 3.28–3.23 (m, 2H), 3.10 (s, 3H), 2.95 (d, J=6.4 Hz, 2H), 1.72–1.66
(m, 5H), 1.60 (brd, J=12.0 Hz, 1H), 1.26–1.10 (m, 3H), 1.02–0.94 ppm
(m, 2H); 13C NMR (D2O + 1 drop TFA, 100 MHz): d=22.01, 24.92,
25.43, 29.18, 29.82, 34.66, 45.65, 54.06, 107.02, 108.70, 111.71, 128.05,
133.54, 136.09, 150.98, 153.07, 161.23 ppm; IR (neat): n˜ =2924 (m), 2847
(w), 2797 (brw), 1675 (s), 1645 (s), 1529 (w), 1444 (s), 1310 (w), 1227 (w),
1153 (w), 1097 (w), 982 (m), 832 (w), 756 cmꢁ1 (w); HR-ESI-MS: m/z:
calcd for C19H28N7O+ [M+H]+: 370.2350; found: 370.2352.
Methyl
1H-benzimidazole-6-carboxylate (12): To a suspension of 11 (200 mg,
0.44 mmol) and [PdCl2A(PPh3)2] (31 mg, 0.04 mmol) in DME (1 mL) and
5-amino-1-[(dimethylamino)sulfonyl]-2-(methylamino)-4-vinyl-
CHTUNGTRENNUNG
H2O (0.2 mL), Et3N (0.18 mL, 1.29 mmol) and 4,4,5,5-tetramethyl-2-
vinyl-1,3,2-dioxaborolane (0.12 mL, 0.69 mmol) were added and the mix-
ture was heated to 808C for 75 min. After evaporation, purification by
CC (SiO2; hexane/AcOEt 67:33!50:50) provided 12 (136 mg, 87%) as a
yellow oil. 1H NMR (CDCl3, 300 MHz): d=7.99 (s, 1H), 6.92 (dd, J=
17.9, 11.8 Hz, 1H), 6.31 (brq, J=4.9 Hz, 1H), 6.21 (dd, J=17.9, 2.1 Hz,
1H), 6.14 (brs, 2H), 5.68 (dd, J=11.8, 2.1 Hz, 1H), 3.86 (s, 3H), 3.17 (d,
J=4.9 Hz, 3H), 2.90 ppm (s, 6H); 13C NMR (CDCl3, 75 MHz): d=29.84,
38.74, 51.50, 104.32, 111.13, 113.20, 119.46, 123.30, 128.63, 146.12, 146.76,
155.79, 168.95 ppm; IR (neat): n˜ =3418 (w), 2948 (w), 1687 (w), 1581 (s),
1406 (w), 1384 (s), 1266 (m), 1189 (s), 1159 (s), 1050 (m), 958 (m), 796
(w), 713 cmꢁ1 (s); HR-MALDI-MS (3-HPA): m/z: calcd for
C14H20N5O4S+ [M+H]+: 354.1231; found: 354.1231.
Methyl 5-amino-1-[(dimethylamino)sulfonyl]-4-[2-(1,3-dioxo-1,3-dihydro-
2H-isoindol-2-yl)ethyl]-2-(methylamino)-1H-benzimidazole-6-carboxylate
(16): To a solution of 13 (374 mg, 1.01 mmol) in CH2Cl2 (10 mL), PPh3
(538 mg, 2.05 mmol), DIAD (0.40 mL, 2.03 mmol), and phthalimide
(299 mg, 2.03 mmol) were added and the mixture was stirred at 228C for
16 h and evaporated. Purification by CC (SiO2; hexane/AcOEt 67:33!
50:50) gave 16 (429 mg, 85%) as
a yellow solid. M.p. 206–2098C;
1H NMR (CDCl3, 300 MHz): d=7.96 (s, 1H), 7.81, 7.71 (AA’MM’, 4H),
6.27 (brs, 2H), 6.16 (brq, J=4.9 Hz, 1H), 3.93 (dd, J=8.4, 6.7 Hz, 2H),
3.85 (s, 3H), 3.20 (dd, J=8.4, 6.7 Hz, 2H), 3.03 (d, J=4.9 Hz, 3H),
2.85 ppm (s, 6H); 13C NMR (CDCl3, 100 MHz): d=24.23, 29.68, 35.78,
38.77, 51.44, 104.23, 109.57, 112.78, 122.79, 123.11, 132.33, 133.79, 147.37,
147.44, 155.60, 168.41, 169.05 ppm; IR (neat): n˜ =3466 (w), 3426 (w),
3362 (w), 2944 (w), 1763 (w), 1703 (m), 1673 (w), 1584 (s), 1428 (w), 1394
(m), 1274 (m), 1206 (s), 1157 (s), 1103 (w), 1066 (m), 965 (m), 792 (w),
716 cmꢁ1 (s); HR-MALDI-MS (3-HPA): m/z: calcd for C22H25N6O6S+
[M+H]+: 501.1551; found: 501.1546.
Methyl 5-amino-1-[(dimethylamino)sulfonyl]-4-(2-hydroxyethyl)-2-(meth-
ylamino)-1H-benzimidazole-6-carboxylate (13): Compound 12 (136 mg,
0.38 mmol) was dissolved in a 9-BBN solution (2.3 mL, 1.2 mmol; 0.5m in
THF) and stirred at 228C for 8 h. At 08C, H2O2 (0.20 mL, 1.96 mmol;
30% in H2O) and 1m aqueous NaOH (2.0 mL, 2.0 mmol) were carefully
added. The mixture was stirred at 228C for 40 min, diluted with saturated
aqueous NH4Cl (20 mL) and extracted with AcOEt (3ꢄ20 mL). The
combined organic phases were dried over Na2SO4, filtered, and evaporat-
ed. Purification by CC (SiO2; AcOEt) afforded 13 (100 mg, 70%) as a
cream-colored solid. An analytical sample was obtained by recrystalliza-
tion from hexane/AcOEt. M.p. 157–1588C; 1H NMR (CDCl3, 300 MHz):
d=7.98 (s, 1H), 6.34 (brq, J=4.9 Hz, 1H), 5.85 (brs, 2H), 4.01 (t, J=
5.5 Hz, 2H), 3.85 (s, 3H), 3.12 (d, J=4.9 Hz, 3H), 2.98 (t, J=5.5 Hz,
2H), 2.92 ppm (s, 6H); 13C NMR (CDCl3, 75 MHz): d=29.51, 29.87,
38.76, 51.57, 61.70, 104.85, 112.11, 112.32, 122.97, 145.90, 147.42, 155.23,
169.00 ppm; IR (neat): n˜ =3427 (w), 3333 (w), 2944 (w), 1687 (w), 1592
(s), 1513 (w), 1413 (m), 1373 (m), 1267 (m), 1200 (s), 1155 (s), 1053 (m),
1024 (w), 961 (s), 790 (w), 750 (w), 712 cmꢁ1 (s); HR-MALDI-MS (3-
HPA): m/z: calcd for C14H22N5O5S+ [M+H]+: 372.1336; found: 372.1329;
elemental analysis calcd (%) for C14H21N5O5S (371.42): C 45.27, H 5.70,
N 18.86; found: C 45.84, H 5.79, N 18.64.
Methyl 5-amino-4-(2-aminoethyl)-1-[(dimethylamino)sulfonyl]-2-(methyl-
amino)-1H-benzimidazole-6-carboxylate (17): To
a solution of 16
(458 mg, 0.92 mmol) in CH2Cl2 (10 mL) and MeOH (10 mL), hydrazine
monohydrate (0.45 mL, 9.3 mmol) was added and the mixture stirred at
228C for 2 d. After evaporation, the mixture was taken up in 1m aqueous
NaOH (50 mL) and extracted with CH2Cl2 (4ꢄ50 mL). The combined or-
ganic phases were dried over Na2SO4, filtered, and evaporated to yield 17
(327 mg, 96%) as a yellow solid. No further purification was done. M.p.
139–1418C; 1H NMR (CDCl3, 300 MHz): d=7.92 (s, 1H), 6.25 (brq, J=
4.8 Hz, 1H), 3.83 (s, 3H), 3.11 (d, J=4.8 Hz, 3H), 2.99 (brs, 4H),
2.88 ppm (s, 6H); 13C NMR (CDCl3, 100 MHz): d=29.24, 29.79, 38.67,
40.83, 51.38, 104.43, 112.05, 112.22, 122.99, 147.22, 147.53, 155.40,
169.04 ppm; IR (neat): n˜ =3466 (w), 3428 (w), 3362 (w), 2944 (w), 1763
(w), 1703 (w), 1672 (w), 1582 (s), 1427 (w), 1394 (m), 1271 (m), 1204 (s),
1157 (s), 1102 (w), 963 (m), 791 (m), 710 cmꢁ1 (s); HR-MALDI-MS (3-
HPA): m/z: calcd for C14H23N6O4S+ [M+H]+: 371.1496; found: 371.1501.
Methyl 5-amino-4-{2-[(cyclohexylmethyl)amino]ethyl}-1-[(dimethylami-
no)sulfonyl]-2-(methylamino)-1H-benzimidazole-6-carboxylate
(14):
From 13: Compound 13 (84 mg, 0.23 mmol) was dissolved in CH2Cl2
(10 mL). pTsCl (85 mg, 0.45 mmol), Et3N (0.03 mL, 0.22 mmol), and
DMAP (12 mg, 0.10 mmol) were added at 08C, and the mixture was
stirred at the same temperature for 3.5 h and evaporated. CC (SiO2;
hexane/AcOEt 50:50) afforded a colorless oil (71 mg), which was dis-
solved in CH2Cl2 (10 mL). At 08C, (cyclohexylmethyl)amine (0.20 mL,
1.54 mmol) was added, the mixture allowed to reach 228C during 26 h,
and the solvent removed by evaporation. Purification by CC (SiO2;
CH2Cl2/MeOH/25% aqueous NH3 97:2:1) gave 14 (40 mg, 38%) as a
white solid. From 17: GP 2 using 17 (21.9 mg, 0.059 mmol) and cyclohex-
anecarbaldehyde (6.5 mg, 0.058 mmol) yielded 14 (16.9 mg, 61%) as a
white solid. M.p. 173–1748C; 1H NMR (CDCl3, 400 MHz): d=7.92 (s,
1H), 6.24 (q, J=4.9 Hz, 1H), 3.84 (s, 3H), 3.13 (d, J=4.9 Hz, 3H), 3.03
(t, J=6.4 Hz, 2H), 2.89 (s, 6H), 2.88 (t, J=6.4 Hz, 2H), 2.47 (d, J=
6.7 Hz, 2H), 1.72–1.66 (m, 5H), 1.47–1.37 (m, 1H), 1.27–1.07 (m, 3H),
0.93–0.83 ppm (m, 2H); 13C NMR (CDCl3, 100 MHz): d=26.09, 26.27,
Methyl
3,6,7,8,9,10-hexahydroimidazo
(28) and methyl 5-amino-4-{2-[benzylACHTUNGTRENNNUG
8-benzyl-3-[(dimethylamino)sulfonyl]-2-(methylamino)-
[4,5-g][1,3]benzodiazepine-5-carboxylate
(methyl)amino]ethyl}-1-[(dimethy-
G
ACHTUNGTRENNUNG
lamino)sulfonyl]-2-(methylamino)-1H-benzimidazole-6-carboxylate (29):
A solution of 21 (63 mg, 0.137 mmol) in CH2Cl2 (2.0 mL) was cooled to
08C, and formic acid (16 mL, 0.42 mmol) and a solution of formaldehyde
(13.5 mL, 0.175 mmol; 36% in H2O) in H2O (0.1 mL) were added. The
mixture was allowed to warm to 228C during 16 h, 1m aqueous NaOH
(10 mL) was added, and the mixture was extracted with CH2Cl2 (2ꢄ
10 mL). The combined organic phases were dried over Na2SO4, filtered,
and evaporated, providing 28 as a pale yellow solid. M.p. 156–1588C;
1H NMR (CDCl3, 300 MHz): d=7.97 (s, 1H), 7.73 (brt, J=4.0 Hz, 1H),
7.44–7.25 (m, 5H), 6.27 (q, J=4.9 Hz, 1H), 4.16 (d, J=4.0 Hz, 2H), 3.95
(s, 2H), 3.84 (s, 3H), 3.39–3.35 (m, 2H), 3.15 (d, J=4.9 Hz, 3H), 3.10–
3.07 (m, 2H), 2.90 ppm (s, 6H); 13C NMR (CDCl3, 100 MHz): d=25.36,
10816
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 10809 – 10817

Inhibitors of tRNA-Guanine Transglycosylase
FULL PAPER
29.83, 38.75, 51.63, 52.04, 56.02, 65.38, 107.35, 111.93, 118.19, 124.16,
127.11, 128.38, 129.39, 138.62, 147.18, 152.42, 155.25, 169.23 ppm; IR
(neat): n˜ =3409 (w), 3320 (w), 2937 (w), 2919 (w), 1681 (w), 1635 (w),
1573 (s), 1427 (m), 1410 (m), 1372 (s), 1354 (s), 1271 (m), 1198 (s), 1133
(s), 1049 (s), 1006 (s), 971 (w), 794 (m), 738 (s), 712 cmꢁ1 (s); HR-
MALDI-MS (3-HPA): m/z: calcd for C22H29N6O4S+ [M+H]+: 473.1966;
found: 473.1966. The crude intermediate was dissolved in 1,2-dichloro-
[6] B. Stengl, E. A. Meyer, A. Heine, R. Brenk, F. Diederich, G. Klebe,
J. Mol. Biol. 2007, 370, 492–511.
[7] S. R. Hçrtner, T. Ritschel, B. Stengl, C. Kramer, W. B. Schweizer, B.
Wagner, M. Kansy, G. Klebe, F. Diederich, Angew. Chem. 2007, 119,
8414–8417; Angew. Chem. Int. Ed. 2007, 46, 8266–8269.
[8] T. Ritschel, S. Hoertner, A. Heine, F. Diederich, G. Klebe, Chem-
BioChem 2009, 10, 716–727.
ethane (2.0 mL), NaBH
G
[9] J. E. Ladbury, Chem. Biol. 1996, 3, 973–980.
mixture heated to reflux for 30 min. Subsequently, 1m aqueous NaOH
(10 mL) was added and the mixture extracted with CH2Cl2 (2ꢄ10 mL).
The combined organic phases were dried over Na2SO4, filtered, and
evaporated. Purification by CC (SiO2; CH2Cl2/MeOH/25% aqueous NH3
97:2:1) afforded 29 (54 mg, 83%) as a colorless oil. 1H NMR (CDCl3,
300 MHz): d=7.93 (s, 1H), 7.34–7.20 (m, 5H), 6.56 (brs, 2H), 6.24 (q,
J=4.9 Hz, 1H), 3.85 (s, 3H), 3.62 (s, 2H), 3.12 (d, J=4.9 Hz, 3H), 3.10
(t, J=6.6 Hz, 2H), 2.89 (s, 6H), 2.69 (t, J=6.6 Hz, 2H), 2.30 ppm (s,
3H); 13C NMR (CDCl3, 100 MHz): d=23.90, 29.79, 38.75, 42.21, 51.38,
56.52, 62.22, 104.27, 111.89, 113.81, 122.89, 127.03, 128.23, 129.12, 138.66,
146.82, 148.18, 155.35, 169.17 ppm; IR (neat): n˜ =3417 (w), 2946 (w),
2840 (w), 2794 (w), 1682 (w), 1638 (w), 1582 (s), 1413 (m), 1392 (m),
1373 (w), 1272 (m), 1202 (s), 1156 (s), 1067 (w), 962 (m), 908 (w), 793
(w), 733 (m), 715 cmꢁ1 (s); HR-MALDI-MS (3-HPA): m/z: calcd for
C22H31N6O4S+ [M+H]+: 475.2122; found: 475.2121.
[10] F. A. Quiocho, D. K. Wilson, N. K. Vyas, Nature 1989, 340, 404–407.
[11] P. Y. S. Lam, P. K. Jadhav, C. J. Eyermann, C. N. Hodge, Y. Ru, L. T.
Bacheler, J. L. Meek, M. J. Otto, M. M. Rayner, Y. N. Wong, C.-H.
Chang, P. C. Weber, D. A. Jackson, T. R. Sharpe, S. Erickson-Viita-
nen, Science 1994, 263, 380–384.
[12] J. R. H. Tame, G. N. Murshudov, E. J. Dodson, T. K. Neil, G. G.
Dodson, C. F. Higgins, A. J. Wilkinson, Science 1994, 264, 1578–
1581.
[13] M. L. Verdonk, G. Chessari, J. C. Cole, M. J. Hartshorn, C. W.
Murray, J. W. M. Nissink, R. D. Taylor, R. Taylor, J. Med. Chem.
2005, 48, 6504–6515.
[14] C. Barillari, J. Taylor, R. Viner, J. W. Essex, J. Am. Chem. Soc. 2007,
129, 2577–2587.
[15] A. Amadasi, J. A. Surface, F. Spyrakis, P. Cozzini, A. Mozzarelli,
G. E. Kellogg, J. Med. Chem. 2008, 51, 1063–1067.
[16] J. D. Dunitz, Science 1994, 264, 670.
[17] J. D. Dunitz, Chem. Biol. 1995, 2, 709–712.
[18] J. Gꢀnther, A. Bergner, M. Hendlich, G. Klebe, J. Mol. Biol. 2003,
326, 621–636.
Acknowledgements
[19] T. Ritschel, P. C. Kohler, G. Neudert, A. Heine, F. Diederich, G.
Klebe, ChemMedChem, DOI: 10.1002/cmdc.200900343.
[20] P. R. Gerber, K. Mꢀller, J. Comput.-Aided Mol. Des. 1995, 9, 251–
268.
[21] X. Zhang, Y. Zhang, J. Huang, R. P. Hsung, K. C. M. Kurtz, J. Op-
penheimer, M. E. Petersen, I. K. Sagamanova, L. Shen, M. R.
Tracey, J. Org. Chem. 2006, 71, 4170–4177.
We are grateful to the ETH Research Council and F. Hoffmann-La
Roche for financial support. We thankfully acknowledge the support for
the work at the Philipps-University, Marburg by the Deutsche For-
schungsgemeinschaft (Grants KL1204/1-3 and KL1204/9-1) and the Gra-
duiertenkolleg (“Protein function at the atomic level”, G. Klebe, Univer-
sity of Marburg).
[22] O. Hromatka, M. Knollmꢀller, H. Deschler, Monatsh. Chem. 1969,
100, 469–478.
[1] N. Okada, C. Sasakawa, T. Tobe, M. Yamada, S. Nagai, K. A. Ta-
lukder, K. Komatsu, S. Kanegasaki, M. Yoshikawa, Mol. Microbiol.
1991, 5, 187–195.
[23] P. G. Gassman, S. J. Lee, J. Org. Chem. 1986, 51, 267–268.
[24] J. M. Gardiner, M. R. Bryce, P. A. Bates, M. B. Hursthouse, J. Org.
Chem. 1990, 55, 1261–1266.
[2] K. L. Kotloff, J. P. Winickoff, B. Ivanoff, J. D. Clemens, D. L. Swer-
dlow, P. J. Sansonetti, G. K. Adak, M. M. Levine, WHO Bull. 1999,
77, 651–666.
[3] D. Iwata-Reuyl, Bioorg. Chem. 2003, 31, 24–43.
[4] G. E. Keyser, N. J. Leonard, J. Org. Chem. 1976, 41, 3529–3532.
[5] E. A. Meyer, N. Donati, M. Guillot, W. B. Schweizer, F. Diederich,
B. Stengl, R. Brenk, K. Reuter, G. Klebe, Helv. Chim. Acta 2006, 89,
573–597.
[25] H. M. Ma, Z. Z. Liu, S. Z. Chen, X. T. Liang, Chin. Chem. Lett.
2004, 15, 759–761.
[26] G. DeStevens, M. Dughi, J. Am. Chem. Soc. 1961, 83, 3087–3091.
[27] W. DeLano, 2006, http://www.pymol.org.
Received: May 13, 2009
Published online: September 10, 2009
Chem. Eur. J. 2009, 15, 10809 – 10817
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10817