Sugar Acetonides are a Superior Motif for Addressing the Large, Solvent-Exposed Ribose-33 Pocket of tRNA-Guanine Transglycosylase

Article abstract of DOI:10.1002/chem.201801756

The intestinal disease shigellosis caused by Shigella bacteria affects over 120 million people annually. There is an urgent demand for new drugs as resistance against common antibiotics emerges. Bacterial tRNA-guanine transglycosylase (TGT) is a druggable target and controls the pathogenicity of Shigella flexneri. We report the synthesis of sugar-functionalized lin-benzoguanines addressing the ribose-33 pocket of TGT from Zymomonas mobilis. Ligand binding was analyzed by isothermal titration calorimetry and X-ray crystallography. Pocket occupancy was optimized by variation of size and protective groups of the sugars. The participation of a polycyclic water-cluster in the recognition of the sugar moiety was revealed. Acetonide-protected ribo- and psicofuranosyl derivatives are highly potent, benefiting from structural rigidity, good solubility, and metabolic stability. We conclude that sugar acetonides have a significant but not yet broadly recognized value in drug development.

Full text of DOI:10.1002/chem.201801756

DOI: 10.1002/chem.201801756
Full Paper
&
Drug Design
Sugar Acetonides are a Superior Motif for Addressing the Large,
Solvent-Exposed Ribose-33 Pocket of tRNA-Guanine
Transglycosylase
Levon D. Movsisyan+,[a] Elisabeth Schꢀfer+,[a] Andreas Nguyen,[b] Frederik R. Ehrmann,[b]
Anatol Schwab,[a] Thomas Rossolini,[a] Daniel Zimmerli,[c] Bjçrn Wagner,[c] Hamina Daff,[c]
Andreas Heine,[b] Gerhard Klebe,*[b] and FranÅois Diederich*[a]
Abstract: The intestinal disease shigellosis caused by Shigel-
la bacteria affects over 120 million people annually. There is
an urgent demand for new drugs as resistance against
common antibiotics emerges. Bacterial tRNA-guanine trans-
glycosylase (TGT) is a druggable target and controls the
pathogenicity of Shigella flexneri. We report the synthesis of
sugar-functionalized lin-benzoguanines addressing the
ribose-33 pocket of TGT from Zymomonas mobilis. Ligand
binding was analyzed by isothermal titration calorimetry and
X-ray crystallography. Pocket occupancy was optimized by
variation of size and protective groups of the sugars. The
participation of a polycyclic water-cluster in the recognition
of the sugar moiety was revealed. Acetonide-protected ribo-
and psicofuranosyl derivatives are highly potent, benefiting
from structural rigidity, good solubility, and metabolic stabili-
ty. We conclude that sugar acetonides have a significant but
not yet broadly recognized value in drug development.
cosylation of guanine.[7] Inhibition of bacterial TGT results in a
significant reduction of virulence in Shigella flexneri, the major
agent of the bacterial dysentery called Shigellosis.[8] Both, sub-
strate specificity and virulence control render TGT a promising
drug target to overcome urgent needs in the development of
new therapeutic agents against multi-drug resistant Shigella
strains.[9]
Introduction
Post-translational modifications of transfer RNAs play a crucial
role in the folding and stability of tRNA molecules,[1] modula-
tion of interactions with other cellular macromolecules,[2] and
translation of mRNAs.[3] Links between defects in tRNA modifi-
cations and human diseases have also been reported.[4] The
homo-dimeric enzyme tRNA-guanine transglycosylase (TGT, EC
2.4.2.29) modifies the tRNAs of His, Tyr, Asn, and Asp and is
found in all organisms.[5] In bacteria, the enzyme exchanges
guanine at position 34 of the anticodon with preQ1, a nucleo-
base which is further transformed to queuine in a downstream
process.[6] Eukaryal TGT uses queuine directly for the transgly-
For many years, TGT from Zymomonas mobilis, a protein
with high crystallization propensity, has been used as substi-
tute for TGT from Shigella spp. The binding site of the enzyme
is highly conserved among these bacteria and consists of a
central guanine/preQ1 nucleobase exchange pocket and two
adjacent ribose-33/uracil-33 and ribose-34/phosphate-34 bind-
ing pockets.[10] In our structure-based drug design program,
we identified tricyclic 6-aminoimidazo[4,5-g]quinazolin-8(7H)-
one (lin-benzoguanine) as a privileged scaffold to occupy the
nucleobase exchange pocket.[11] lin-Benzoguanines can be sub-
stituted at C(2’), C(4’), or both positions (Figure 1a). Substitu-
ents at position C(4’) allow to fill the shallow ribose-34/phos-
phate-34 pocket and further extend towards the TGT homo-
dimer interface, which was addressed and modulated with
“spiking” ligands.[12]
[a] Dr. L. D. Movsisyan,+ Dr. E. Schꢀfer,+ Dr. A. Schwab, T. Rossolini,
Prof. Dr. F. Diederich
Laboratorium fꢁr Organische Chemie, ETH Zurich
Vladimir-Prelog-Weg 3, HCI, 8093 Zurich (Switzerland)
E-mail: diederich@org.chem.ethz.ch
[b] A. Nguyen, Dr. F. R. Ehrmann, Prof. Dr. A. Heine, Prof. Dr. G. Klebe
Institut fꢁr Pharmazeutische Chemie
Philipps-Universitꢀt Marburg
Marbacher Weg 6, 35032 Marburg (Germany)
E-mail: klebe@mailer.uni-marburg.de
The ribose-33 pocket is large and bowl-shaped, with water-
exposed polar and non-polar residues to recognize the ribose
sugar of the tRNA substrate.[10] Previously introduced hydro-
phobic and hydrophilic substituents at position C(2’) of the lin-
benzoguanine core occupied the ribose-33 pocket loosely, ex-
hibiting in most cases a high degree of conformational flexibili-
ty, as revealed by co-crystal structure analysis and MD simula-
tions.[13] Ligands with single-digit nm activity had been devel-
[c] D. Zimmerli, B. Wagner, H. Daff
F. Hoffmann-La Roche Ltd
Discovery Technologies
Bldg 92, 4070 Basel (Switzerland)
[+] These authors contributed equally to this work.
Supporting information and the ORCID identification number(s) for the au-
thor(s) of this article can be found under:
https://doi.org/10.1002/chem.201801756.
Chem. Eur. J. 2018, 24, 1 – 12
1
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
ing Information, Section S1). Attaching a 5-amino ribofurano-
side at position C(2’) of the lin-benzoguanine core was consid-
ered first in our design of ligand 2 as it mimics the natural
tRNA substrate.[10] The space filling capacity of the C(2’) side
chain was subsequently increased by attaching one acetonide-
protecting group to ribofuranoside ligand 1, or two acetonides
to psicofuranose ligand 3 and fructopyranose ligand 4. The
acetonide motif has been utilized previously to modulate lipo-
philicity, chemical and metabolic stability, as well as selectivity
of bioactive compounds, and we anticipated our designed li-
gands to benefit from these improved properties.[16] For exam-
ple, the antiepileptic drug topiramate, a sulfamate-modified
fructose diacetonide, has fast oral absorption and high meta-
bolic stability, mostly being excreted in urine unchanged.[16] In
fact, a search of the DrugBank database furnished 14 approved
drugs having an acetonide (isopropylidene) motif in their
structure.[17]
Synthesis and characterization
The transformation of N-(dimethylamino)sulfonyl-protected 6-
nitrobenzimidazole-5-carboxylates into lin-benzoguanines is
well established.[11b,13,18] The last step, the cleavage of the (di-
methylamino)sulfonyl group from the imidazole moiety re-
quires strong acidic conditions (2m aqueous HCl/MeOH 1:1,
658C). Under these conditions, the isopropylidene protecting
groups of the carbohydrate moiety would also be removed.
Therefore, in this work, the (dimethylamino)sulfonyl group is
replaced by a tert-butyloxycarbonyl (Boc) group which can be
hydrolyzed under basic conditions (1m NaOH in MeOH at
408C). The required isopropylidenated carbohydrates, ribofura-
nose 5r,[19] fructopyranose 5 f,[20] and psicofuranose 5p[20a,21] (r,
f, p are indices for the three different sugars), all possessing a
primary hydroxyl group, are well known (Scheme 1).
Figure 1. a) Structure of ligands 1–4. b) Active site of Z. mobilis TGT with
preQ1-tRNA substrate[10] (Cprotein gray, CtRNA yellow, 3.20 ꢂ resolution, PDB
code: 1Q2S) aligned with designed ligand 4. One of the acetonide groups
approaches the uracil-33 subpocket. Color code: Cligand maroon, O red, N
dark blue, P orange. The protein surface is represented as gray mesh.
oped, however, the optimal occupancy of the flat ribose-33
pocket had not been accomplished yet, even when bearing
C(2’) substituents that reach up to the ribose-32 subpocket.[13e]
Therefore, we were interested in the elaboration of new func-
tional moieties at position C(2’) of the lin-benzoguanine core
for superior occupancy of the ribose-33 pocket and improved
potency and physicochemical properties.
Recently, we introduced C(4’)-furanoside-functionalized lin-
benzoguanines which benefit from good water solubility,
though have weaker binding affinities, with Kd values in a
range of 217–353 nm.[14] Co-crystal structures revealed that the
decrease in affinity is due to unfavorable displacement of the
water molecules from a five-membered polygonal water clus-
ter solvating the side chains of the catalytic Asp102 and
Asp280 in the ribose-34 pocket. The crystal structure analysis
suggested that ether-protected hydroxy groups on the furano-
sides can potentially occupy the hydrophobic regions of this
pocket and consequently enhance the binding affinity. These
findings motivated us to investigate in this study the binding
of lin-benzoguanine ligands featuring acetonide-protected fur-
anose and pyranose substituents at C(2’) to fill the ribose-33
pocket of TGT.
Ribofuranose 5r and psicofuranose 5p were transformed
with phthalimide under Mitsunobu conditions into 7r (96%)[19]
and 7p (74%), respectively. The sterically more hindered fruc-
topyranose 5 f was triflated to 6 f[22] which upon substitution
with potassium phthalimide gave 7 f. Hydrazinolysis of 7r, 7 f,
and 7p at 658C afforded the desired primary amines 8r,[19] 8 f,
and 8p (50–73% yield).
Boc protection of benzimidazole 9[11b] and flash chromatog-
raphy gave the crystalline, regioisomeric carbamates 10a
(47%) and 10b (19%) (Scheme 2). They were separately bromi-
nated with NBS in THF affording 11 a and 11 b in 36 and 19%
yield, respectively. Nucleophilic substitution of 11 a with 8r
and of 11 b with 8 f or 8p gave the 2’-(glycosylamino)benzimi-
dazoles 12r (49%), 12 f (61%), and 12p (74%). Reduction of
the nitro group in these coupling products with zinc dust in
AcOH/H2O 5:1 provided the benzimidazolediamines 13r, 13 f,
and 13p (73–82%). The milder and less toxic method of Fabis
and co-workers[22] was implemented for the preparation of the
protected lin-benzoguanines 14 from 13. Thus, treatment of
13r, 13 f, and 13p with ethoxycarbonyl isothiocyanate gave
the corresponding N-(ethoxycarbonyl)thioureas, which were
transformed with hexamethyldisilazane into the corresponding
N-(ethoxycarbonyl)amidines. The latter, upon the dehydrating
Results and Discussion
Ligand design
The lin-benzoguanine scaffold in all ligands 1–4 (Figure 1a)
binds to the nucleobase exchange pocket similar to the natural
[15]
substrates preQ1 and preQ1-tRNA[10] (Figure 1b and Support-
&
&
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
2
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Scheme 1. Synthesis of the aminosugar precursors. Reagents and conditions: a) Tf2O, pyridine, CH2Cl2; 86%; b) PPh3, phthalimide, DIAD, THF; 96% (7r),
94% (7 f), 74% (7p); c) potassium phthalimide, DMF, 1208C; 94%; d) 64% hydrazine hydrate, EtOH, 658C; 50% (8r), 73% (8 f), 71% (8p). Tf=trifluoromethyl-
sulfonyl; DIAD=diisopropyl azodicarboxylate; THF=tetrahydrofuran; DMF=dimethylformamide.
Scheme 2. a) N,N-Diisopropylethylamine, Boc2O, 4-(dimethylamino)pyridine, DMF; 47% (10a), 19% (10b); b) NBS, THF; 36% (11 a), 19% (11 b); c) Et3N, EtOAc/
EtOH; 49% (12r), 61% (12 f), 74% (12p); d) Zn dust, AcOH/H2O 5:1; 82% (13r) 73% (13 f), 78% (13p); e) ethoxycarbonyl isothiocyanate, hexamethyldisila-
zane, 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide, MeCN; 57% of 14r, 48% of 14 f, 85% of 14p. f) 1m NaOH, MeOH, 408C, HPLC with NH4OAc in the
eluent; 30% (1·0.8AcOH), 65% (3·0.7AcOH), 77% (4·0.8AcOH); g) 1m aq. HCl, MeOH, 408C, HPLC with HFBA in the eluent; 36% of 2·2.8 HFBA. Boc=tert-
butyloxycarbonyl; NBS=N-bromosuccinimide; HFBA=heptafluorobutyric acid.
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
3
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
action of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide gave
the desired 14r (57%), 14 f (48%), and 14p (85%). The treat-
ment of these protected lin-benzoguanines with 1m NaOH in
MeOH cleaved both carbamate groups. HPLC with NH4OAc in
the eluent gave 30% of 1·0.8AcOH, 66% of 3·0.7AcOH, and
77% of 4·0.8AcOH. Finally, the isopropylidene group of 1 was
cleaved with 1m aq. HCl in MeOH at 408C leaving the glycosi-
dic moiety intact. HPLC with heptafluorobutyric acid (HFBA) in
the eluent afforded 2·2.8 HFBA in 36% yield.
Broad 13C NMR signals of the N(1’)-unprotected lin-benzo-
guanines 1–4, especially for C(3’a) and C(9’a) revealed a slow
equilibrium on the NMR time scale of the 1H and 3H tauto-
mers. Shown in Scheme 1 are the 1H tautomers since the li-
gands bind in this form to TGT.
formation, Section S4. The ribose-appended ligand 2 has the
weakest affinity with a Kd value of 1120 nm, while its acetonide
1 is the strongest binder (Kd =14 nm) amounting to an activity
difference by a factor of 80. Psicofuranose diacetonide 3 also
binds in the lower double-digit nm range (Kd =32 nm). Moving
to bisacetonide ligand 4 with the larger fructopyranose sub-
stituent, the affinity to TGT is reduced by a factor of 23 com-
pared to ligand 1 (Kd =321 nm).
Advantageously, compounds 1–4 are readily water-soluble,
and by introducing acetonide-protecting groups, physico-
chemical properties are significantly improved. The calculated
clogD and clogP values (using ACD/Labs software)[24] reflect
these changes. The experimentally measured logD values for 1
and 3 provide certain confidence in the calculated values
(Table 1), though clogD of compound 1 underestimates logD
by over 40%. Ribose 2 lacking the acetal motif is the most hy-
drophilic compound in the series (clogD=ꢀ1.05, clogP=
ꢀ0.87), while acetonides 1, 3, and 4 have increased lipophilic
character, that is, they show more favorable partitioning
(Table 1). The clogD/logD values become more positive upon
changing from mono- to bisacetonide ligands. Extending the
ring size from furanose 3 (clogD=2.52) to pyranose 4 (clogD=
3.33) increases lipophilicity further, which is not unexpected in
view of the values measured for tetrahydrofuran (logD=0.46)
and tetrahydropyran (logD=0.82).[25]
The conformation of the sugars was analyzed according to
the assignments for 6-amino-6-deoxy-hexopyranosides (for
more details, see Section S2 in the Supporting Information).[23]
The vicinal 1H coupling constants J(4,5a)=8.5–8.9 Hz and
J(4,5b)=5.4–5.7 Hz of the ribose in precursors 12r, 13r, and
14r as well as J(5,6a)=8.9–9.1 Hz and J(5,6b)=4.8–6.1 Hz of
the psicofuranose derivatives 12p, 13p, 14p, and 3·0.7AcOH
evidence a ca. 1:1 equilibrium of the sugar gt and tg rotamers
(see Section S3). The broad triplet of HꢀC(4) of 1·0.8AcOH (J=
7.5 Hz) and the quartet of HꢀC(4) of 2·2.8 HFBA (J=7.5 Hz)
suggest also a ca. 1:1 equilibrium of the gt and tg rotamers, as
J(4,5a) ꢁ8.5 Hz and J(4,5b)=5.5–6.5 Hz may contribute to the
mean observed couplings. Thus, protonation of 1–4, which ac-
cording to previous pKa measurements[18] most probably
occurs in aqueous solution, different from protein-bound
state,[13c] on the aminoimidazole moiety, has a negligible influ-
ence upon this rotameric equilibrium.
Table 1 shows the thermodynamic profiles of the TGT com-
plexes with ligands 1–4.[13c,d] Binding is enthalpically driven,
and across the series enthalpy/entropy compensation is ob-
served.[26] The weakest binder 2 features the lowest enthalpic
gain but also the smallest entropic compensation. The highest
enthalpic gains are measured for the two strongest binders 1
and 3, accompanied by a correspondingly larger entropic com-
pensation. The ribose substituent in ligand 2 has the poorest
fit to the spacious 33-pocket (see the co-crystal structures
below) and features the largest desolvation costs upon bind-
ing.
Binding activity and physicochemical properties
The dissociation constants Kd of ligands 1–4 were measured by
isothermal titration calorimetry (ITC, Table 1).[13c,d] The details of
thermodynamic measurements are provided in Supporting In-
Table 1. TGT Binding affinities and thermodynamic profiles of ligands 1–4. The Kd, clogD7.4, clogP, and measured logD values of ligands 1–4 are also shown
together with microsomal stability measurements on human, rat, and mouse microsomes.
Ligand
Kd
[nm]
DG0
[kJmolꢀ1
Buffer
DH0
[kJmolꢀ1
ꢀTDS0
clogD
(logD)
clogP
Clint
]
]
[kJmolꢀ1
]
[mLminꢀ1 mgꢀ1
]
1
14ꢂ7
ꢀ45.0ꢂ0.7
ꢀ34.0ꢂ0.03
ꢀ43.0ꢂ0.4
ꢀ37.1ꢂ1.5
HEPES
Tricin
TRIS
Ion corrected
HEPES
Tricin
TRIS
Ion corrected
HEPES
Tricin
TRIS
Ion corrected
HEPES
Tricin
TRIS
ꢀ64.2ꢂ1.9
ꢀ53.0ꢂ2.3
ꢀ42.3ꢂ0.4
ꢀ79.5ꢂ1.5
ꢀ41.7ꢂ1.0
ꢀ33.9ꢂ0.2
ꢀ17.8ꢂ0.6
ꢀ46.9ꢂ0.6
ꢀ55.9ꢂ3.2
ꢀ45.5ꢂ3.2
ꢀ34.6ꢂ0.6
ꢀ71.1ꢂ2.3
ꢀ41.7ꢂ1.0
ꢀ33.9ꢂ0.2
ꢀ17.8ꢂ0.6
ꢀ60.6ꢂ0.6
19.6ꢂ3.8
7.4ꢂ3.3
0.63
(1.08)
0.72
<10
ꢀ2.4ꢂ0.9
34.5ꢂ2.7
3.6ꢂ1.7
2
3
4
1120ꢂ110
32ꢂ13
ꢀ1.05
ꢀ0.87
2.61
–
ꢀ0.6ꢂ1.4
ꢀ8.7ꢂ1.7
12.9ꢂ1.6
12.5ꢂ1.6
2.9ꢂ0.3
2.52
(2.04)
<10
<10
ꢀ9.9ꢂ0.9
28.1ꢂ0.9
3.2ꢂ1.0
321ꢂ193
3.33
3.47
ꢀ0.2ꢂ0.5
ꢀ18.4ꢂ0.9
23.6ꢂ0.8
Ion corrected
&
&
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
4
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
The improved water solubility of lin-benzoguanines with
acetonide-protected carbohydrate substituents provides a sub-
stantial benefit for further optimization of the ligand class. Fur-
thermore, the metabolic stability (intrinsic clearance, Clint) of
acetonides 1, 3, and 4 was determined in a microsomal stabili-
ty assay with human, mouse, and rat microsomes (Table 1).[27]
All three ligands show low clearance (<10 mLminꢀ1 mgꢀ1) and
therefore promise long duration of action.
Crystallographic analysis
High resolution (between 1.17 and 1.36 ꢂ) X-ray crystal struc-
tures of all four ligands in complex with Z. mobilis TGT were
determined. In all complexes, the lin-benzoguanine core is ac-
commodated in the guanine/preQ1-binding sub-site. In the
complexes of 1–3, the core is sandwiched between the side
chains of Met260 and Tyr106 (Figure S2c in the Supporting In-
formation; for the particular case of the complex with 4, see
below). Hydrogen bonds are established between the lin-ben-
zoguanine core and the side chains of Asp102, Asp156,
Gln203, the backbone NH of Gly230, and the backbone car-
bonyl groups of Leu231 and Ala232, a pattern typical for the
lin-benzoguanine scaffold bound to TGT (Figure S2a in the
Supporting Information).[11–14,18] The polar ribose-34 pocket is
solvated by a conserved water cluster arranged as a five-mem-
bered water cluster (W1*-W5*, labeled with asterisk and col-
ored red, Figure S2b). W1* interacts with the Asp102 carboxyl-
ate and forms a hydrogen bond to W2*, which itself hydrates
the Asp280 side chain. W3* also solvates the Asp280 side
chain and additionally forms hydrogen bonds to W4* and W5*.
In this cluster, only W4* makes a contact to the lin-benzogua-
nine imidazole nitrogen N(3’). In the case of ligand 3, this
water (W4*) is replaced by a glycerol molecule from the cryo
buffer (Figure S2b) which can be treated as surrogate for
water molecules[28] and is expected to have no influence on
the binding of the carbohydrate substituents in the ribose-33
pocket.
Figure 2. Binding modes of ribofuranosides 1 (Ccyan, 1.17 ꢂ resolution, PDB
code: 5JSV, panels a) and b)) and 2 (Cgreen, 1.36 ꢂ resolution, PDB co-
de: 6FMN, panels c) and d)) in the active site of TGT Z. mobilis. In panels a)
and c), the 2 Fo-Fc electron density maps are shown (blue, contour level
1.0s). Water molecules are depicted as spheres and colored in the polycyclic
cluster in red, while the second water cluster molecules are shown in blue.
In panels b) and d), B factors (ꢂ2) of the water molecules are displayed in
green and the hydrogen bond distances are given in ꢂ. The protein surface
is colored in light grey, hydrogen bonds as black dotted lines. The red
dotted line indicates a repulsive interaction. Color code: Cprot gray, Clig cyan
or green, O red, N dark blue, S yellow.
within van der Waals distances (3.6–4.7 ꢂ). The [5+5+4] cluster
is additionally stabilized through hydrogen bonding to the
side chains of Arg286, Cys281, and Gln292 (shown in Figure S2)
as well as via networking with the first solvation layer, connect-
ing to the bulk phase. The second water cluster is formed at
the lower left side of the pocket (Uracil-33 subpocket), close to
the Gly234/Glu235/Gly236 loop region (Figure 2) and consists
of five water molecules (W1’-W5’, labeled with prime symbol
and colored blue) connected in a chain-like fashion. The mole-
cules W2’, W3’, and W4’ solvate the Asp267 and Glu235 car-
boxylates, respectively. Water molecule W5’ is engaged in hy-
drogen bonding with the backbone NH groups of the distort-
ed Gly234/Gly236 loop. The deeper buried W1’ bridges W2’ to
W3 in the [5+5+4] cluster. Additionally, it solvates the Gly261
carbonyl group and an oxygen atom of the 2,3-acetonide of
the ligand.
In the crystal structure of TGT in complex with 1, the ribo-
side residue has well-defined electron density with the average
B value (22 ꢂ2) twice exceeding the lin-benzoguanine core
(11 ꢂ2, Figures 2a,b). The ribofuranoside ring adopts an EO en-
velope (see Sections S3) and the energetically favorable tg con-
formation about the C(4)ꢀC(5) bond (gt/tg ca. 1:1 in uncom-
plexed 1, vide supra),[29] exposing the MeO group towards the
ribose-34 pocket. The acetonide moiety optimally occupies the
ribose-33 pocket. The ribose Oring atom gets hydrated from
W5* which is connected to a water cluster in the ribose-34
pocket. High resolution structural data show two distinctive
water clusters in the ribose-33 pocket: One of them consists of
9 water molecules forming a tricyclic cluster featuring two five-
and one four-membered rings ([5+5+4] cluster) on the lower
right side of the pocket (labelled W1ꢀW9 and colored red, Fig-
ures 2a,b). The pentagonal structure of clustered water is well
known to be energetically favorable.[30] This water network
wraps around the Leu283 residue, making van der Waals con-
tacts (3.6–5.0 ꢂ) with its isobutyl side chain. One pentagon of
the cluster faces toward the methyl group of the 2,3-acetonide
In the active site of TGT complexed with 2, the 2,3-unpro-
tected riboside shows a diffuse electron density (Figure 2c). At-
tempts to assign the surplus electron density to different con-
formations of the ribofuranosyl unit did not afford reasonable
structural solutions. We restrict the structural analysis to one,
most populated conformation, though we do not exclude that
more conformations can contribute to the observed electron
density. The average temperature factor B of the ribofuranosyl
residue (33 ꢂ2) is significantly higher than the B values of the
core (10 ꢂ2) and the protected ribofuranosyl unit in ligand 1.
1
The ribofuranosyl ring in 2 adopts an E envelope configura-
tion, and the conformation about the C(4)ꢀC(5) bond is tg,
which orients the OMe group to the ribose-33 pocket (see Sec-
tion S3 in the Supporting Information). The OMe methyl group
is not resolved in the electron density map. Two water clusters
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
5
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
are observed in the ribose-33 pocket, similarly to TGT in com-
plex with 1. The chain-like 5-membered cluster on the left side
of the pocket (phosphate-32 binding subpocket) essentially
stays the same, when compared to TGT in complex with 1, sol-
vating Asp267, Gly234, Glu235, Gly236, and Gly261. The integ-
rity of the polycyclic [5+5+4] cluster on the right side is par-
tially destroyed and W6, W7, and W9 waters are not observed
in the high-resolution electron density map (Figure 2c). The
cluster solvates Arg286 and Cys281 residues and continues the
hydrogen-bonding network with the external water shell (not
shown in Figure 2). The C(2) hydroxyl group is solvated by
water molecules W1 and W2 while the OMe oxygen forms a
hydrogen bond with water W1’ (Figures 2c,d).
In the TGT·3 crystal structure, a well-defined electron density
of the ligand shows the psicofuranose ring in an EO envelope
and the bond about C(5)ꢀC(6) in the tg conformation (Fig-
ures 3a,b, Section S3 in the Supporting Information). The 1,2-
acetonide moiety points towards the ribose-33 pocket, while
the 3,4-acetonide reaches to the rim of the ribose-34 pocket.
the cluster solvates the side chains of Arg286, Cys281, and
Gln292 (not shown) and makes contacts with the ribose-34
pocket water W5* and the W1’ water of the chain-like cluster
from the other side of the ribose-33 pocket. Further hydrogen-
bonding interactions occur with the external solvent shell of
the protein. In contrast to the TGT·1 and TGT·2 complexes, the
sugar residue of ligand 3 interacts neither with the tricyclic nor
the water clusters in the ribose-33 pocket.
In the crystal structure of TGT in complex with 4, the fructo-
2
pyranose ring adopts a flattened C5 chair conformation. The
C(1)ꢀN bond is synclinal to the C(2)ꢀOring bond (see Section S3,
and Figures 3c,d). The average B values for the lin-benzogua-
nine core (14 ꢂ2) and pyranosyl residue (34 ꢂ2) are comparable
to the values in the complex of ligand 3. Although only one
out of four acetonide methyl groups is resolved in the electron
density map, reconstruction of the 2,3-acetonide shows signifi-
cant steric repulsion with the Tyr106 side chain. Tyr106 mini-
mizes this repulsion by turning the phenolic ring about 908,
forming new short CH···p (edge-to-face) contacts (3.4–4.0 ꢂ)
with the lin-benzoguanine core (Figure 4a). The 4,5-acetonide
moiety is solvated by water W5* (Figure 3c,d).
Striking differences are observed in the solvation pattern of
the ribose-33 pocket in the complex of 4, as compared to
those of 1 and 3. The tricyclic [5+5+4] water cluster is com-
pletely removed from the pocket. The 4,5-acetonide residue is
not completely resolved in the density map, however, the re-
Figure 3. Binding modes of psicofuranose 3 (Cblue, 1.22 ꢂ resolution, PDB
code: 5JSW) and fructopyranose 4 (Cmaroon, 1.36 ꢂ resolution, PDB code: 6F-
PU) in the active site of TGT Z. mobilis. In panels a) and c), the 2 Fo-Fc elec-
tron density maps (blue) are shown (at contour level 1.0s). Water molecules
are depicted as spheres and colored in the polycyclic cluster by red, while
the waters in the water cluster are shown in blue. In panels b) and d), B fac-
tors (ꢂ2) of the water molecules are shown in green and the hydrogen bond
distances are given in ꢂ. The protein surface is colored in light gray, hydro-
gen bonds as black dotted lines. The red dotted line indicates a repulsive in-
teraction. Color code: O red, N dark blue, S yellow.
The average B value of the lin-benzoguanine core (12 ꢂ2) is
similar to the TGT complex of 1 and 2. The psicofuranose sub-
stituent of 3 has an average B value (35 ꢂ2) comparable to the
average B value of the ribose moiety in the TGT·2 complex. Psi-
cofuranose 3 establishes contacts within van der Waals distan-
ces in the ribose-33 pocket. Two distinguished water clusters
are again resolved clearly in the density map (Figure 3a). All
nine water molecules of the [5+5+4] tricyclic cluster have the
same arrangement as the one in the TGT·1 complex. This clus-
ter forms a water shell between the hydrophobic side chain of
Leu283 (vdW distances=3.5–4.0 ꢂ) and a methyl group of the
1,2-acetonide moiety (vdW distance=3.4–4.3 ꢂ). Additionally,
Figure 4. a) Comparison of the active site of Z. mobilis TGT complexed with
ligands 3 (Cblue, 1.22 ꢂ resolution, PDB code: 5JSW) and 4 (Cmaroon, 1.36 ꢂ res-
olution, PDB code: 6FPU). b) Comparison of water molecules in the ribose-
33 pocket of Z. mobilis TGT bound to ligands 1 (Ccyan, 1.17 ꢂ resolution, PDB
code: 5JSV) 2 (Cgreen, 1.36 ꢂ resolution, PDB code: 6FMN), 3 (Cblue, 1.22 ꢂ reso-
lution, PDB code: 5JSW), and 4 (Cmaroon, 1.36 ꢂ resolution, PDB code: 6FPU).
Water molecules are represented as spheres and colored according to the
corresponding ligand: 1 cyan, 2 green, 3 maroon, 4 blue. The protein surfa-
ce (PDB code: 5JSV) is colored in light gray, and the ribose-33 pocket is high-
lighted in blue. Color code: O red, N dark blue, S yellow.
&
&
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
6
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
construction (modeling) suggests that one of its methyl
groups forms van der Waals contacts (3.4–4.1 ꢂ) to the side
chain of Leu283, which is resolved in two equally populated
conformations with one of them shifted towards the carbohy-
drate. This dispersion interaction between methyl group and
the Leu283 side chain leaves not sufficient space for solvation
by a tricyclic water cluster, which completely disappears. Addi-
tionally, the Cys281 side chain adopts a conformation enabling
weak interactions with W5* and Gly261, which were not ob-
served in the complexes of TGT with 1–3. The water molecules
of the second, 5-water cluster in the complex of 4 are resolved
in the pocket and four of them, W1’-W4’, adopt the same posi-
tions as those seen for the complexes with 1–3. W5’ in the
density map gives rise to a diffuse distribution and two posi-
tions are assigned to it (Figure 3c and Figure 4b).
ligand and protein merge at the position where the tricyclic
cluster is found in the complexes of 1 and 3, and partially 2.
All nine water molecules in the [5+5+4] cluster in TGT·1
and TGT·3 are arranged at optimal hydrogen bond distances of
2.6–3.4 ꢂ. Properties and structure of ordered water molecules
on hydrophobic surfaces are pivotal in protein function, molec-
ular recognition, and development of drugs in medicinal
chemistry.[32] In one recent example, some of us showed that
small alkyl substituents can modulate to a large extent the
structure and energetics of the water bound on the surface of
the protein thermolysin.[28,33] The complexes presented here
highlight the ability of large, uncharged molecules, as sugars
are, to stabilize water molecules in a form of clusters inside a
flat, solvent exposed pocket with mixed hydrophobic and hy-
drophilic character. The [5+5+4] tricyclic and five-membered
chain-like water clusters in the ribose-33 pocket are connected
via water W1’ but also seem to exist independently from each
other. This hypothesis is supported by the fact that total dis-
placement of the [5+5+4] cluster in the TGT·4 complex does
not affect the position of the water cluster. In the majority of
crystal structures of ligands targeting the ribose-33 pocket of
Z. mobilis TGT, the cluster is fully conserved,[13] and only in one
crystal structure (PDB code: 4PUJ)[13c] the polycyclic water clus-
ter is conserved partially. In this case, the C(2’)-morpholinoethyl
substituent of the lin-benzoguanine instead of occupying the
ribose-33 pocket is located outside the pocket (Figure S4). The
formation of the [5+5+4] water cluster in the ribose-33 pocket
could be a result of the interruption of the first layer of protein
solvent by the sugar moiety which facilitates directional bond-
ing of water molecules around Leu283. Water molecules are
further immobilized when they are sandwiched between two
aliphatic moieties of acetonides and the isobutyl side chain of
Leu283 in the TGT·1 and TGT·3 complexes. In contrast, water
molecules of the chain-like clusters are stabilized primarily by
polar interactions with the charged Glu235 and Asp267 side
chains and backbone NH group of Gly234.
High resolution crystal structures allow to correlate the ex-
perimental thermodynamic parameters with structural features
revealed by crystallography. In all four ligands, the carbohy-
drate substituents at position C(2’) of the lin-benzoguanine
core occupy only a single (TGT·1, TGT·3 and TGT·4) or a domi-
nant (TGT·2) conformation in the ribose-33 pocket. This was
not the case with previously developed ligands functionalized
at position C(2’) with aromatic rings or flatter heteroalicycles
(such as morpholine rings), which show very high residual mo-
bility in the pocket.[13]
The highly defined binding poses of 1–4 reflect superior
space-filling properties in particular of the acetonide-protected
sugars, which benefit from their unique positioning with low
mobility in the ribose-33 pocket. This pocket clearly has spatial
boundaries. On the one hand, the diffuse electron density and
high B values for the 2,3-unprotected ribosyl moiety in TGT·2
suggest that this smaller sugar provides only a suboptimal oc-
cupation of the pocket. On the other extreme, the bis-aceto-
nide-protected fructopyranosyl moiety occupies the 33-pocket
beyond its limits, provoking structural changes in the guanine-
binding site of the enzyme (shift of Tyr106) to enable accom-
modation. With 4, the limits for space filling of the ribose-33
pocket have been reached. For the quantitative description of
pocket occupancy, buried solvent accessible surface areas
(SASA) were calculated for complexes using the program
MOLOC.[31] For comparability reasons, geometries of the li-
gands in the TGT·1 and TGT·4 crystal structures were comple-
mented by modelling, and for the MeO group in ligand 1 the
position of the methyl group was added. Calculated buried
SASAs for TGT·1 (627 ꢂ2), TGT·2 (535 ꢂ2), TGT·3 (737 ꢂ2), and
TGT·4 (780 ꢂ2) complexes confirm the gradual increase of
ligand space filling, although, the difference in SASA between
the last two complexes is small. Figure S3 in the Supporting In-
formation shows the solvent accessible surfaces of the four
complexes. All water molecules of the tricyclic cluster in
TGT·1–3 do neither merge with the solvent-accessible surface
of the ligand nor the protein. This suggests that the ordered
[5+5+4] water cluster is formed between the sugar and pro-
tein hydrophobic surfaces to complete the optimal space fill-
ing of the 33-pocket. The situation is dramatically different for
the TGT·4 complex, where the solvent accessible surfaces of
Structural features seen in the crystal structures correlate
well with the measured thermodynamic profiles (Table 1). In
the previously studied C(2’)-substituted lin-benzoguanines and
lin-benzohypoxanthines, protein binding was driven by enthal-
py and a clear enthalpy/entropy compensation was observ-
ed.[13d] Similarly, for ligands 1–4, the Gibbs binding free energy
has strong exothermic and unfavorable entropic contributions
and enthalpy/entropy compensation takes place within the
series. A large drop in the enthalpy contribution is observed
for the TGT·2 complex (DDH1!2 =32.6 kJmolꢀ1) with respect to
TGT·1. This can be attributed to reduced short intermolecular
contacts established by the smaller guest bound in different
conformations as reflected by the diffuse electron density, and
in particular the unfavorable desolvation of the 2,3-deprotect-
ed riboside. The bound ribose forms three hydrogen bonds to
W1, W2, and W1’ of the water clusters (Figure 2c,d), nonethe-
less, these are not sufficient for compensating the desolvation
enthalpy of the free ligand. The favorable entropy difference
ꢀDTDS1!2 =ꢀ21.6 kJmolꢀ1 between TGT·2 and TGT·1 com-
plexes can be rationalized by higher residual mobility of ligand
2 in the pocket. Also, the polycyclic water cluster in the com-
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
7
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
plex of 2 is not completely established, which could contribute
to the changes in thermodynamic profile.
The proper filling of the ribose-33 cavity seen in the crystal
structures of 1 and 3 would clearly be difficult to realize with
substituents other than the acetonide-protected sugars report-
ed herein. Except for isolated cases,[16] acetonide-protected
sugars have not been investigated broadly for drug develop-
ment, and this study suggests that they have the potential to
become privileged building blocks for the filling of pockets
that are not complementary to (hetero)aromatic rings or het-
eroalicycles in many medicinal chemistry projects.
The thermodynamic quantities for acetonides 1 and 3 are
quite similar, with the highest enthalpic driving forces and the
highest unfavorable entropic terms. Since the two water clus-
ters are nearly identical in the two complexes, these small dif-
ferences (Table 1) in DH and TDS presumably reflect the better
fit and interactions of 1, with a twofold lower Kd value, rather
than differences in solvation.
The difference between the Gibbs binding free energies of
the two bis-acetonides 3 and 4 is DDG3!4 =5.9 kJmolꢀ1. It is
Experimental Section
tempting to speculate that the enthalpic drop (DDH3!4
=
10.5 kJmolꢀ1) encountered by 4 could be associated with the
complete displacement of the ordered tricyclic water cluster
from the ribose-33 pocket, which also could be at the origin of
The synthesis of 1 and 2 starting with the condensation of 8r with
11 a (Scheme 2) is described here. For general remarks and the syn-
thesis and characterization of all other products, see the SI. Images
of the structures were prepared using PyMol software.[36]
the more favorable entropic contribution (ꢀDTDS3!4
=
ꢀ4.5 kJmolꢀ1).[34] The displacement of this water cluster by the
dispersion interactions with Leu283 might be energetically less
favorable. The turning of Tyr106 from a p–p-stacking to an
edge-to-face geometry with the lin-benzoguanine core of 4
should provide a similar energetic contribution.[35]
Methyl 5-{1-[(tert-butoxy)carbonyl]-6-(methoxycarbonyl)-5-nitro-
1H-benzo[d]imidazol-2-yl}-5-deoxy-2,3-O-isopropylidene-b-d-ri-
bofuranoside (12r): A solution of 8r (360 mg, 0.91 mmol) in
EtOAc/EtOH 1.8:1 (17 mL) was treated with 11a (203 mg, 1 mmol)
and Et3N (1 mL) and stirred for 3 h at 408C. Evaporation and flash
column chromatography FC (SiO2; hexane/EtOAc 3:2) gave 12r
(232 mg, 49%) as a yellow solid. Rf =0.40 (SiO2; hexane/EtOAc 3:2,
UV 254 nm); m.p. 998C; ½aꢃ2D5 =ꢀ23.8 (c=0.1 in CHCl3); 1H NMR
(400 MHz, CDCl3; contains 2 equiv of EtOAc): d=1.32 and 1.49 (2 s,
6H; CMe2), 1.73 (s, 9H; CMe3), 3.41 (s, 3H; MeO-C(1)), 3.64 (ddd, J=
13.4, 8.5, 4.8 Hz, 1H; Ha-C(5)), 3.85 (ddd, J=13.6, 6.8, 5.6 Hz, 1H;
Hb-C(5)), 3.89 (s, 3H; CO2Me), 4.51 (ddd, J=8.5, 5.5, 1.0 Hz, 1H; H-
C(4)), 4.66 (d, J=5.9 Hz, 1H; H-C(2)), 4.71 (dd, J=6.0, 1.0 Hz, 1H; H-
C(3)), 5.04 (s, 1H; H-C(1)), 7.73 (s, 1H; H-C(4’)), 7.76 (br. t, J=5.4 Hz,
1H; NH), 8.00 ppm (s, 1H; H-C(7’)); 13C NMR (101 MHz, CDCl3; con-
tains 0.5 equiv of EtOAc): d=25.18 and 26.62 (CMe2), 28.17 (CMe3),
46.29 (C(5)), 53.19 (CO2Me), 55.58 (MeO-C(1)), 82.24 (C(3)), 85.10
(C(4)), 85.54 (C(2)), 88.08 (OCMe3), 110.05 (C(1)), 111.94 (C(7’)),
112.83 (CMe2), 115.46 (C(4’)), 118.74 (C(6’)), 132.76 (C(7’a)), 145.47
(C(5’)), 147.03 (C(3’a)), 156.86 (C(2’)), 166.33 ppm (CO2Me), NC=O
hidden by the noise; IR (ATR): n˜ =3357 (br., w), 2987 (w), 2940 (w),
1728 (s), 1630 (s), 1605 (s), 1524 (m), 1438 (m), 1303 (s), 1208 (m),
Conclusion
We developed a new series of lin-benzoguanine ligands with
sugar substituents to address the ribose-33 pocket of Z. mobilis
TGT. The Kd values and the thermodynamic profile for the bind-
ing of four ligands were measured by ITC. Experimental param-
eters correlate well with structural features determined by X-
ray crystallography, revealing a multi-faceted structure–activity
relationship. Under the applied conditions, ligands 1–4 bind to
TGT exothermally without favorable entropic contribution. The
two acetonide-protected furanoses 1 and 3 have Kd values in
the low, double-digit nanomolar range, thereby showing sub-
stantially higher potency than lin-benzoguanine ligands with
furanoside substituents addressing the ribose-34 pocket.[14]
Structural data particularly highlighted the role of acetonide-
protected sugar moieties in the stabilization of a tricyclic
[5+5+4]-ring water cluster around their hydrophobic surface.
However, the cluster is removed from the pocket by binding
the larger acetonide ligand 4, which extends beyond the size
of the ribose-33 pocket. Solvent-accessible surface area (SASA)
evaluations suggest, that the ordered [5+5+4] water cluster is
formed between the sugar and protein hydrophobic surface to
complete the optimal space filling of the 33-pocket. While X-
ray structural data explain the overall thermodynamic recogni-
tion profile of ligands 1–4 in the active site of Z. mobilis TGT,
the presence of water clusters in the complexes renders a
more detailed characterization of each enthalpic/entropic con-
tribution to the binding events more difficult. The protected
carbohydrate moieties not only enhance binding affinity, but
also greatly improve the physicochemical properties (solubility,
logD, logP, microsomal stability) of lin-benzoguanine ligands.
They add fascinating structural diversity to the family of lin-
benzoguanine ligands, as their acetonide-protecting groups
grant structural rigidness with a unique combination of polar
and apolar areas on the surface of the uncharged molecules.
1105 (s), 1091 (s), 958 (m), 866 (m), 780 cmꢀ1 (m); HR-MALDI-MS:
m/z (%): 545.1854 (23, [M+Na]+, calcd for C23H30N4NaO10
545.1854), 523.2034 (21, [M+H]+, calcd for C23H31N4O10+: 523.2035),
468.1442 (21), 467.1408 (100, [MꢀCH2 =CMe2 +H]+, calcd for
C19H23N4O10+: 467.1049), 423.1510 (42, [MꢀCH2 =CMe2ꢀCO2 +H]+,
calcd for C18H23N4O8+: 423.1510).
+
:
Methyl 5-({5-Amino-1-[(tert-butoxy)carbonyl]-6-(methoxycarbon-
yl)-1H-benzo[d]imidazol-2-yl}amino)-5-deoxy-2,3-O-isopropyli-
dene-b-d-ribofuranoside (13r):
A solution of 12r (480 mg,
0.92 mmol) in AcOH/H2O 5:1 (5.2 mL) was treated with Zn dust
(565 mg, 8.6 mmol), stirred for 2 h at 08C, and filtered over a plug
of Celite (washing with CH2Cl2 and MeOH). Evaporation and FC
(SiO2; cyclohexane/EtOAc/Et3N 1:1:0.01) gave 13r (375 mg, 82%) as
a yellow solid. Rf =0.26 (SiO2; cyclohexane/EtOAc 3:2, UV 254 nm);
m.p. 748C; ½aꢃ2D5 =ꢀ12.8 (c=0.05 in CHCl3); 1H NMR (600 MHz,
CDCl3): d=1.31 and 1.48 (2 s, 6H; CMe2), 1.70 (s, 9H; CMe3), 3.40 (s,
3H; MeO-C(1)), 3.60 (ddd, J=13.4, 8.5, 4.6 Hz, 1H; Ha-C(5)), 3.81
(ddd, J=13.5, 6.8, 5.7 Hz, 1H; Hb-C(5)), 3.85 (s, 3H; CO2Me), 4.50
(ddd, J=8.9, 5.7, 1.0 Hz, 1H; H-C(4)), 4.65 (d, J=6.0 Hz, 1H; H-C(2)),
4.70 (dd, J=5.9, 1.0 Hz, 1H; H-C(3)), 5.02 (s, 1H; H-C(1)), 5.66 (br. s,
2H; NH2), 6.57 (s, 1H; H-(C4’)), 7.74 (br. s, 1H; NH), 8.09 ppm (s,
1H; H-C(7’)); 13C NMR (151 MHz, CDCl3): d=25.21 and 26.62 (CMe2),
28.19 (CMe3), 46.08 (C(5)), 51.54 (CO2Me), 55.48 (MeO-C(1)), 82.35
(C(3)), 85.23 (C(4)), 85.58 (C(2)), 85.79 (OCMe3), 102.52 (C(4’)), 104.11
&
&
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
8
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
(C(6’)), 109.98 (C(1)), 112.70 (CMe2), 115.99 (C(7’)), 123.08 (C(7’a)),
148.79 (C(3’a)), 149.40 (C(5’)), 150.78 (NC=O), 156.75 (C(2’)),
168.93 ppm (CO2Me); IR (ATR): n˜ =3467 (br., w), 3365 (br., w), 2980
(w), 2938 (w), 1721 (s), 1686 (s), 1585 (s), 1463 (m), 1345 (s), 1211
(s), 1145 (s), 1106 (s), 1056 (m), 867 (m), 764 cmꢀ1 (m); HR-MALDI-
MS: m/z (%): 404.1757 (20), 403.1724 (100, [M+H]+, calcd for
C18H23N6O5+: 403.1724).
Methyl 5-[(6-amino-8-oxo-7,8-dihydro-1/3H-imdazo[4,5-g]quina-
zolin-2-yl)amino]-5-deoxy-b-d-ribofuranoside·2.8 HFBA (2): A so-
lution of 1 (62 mg, 0.17 mmol) in MeOH (10 mL) was cooled to
08C, treated with 1m aqueous HCl (1.35 mL, 1.35 mmol), and
stirred for 24 h at 408C. Evaporation and HPLC (Merck LiChrosorb
RP-18, 7 mm, 250ꢃ25 mm; flow rate 10 mLminꢀ1, MeOH/H2O/ hep-
tafluorobutyric acid (HFBA) 2:3:0.005) gave 2 (22 mg, 36%) con-
taining 2.8 equiv of HFBA as a white foam. ½aꢃ2D5 =- 6.0 (c=0.05 in
MS: m/z (%): 494.2327 (6), 493.2293 (24, [M+H]+, calcd for
+
C23H33N4O8
:
493.2293), 438.1699 (5), 437.1666 (25, [MꢀCH2 =
CMe2 +H]+, calcd for C19H25N4O8+: 437.1667), 393.1768 (87, [Mꢀ
CH2 =CMe2ꢀCO2 +H]+, calcd for C18H25N4O6+: 393.1769), 392.1690
(100, [MꢀCH2 =CMe2ꢀCO2]+, calcd for C18H24N4O6+: 392.1690).
1
CHCl3); H NMR (600 MHz, CD3CN): d=3.19 (s, 3H; OMe), 3.52 (dt, J
Methyl ({1-[(tert-butoxy)carbonyl]-6-[(ethoxycarbonyl)amino]-8-
oxo-7,8-dihydro-1H-imidazo[4,5-g]quinazolin-2-yl}amino)-5-
deoxy-2,3-O-isopropylidene-b-d-ribofuranoside (14r): A solution
of 13r (200 mg, 0.41 mmol) in anhydrous MeCN (4 mL) was treated
with ethoxycarbonyl isothiocyanate (76 mL, 0.65 mmol), stirred for
15 min at 238C, treated with hexamethyldisilazane (860 mL,
4.07 mmol) and 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide
free base (280 mL, 1.6 mmol), and stirred for 4.5 h at 238C. Evapora-
tion and FC (SiO2; cyclohexane/EtOAc 3:7) gave 14r (132 mg, 57%)
as an off-white solid. Rf =0.46 (SiO2; CH2Cl2/MeOH 9:1, UV 254 nm);
m.p. 2908C; ½aꢃ2D5 =ꢀ13.0 (c=0.1 in CHCl3); 1H NMR (600 MHz,
CDCl3): d=1.31 (t, J=7.2 Hz, 3H; OCH2Me), 1.32 and 1.48 (2 s, 6H;
CMe2), 1.75 (s, 9H; CMe3), 3.42 (s, 3H; OMe), 3.65 (ddd, J=13.6, 8.6,
4.6 Hz, 1H; Ha-C(5)), 3.87 (ddd, J=13.6, 6.8, 5.6 Hz, 1H; Hb-C(5)),
4.29 (q, J=7.2 Hz, 2H; OCH2Me), 4.54 (ddd, J=8.8, 5.4 Hz, 0.5 Hz,
1H; H-C(4)), 4.67 (d, J=5.9 Hz, 1H; H-C(2)), 4.72 (dd, J=5.9, 0.5 Hz,
1H; H-C(3)), 5.04 (s, 1H; H-C(1)), 7.33 (br. s, 1H; H-C(4’)), 7.89 (br. s,
1H; HN-C(2’)), 8.36 (s, 1H; H-C(9’)), 11.20 ppm (br. s, 1H; NH);
13C NMR (151 MHz, CDCl3): d=14.36 (OCH2Me), 25.20 and 26.62
(CMe2), 28.29 (CMe3), 46.18 (C(5)), 55.55 (OMe), 63.09 (OCH2Me),
82.30 (C(3)), 85.15 (C(4)), 85.57 (C(2)), 87.07 (OCMe3), 110.02 (C(1)),
111.23 (br., C(4’,9’)), 112.76 (CMe2), 113.26 (C(8’a)), 129.70 (C(4’a)),
144.79 (N(1)C=O), 146.37 (C(9’a)), 149.46 (C(3’a)), 150.73 (C(6’)),
154.27 (CO2Et), 156.84 (C(2’)), 161.49 ppm (C(8’)); IR (ATR): n˜ =3351
(br., w), 2985 (w), 2935 (w), 1730 (m), 1682 (m), 1559 (s), 1438 (s),
1346 (s), 1232 (s), 1186 (s), 1154 (s), 1096 (s), 868 (m), 765 (m),
704 cmꢀ1 (m); HR-ESI-MS: m/z (%): 576.2493 (25), 575.2459 (92,
ꢁ13.9, 5.1 Hz, 1H; Ha-C(5)), 3.75 (br. dt, J=13.5, 4.3 Hz, 1H; Hb-
C(5)), 3.90 (d, J=4.5 Hz, 1H; H-C(2)), 4.07 (br. q, J ꢁ5.2 Hz, 1H; H-
C(4)), 4.11 (br. t, J ꢁ4.9 Hz, 1H; H-C(3)), 4.75 (s, 1H; H-C(1)), 7.24 (s,
1H; H-C(4’)), 7.92 (s, 1H; H-C(9’)), 9.52 ppm (br. s, 1H; NH); 13C NMR
(151 MHz, CD3CN): d=47.05 (C(5)), 55.81 (OMe), 72.86 (C(3)), 75.61
(C(2)), 82.02 (C(4)), 99.59 (C(4’)), 109.91 (C(1)), 110.15 (C(9’)), 112.59
(C(8’a)), 128.83 (C(9’a)), 137.02 and 137.07 (C(3’a,4’a)), 153.10 (C(6’)),
154.52 (C(2’)), 160.42 ppm (C(8’)); IR (ATR): n˜ =3600-2200 (w), 2970
(w), 1680 (s), 1650 (s), 1477 (m), 1407 (m), 1335 (s), 1209 (s), 1115
(s), 1080 (m), 965 (m), 929 (m), 716 cmꢀ1 (s); HR-ESI-MS: m/z (%):
364.1445 (16), 363.1412 (100, [M+H]+, calcd for C15H19N6O5
:
+
363.1411).
6-[(6-Amino-8-oxo-7,8-dihydro-1/3H-imidazo[4,5-g]quinazolin-2-
yl)amino]-6-deoxy-1,2:3,4-di-O-isopropylidene-b-d-psicofura-
nose·0.7 acetic acid (3): A solution of 14p (132 mg, 0.21 mmol) in
MeOH (25 mL) was treated with 1m NaOH (0.41 mL), stirred for
31 h at 408C, and evaporated. HPLC (Reprosil Chiral NR, 8 mm,
250ꢃ50 mm; flow rate 35 mLminꢀ1, EtOH (0.01 molLꢀ1 NH4OAc)/
heptane 2:3) gave 3 (66 mg, 65%) containing 0.7 equiv of AcOH as
1
a white foam. M.p. 2788C; ½aꢃ2D5 =ꢀ25.1 (c=0.1 in MeOH); H NMR
(600 MHz, (CD3)2SO): d=1.25 and 1.35 (2 s, 6H; CMe2), 1.34 and
1.40 (2 s, 6H; CMe2), 1.85 (s, 2.1H; Me of AcOꢀ), 3.41 (dt, J=13.4,
5.3 Hz, 1H; Ha-C(6)), 3.47 (ddd, J=13.4, 8.8, 5.8 Hz, 1H; Hb-C(6)),
3.94 (d, J=9.6 Hz, 1H; Ha-C(1)), 4.16 (d, J=9.7 Hz, 1H; Hb-C(1)),
4.23 (dd, J=9.1, 6.1 Hz, 1H; H-C(5)), 4.68 (d, J=5.8 Hz, 1H; H-C(3)),
4.84 (d, J=5.6 Hz, 1H; H-C(4)), 6.23 (br. s, 2H; NH2), 6.89 (s, 1H; H-
C(4’)), 7.25 (br. s, 1H; NH), 7.57 (s, 1H; H-C(9’)), 10.8–11.6 ppm (br.
s, 1H; NH); 13C NMR (151 MHz, (CD3)2SO): d=22.42 (Me of AcOꢀ),
24.78 and 26.21 and 26.25 and 26.35 (2 CMe2), 44.72 (C(6)), 69.12
(C(1)), 82.11 (C(4)), 83.21 (C(5)), 84.68 (C(3)), 104.95 and 105.80 (br.,
C(4’,9’)), 109.89 (br., C(8’a)), 110.98 (CMe2), 111.70 (CMe2), 113.11
(C(2)), 131.38 (br., C(9’a)), 140.34 (br., C(3’a)), 144.71 (br., C(4’a)),
150.31 (C(6’)), 157.78 (br., C(2’)), 163.72 (C(8’)), 173.30 ppm (C=O of
AcOꢀ); IR (ATR): n˜ =3600-2000 (w), 3144 (w), 2987 (w), 2938 (w),
1574 (s), 1452 (s), 1372 (s), 1292 (m), 1207 (s), 1067 (s), 1022 (s),
854 (s), 649 (m), 616 cmꢀ1 (m); HR-ESI-MS: m/z (%): 460.2020 (26),
459.1986 (100, [M+H]+, calcd for C21H27N6O6+: 459.1987).
[M+H]+, calcd for C26H35N6O9+: 575.2460), 520.1867 (24), 519.1833
(100, [MꢀCH2 =CMe2 +H]+, calcd for C22H27N6O9
: 519.1834),
+
476.1969 (21), 475.1935 (96, [MꢀCH2 =CMe2ꢀCO2 +H]+, calcd for
C21H27N6O7+: 475.1936).
Methyl 5-[(6-amino-8-oxo-7,8-dihydro-1/3H-imdazo[4,5-g]quina-
zolin-2-yl)amino]-5-deoxy-2,3-O-isopropylidene-b-d-ribofuranosi-
de·0.8acetic acid (1): A solution of 14r (61 mg, 0.11 mmol) in
MeOH (13 mL) was treated with 1m NaOH (0.21 mL), stirred for
26 h at 408C, and evaporated. HPLC (Reprosil chiral NR, 8 mm,
250ꢃ50 mm; flow rate 35 mLminꢀ1, EtOH (0.01 molLꢀ1 NH4OAc)/
heptane 4:6) gave 1 (13 mg, 30%) containing 0.8 equiv of AcOH as
1’-[(6-Amino-8-oxo-7,8-dihydro-1/3H-imidazo[4,5-g]quinazolin-2-
yl)amino]-1-deoxy-2,3:4,5-di-O-isopropylidene-b-d-fructopyra-
nose·0.8 acetic acid (4): A solution of 14f (76 mg, 0.12 mmol) in
MeOH (13 mL) was treated with 1m NaOH (0.24 mL), stirred for
25.5 h at 408C, and evaporated. HPLC (Reprosil Chiral NR, 8 mm,
250ꢃ50 mm; flow rate 35 mLminꢀ1, EtOH (0.01 molLꢀ1 NH4OAc)/
heptane 4:6) gave 4 (43 mg, 77%) containing 0.8 equiv of AcOH as
a white foam. M.p. 2288C; ½aꢃ2D5 =ꢀ38.2 (c=0.05 in MeOH);
1H NMR (600 MHz, (CD3)2SO): d=1.25 (s, 3H) and 1.28 (s, 3H) and
1.42 (s, 6H) (2 CMe2), 1.88 (s, 2.4H; Me of AcOꢀ), 3.57 (br. d, J=
13.9 Hz, 1H; Ha-C(1)), 3.61 (d, J=12.9 Hz, 1H; Heq-C(6)), 3.72 (br. d,
J=14.0 Hz, 1H; Hb-C(1)), 3.75 (dd, J=13.1, 1.9 Hz, 1H; Hax-C(6)),
4.23 (br. dd, J ꢁ8.0, 1.0 Hz, 1H; H-C(5)), 4.41 (d, J=2.5 Hz, 1H; H-
C(3)), 4.58 (dd, J=7.8, 2.5 Hz, 1H; H-C(4)), 6.34 (br. s, 2H; NH2), 6.88
(s, 1H; H-C(9’)), 7.12 (br. s, 1H; NH), 7.57 ppm (s, 1H; H-C(4’));
1
a white foam. M.p. 2748C; ½aꢃ2D5 =ꢀ7.0 (c=0.04 in MeOH); H NMR
(600 MHz, (CD3)2SO): d=1.23 and 1.36 (2 s, 6H; CMe2), 1.82 (s,
2.4H; Me of AcOꢀ), 3.29 (s, 3H; OMe), 3.40 (br. t, J ꢁ6.2 Hz, 2H;
H2C(5)), 4.29 (t, J=7.5 Hz, 1H; H-C(4)), 4.62 (d, J=5.9 Hz, 1H; H-
C(2)), 4.76 (d, J=5.9 Hz, 1H; H-C(3)), 4.95 (s, 1H; H-C(1)), 6.40 (br. s,
2H; NH2), 6.89 (s, 1H; H-C(4’)), 7.49 (br. s, 1H; NH), 7.56 ppm (s,
1H; H-C(9’)); 13C NMR (151 MHz, (CD3)2SO): d=23.09 (Me of AcOꢀ),
24.70 and 26.26 (CMe2), 45.05 (C(5)), 54.30 (OMe), 81.78 (C(3)),
84.64 and 84.66 (C(2,4)), 102–107 (br.; C(4’,9’)), 108.58 (C(1)), 110.05
(br., C(8’a)), 111.40 (CMe2), 144.36 (br., C(4’a)), 150.65 (C(6’)), 157.86
(br., C(2’)), 164.14 (C(8’)), 174.02 ppm (C=O of AcOꢀ), C(3’a,9’a)
hidden by the noise; IR (ATR): n˜ =3600-2300 (w), 3145 (w), 2987
(w), 2936 (w), 1649 (m), 1573 (s), 1519 (m), 1452 (s), 1372 (m), 1290
(m), 1208 (s), 1105 (s), 1090 (s), 867 (m), 616 cmꢀ1 (m); HR-MALDI-
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
9
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
13C NMR (151 MHz, (CD3)2SO): d=21.63 (Me of AcOꢀ), 24.05 and
25.08 and 25.84 and 26.23 (2 CMe2), 48.16 (C(1)), 60.60 (C(6)), 69.62
(C(4)), 70.08 (C(5)), 70.68 (C(3)), 102.67 (C(2)), 102.7–107 (br.,
C(4’,9’)), 107.83 and 108.16 (2 CMe2), 109.88 (br., C(8’a)), 144.18
(br.,), 150.52 (C(6’)), 158.24 (br., C(2’)), 164.14 (C(8’)), 172.71 ppm
(C=O of AcOꢀ), C(3’a,9’a) hidden by the noise; IR (ATR): n˜ =3600-
2100 (w), 2987 (w), 2936 (w), 1647 (m), 1577 (s), 1449 (s), 1372 (s),
1290 (m), 1250 (m), 1209 (s), 1066 (s), 997 (m), 862 (m), 615 cmꢀ1
(m); HR-ESI-MS: m/z (%): 460.2020 (23), 459.1984 (100, [M+H]+,
calcd for C21H27N6O6+: 459.1987).
Z. mobilis TGT crystallization: Protein was cloned, overexpressed,
and purified as described in detail elsewhere.[15,37] Crystals of TGT
appropriate for data collection were obtained by using the hang-
ing-drop, vapor-diffusion method at 288 K. TGT was co-crystallized
with the inhibitors. A protein solution (12 mgmLꢀ1 TGT, 10 mm Tris
(Tris=tris(hydroxymethyl)-aminomethane), pH 7.8, 2m NaCl, 1 mm
EDTA (ethylenediaminetetraacetic acid), 15% (v/v) DMSO) was incu-
bated with 3.2 mm inhibitor. A total of 1.5 mL of this solution was
mixed with 1.5 mL reservoir solution (100 mm MES (2-(N-morpholi-
no)ethanesulfonic acid), pH 5.5, 1 mm DTT (dithiothreitol), 10%
(v/v) DMSO, 13% (w/v) polyethylene glycol (PEG) 8000). After three
weeks of crystal growth, crystals reached dimensions around 0.7ꢃ
0.7ꢃ0.2 mm3.
Acknowledgements
The authors thank Dr. Bruno Bernet (ETH Zurich) for help in
the preparation of the manuscript. Work at ETH was supported
by F. Hoffmann-La Roche Ltd., Basel and by a grant from the
ETH Research Council (ETH-01 13-2). The group at Marburg uni-
versity is grateful to the Deutsche Forschungsgemeinschaft
(DFG) for financial support by the grant KL1204/23-1. We are
grateful to the beamline staff at Bessy II (Helmholtz-Zentrum
Berlin, HZB) and EMBL beamline at Desy (Hamburg). We also
thank the HZB for travel support.
Conflict of interest
The authors declare no conflict of interest.
Keywords: acetonides · carbohydrate ligands · drug design ·
enzymes · water clusters
[1] a) P. F. Agris, EMBO Rep. 2008, 9, 629–635; b) Y. Motorin, M. Helm, Bio-
chemistry 2010, 49, 4934–4944; c) H. Grosjean, V. de Crꢄcy-Lagard, C.
Marck, FEBS Lett. 2010, 584, 252–264.
[2] R. Giegꢄ, J. Lapointe in DNA and RNA Modification Enzymes: Structure,
Mechanism, Function and Evolution (Ed.: H. Grosjean), Landes Biosci-
ence, Austin, TX, 2009, pp. 475–492.
[3] a) J. E. Jackman, J. D. Alfonzo, WIREs RNA 2013, 4, 35–48; b) M. Helm,
J. D. Alfonzo, Chem. Biol. 2014, 21, 174–185; c) P. Schimmel, Nat. Rev.
Mol. Cell Biol. 2018, 19, 45–58.
[4] a) A. G. Torres, E. Batlle, L. R. de Pouplana, Trends Mol. Med. 2014, 20,
306–314; b) N. Jonkhout, J. Tran, M. A. Smith, N. Schonrock, J. S. Mat-
tick, E. M. Novoa, RNA 2017, 23, 1754–1769.
[5] B. Stengl, K. Reuter, G. Klebe, ChemBioChem 2005, 6, 1926–1939.
[6] a) S. G. Van Lanen, S. D. Kinzie, S. Matthieu, T. Link, J. Culp, D. Iwata-
Reuyl, J. Biol. Chem. 2003, 278, 10491–10499; b) Z. D. Miles, R. M.
McCarthy, G. Molnar, V. Bandarian, Proc. Natl. Acad. Sci. USA 2011, 108,
7368–7372.
[7] a) N. Okada, S. Nishimura, J. Biol. Chem. 1979, 254, 3061–3066>; b) Y.-
C. Chen, A. F. Brooks, D. M. Goodenough-Lashua, J. D. Kittendorf, H. D.
Showalter, G. A. Garcia, Nucleic Acids Res. 2011, 39, 2834–2844.
[8] a) J. M. B. Durand, N. Okada, T. Tobe, M. Watarai, I. Fukuda, T. Suzuki, N.
Nakata, K. Komatsu, M. Yoshikawa, C. Sasakawa, J. Bacteriol. 1994, 176,
4627–4634; b) J. M. B. Durand, G. R. Bjçrk, A. Kuwae, M. Yoshikawa, C.
Sasakawa, J. Bacteriol. 1997, 179, 5777–5782; c) K. L. Kotloff, J. P. Winick-
off, B. Ivanoff, J. D. Clemens, D. L. Swerdlow, P. J. Sansonetti, G. K. Adak,
M. M. Levine, Bull. World Health Organ. 1999, 77, 651–666; d) J. M. B.
Durand, B. Dagberg, B. E. Uhlin, G. R. Bjçrk, Mol. Microbiol. 2000, 35,
924–935.
[9] U. Grꢀdler, H.-D. Gerber, D. M. Goodenough-Lashua, G. A. Garcia, R.
Ficner, K. Reuter, M. T. Stubbs, G. Klebe, J. Mol. Biol. 2001, 306, 455–467.
[10] W. Xie, X. Liu, R. H. Huang, Nat. Struct. Biol. 2003, 10, 781–788.
[11] a) T. Ritschel, S. Hoertner, A. Heine, F. Diederich, G. Klebe, ChemBioChem
2009, 10, 716–727; b) P. C. Kohler, T. Ritschel, W. B. Schweizer, G. Klebe,
F. Diederich, Chem. Eur. J. 2009, 15, 10809–10817.
[12] F. Immekus, L. J. Barandun, M. Betz, F. Debaene, S. Petiot, S. Sanglier-
Cianferani, K. Reuter, F. Diederich, G. Klebe, ACS Chem. Biol. 2013, 8,
1163–1178.
X-ray data collection: Crystals were transferred for 10 s into a cry-
oprotectant-containing buffer (50 mm MES, pH 5.5, 300 mm NaCl,
0.5 mm DTT, 2% (v/v) DMSO, 4% (w/v) PEG 8000, 30% (v/v) glycer-
ol). Afterwards, crystals were flash-frozen in liquid nitrogen. The
data sets for TGT·1 and TGT·3 were collected under cryo conditions
(100 K) at the PETRA III P14 (Pixel; Dectris Pilatus3 6m) and for
TGT·2 and TGT·4 at the BESSY-PSF Beamline 14.2 (Pixel; Dectris Pila-
tus3 6m) in Berlin at a wavelength of l=0.91841 ꢂ. TGT crystals
exhibited monoclinic space group symmetry C2 with one mono-
mer per asymmetric unit. Data processing and scaling was carried
out using XDSAPP2. Unit-cell dimensions, data collection, and proc-
essing statistics are given in Table S2.
Structure determination and refinement: The crystal structures
were determined by molecular replacement with the coordinates
of a TGT search model (PDB ID: 1PUD)[15] using Phaser MR from the
CCP4.[38] A subset of randomly selected 5% of all reflections were
omitted during refinement, but used for the calculation of Rfree
.
Model building was performed by using COOT[39] and refinement
in PHENIX.refine version 1.10.1_2155.[40] Rigid body refinement was
applied for the first round of refinement, followed by Cartesian si-
mulated annealing. Subsequently, alternate cycles of structural
adaption and refinement were performed until Rfree values reached
convergence. The refinement cycles comprised the coordinate xyz,
occupancy, and individual B-factor refinement as well as applying
metal restraints for the zinc ion. Temperature factors of all atoms
were refined anisotropically. Ligand restraints were generated by
the grade web server.[41] Alternate conformations of side chains
were fitted into well-defined positive Fo–Fc electron densities,
even if the resulting 2Fo–Fc density for the second conformations
turned out to be weak. Alternative conformations of side chains
were only modeled if their occupancy exceeded 20% after refine-
ment. For adding water molecules, the option update waters in-
cluded in PHENIX was used. Only those water molecules which
were, after refinement, characterized by a 2Fo–Fc electron density
peak with sꢄ1.0. Final validation of the model was performed
with PROCHECK.[42] Coordinate files were deposited in the PDB
with the following access codes: 5JSW, 5JSV, 6FMN, 6FPU.
[13] a) S. R. Hçrtner, T. Ritschel, B. Stengl, C. Kramer, W. B. Schweizer, B.
Wagner, M. Kansy, G. Klebe, F. Diederich, Angew. Chem. Int. Ed. 2007, 46,
8266–8269; Angew. Chem. 2007, 119, 8414–8417; b) L. J. Barandun, F.
Immekus, P. C. Kohler, T. Ritschel, A. Heine, P. Orlando, G. Klebe, F. Die-
derich, Acta Crystallogr. Sect. D 2013, 69, 1798–1808; c) M. Neeb, P. Czo-
drowski, A. Heine, L. J. Barandun, C. Hohn, F. Diederich, G. Klebe, J. Med.
Chem. 2014, 57, 5554–5565; d) M. Neeb, M. Betz, A. Heine, L. J. Baran-
dun, C. Hohn, F. Diederich, G. Klebe, J. Med. Chem. 2014, 57, 5566–
&
&
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
10
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
5578; e) M. Neeb, C. Hohn, F. R. Ehrmann, A. Hꢀrtsch, A. Heine, D. Die-
derich, G. Klebe, Bioorg. Med. Chem. 2016, 24, 4900–4910.
[14] a) L. J. Barandun, F. R. Ehrmann, D. Zimmerli, F. Immekus, M. Giroud, C.
Grꢅnenfelder, W. B. Schweizer, B. Bernet, M. Betz, A. Heine, G. Klebe, F.
Diederich, Chem. Eur. J. 2015, 21, 126–135; b) F. R. Ehrmann, J. Stojko,
A. Metz, F. Debaene, L. J. Barandun, A. Heine, F. Diederich, S. Cianfꢄrani,
K. Reuter, G. Klebe, PLoS One 2017, 12, e0175723.
Touboul, D. Egli, L. Herdeis, A. Vasella, Helv. Chim. Acta 2014, 97, 1244–
1268.
[30] a) M. M. Teeter, Proc. Natl. Acad. Sci. USA 1984, 81, 6014–6018; b) S. Vai-
theeswaran, H. Yin, J. C. Rasaiah, G. Hummer, Proc. Natl. Acad. Sci. USA
2004, 101, 17002–17005; c) J. Lee, S. H. Kim, Protein Sci. 2009, 18,
1370–1376; d) M. Betz, T. Wulsdorf, S. G. Krimmer, G. Klebe, J. Chem. Inf.
Model. 2016, 56, 223–233.
[31] P. R. Gerber, K. Mꢅller, J. Comput.-Aided Mol. Des. 1995, 9, 251–268.
[32] a) A. Nicholls, K. A. Sharp, B. Honig, Proteins Struct. Funct. Bioinf. 1991,
11, 281–196; b) J. E. Ladbury, Chem. Biol. 1996, 3, 973–980; c) S. W.
Homans, Drug Discovery Today 2007, 12, 534–539; d) R. Baron, P. Setny,
J. A. McCammon, J. Am. Chem. Soc. 2010, 132, 12091–12097; e) P. W.
[15] C. Romier, K. Reuter, D. Suck, R. Ficner, EMBO J. 1996, 15, 2850–2857.
[16] a) J. S. Mills, A. Bowers, C. Djerassi, H. J. Ringold, J. Am. Chem. Soc. 1960,
82, 3399–3404; b) M. Britzi, E. Perucca, S. Soback, R. H. Levy, C. Fatorre,
F. Crema, G. Gatti, D. R. Doose, B. E. Maryanoff, M. Bialer, Epilepsia 2005,
46, 378–338; c) M. Katz, E. H. Gans, J. Pharm. Sci. 2008, 97, 2936–2947;
d) C. Priem, A. Wuttke, M. Berditsch, A. S. Ulrich, A. Geyer, J. Org. Chem.
2017, 82, 12366–12376.
[17] D. S. Wishart, C. Knox, A. C. Guo, S. Shrivsatava, M. Hassanali, P. Stothard,
Z. Chang, J. Woolsey, Nucleic Acids Res. 2006, 34, D668–D672.
[18] L. J. Barandun, F. Immekus, P. C. Kohler, S. Tonazzi, B. Wagner, S. Wendel-
spiess, T. Ritschel, A. Heine, M. Kansy, G. Klebe, F. Diederich, Chem. Eur.
J. 2012, 18, 9246–9257.
[19] N. J. Leonard, K. L. Carraway, J. Heterocycl. Chem. 1966, 3, 485–489.
[20] a) T. N. Montgomery, J. Am. Chem. Soc. 1934, 56, 419–423; b) S. Mio, Y.
Kumagawa, S. Sugai, Tetrahedron 1991, 47, 2133–2144.
[21] M. Steiger, T. Reichstein, Helv. Chim. Acta 1936, 19, 184–189.
[22] C. Lecoutey, C. Fossey, S. Rault, F. Fabis, Eur. J. Org. Chem. 2011, 2785–
2788.
[23] a) K. Bock, J. Ø. Duus, J. Carbohydr. Chem. 1994, 13, 513–543; b) T. Kato,
A. Vasella, D. Crich, Carbohydr. Res. 2017, 448, 10–17.
[24] ACD/Labs, Version 12.01, Advanced Chemistry Development, Toronto,
2009.
[25] J. Sangster, J. Phys. Chem. Ref. Data 1989, 18, 1111–1227.
[26] a) E. Gallicchio, M. M. Kubo, R. M. Levy, J. Am. Chem. Soc. 1998, 120,
4526–4527; b) T. S. G. Olsson, J. E. Ladbury, W. R. Pitt, M. A. Williams,
Protein Sci. 2011, 20, 1607–1618; c) J. D. Chodera, D. L. Mobley, Annu.
Rev. Biophys. 2013, 42, 121–142.
[27] L. Di, C. Keefer, D. O. Scott, T. J. Strelevitz, G. Chang, Y.-A. Bi, Y. Lai, J.
Duckworth, K. Fenner, M. D. Troutman, R. S. Obach, Eur. J. Med. Chem.
2012, 57, 441–448.
´
Snyder, J. Mecinovic, D. T. Moustakas, S. W. Thomas III, M. Harder, E. T.
Mack, M. R. Lockett, A. Hꢄroux, W. Sherman, G. M. Whitesides, Proc. Natl.
Acad. Sci. USA 2011, 108, 17889–17894; f) F. Spyrakis, M. H. Ahmed,
A. S. Bayden, P. Cozzini, A. Mozzarelli, G. E. Kellogg, J. Med. Chem. 2017,
60, 6781–6827.
[33] a) A. Biela, N. N. Nasief, M. Betz, A. Heine, D. Hangauer, G. Klebe, Angew.
Chem. Int. Ed. 2013, 52, 1822–1828; Angew. Chem. 2013, 125, 1868–
1876; b) S. G. Krimmer, M. Betz, A. Heine, G. Klebe, ChemMedChem
2014, 9, 833–846; c) J. Cramer, S. G. Krimmer, A. Heine, G. Klebe, J. Med.
Chem. 2017, 60, 5791–5799.
[34] So-called ‘classic“ hydrophobic approach. For examples, see: G. Klebe,
Nat. Rev. Drug Discovery 2015, 14, 95–110.
[35] E. A. Meyer, R. C. Castellano, F. Diederich, Angew. Chem. Int. Ed. 2003, 42,
1210–1250; Angew. Chem. 2003, 115, 1244–1287.
[36] The PyMOL Molecular Graphics System, Version 1.8.6.0, Schrçdinger
Inc., New York, 2017.
[37] For a detailed description of Z. mobilis TGT crystallization, see: a) C.
Romier, R. Ficner, K. Reuter, D. Suck, Proteins Struct. Funct. Bioinf. 1996,
24, 516–519; b) K. Reuter, R. Ficner, J. Bacteriol. 1995, 177, 5284–5288.
[38] A. McCoy, Acta Crystallogr. Sect. D 2007, 63, 32–41.
[39] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D 2004, 60, 2126–2132.
[40] P. D. Adams, P. V. Afonine, G. Bunkꢆczi, V. B. Chen, I. W. Davis, N. Echols,
J. J. Headd, L.-W. Hung, G. J. Kapral, R. W. Grosse-Kunstleve, A. J. McCoy,
N. W. Moriarty, R. Oeffner, R. J. Read, D. C. Richardson, J. S. Richardson,
T. C. Terwilliger, P. H. Zwart, Acta Crystallogr. Sect. D 2010, 66, 213–221.
[41] http://grade.globalphasing.org.
[28] S. G. Krimmer, J. Cramer, M. Betz, V. Fridh, R. Karlsson, A. Heine, G.
Klebe, J. Med. Chem. 2016, 59, 10530–10548.
[42] R. A. Laskowski, J. Appl. Crystallogr. 1993, 26, 283–291.
[29] a) K. Chiesa, A. Shvoryna, B. Bernet, A. Vasella, Helv. Chim. Acta 2010, 93,
668–691; b) K. Chiesa, B. Bernet, A. Vasella, Helv. Chim. Acta 2010, 93,
1822–1843; c) M. Schulze-Adams, D. Touboul, B. Bernet, A. Vasella, Helv.
Chim. Acta 2014, 97, 1037–1054; d) M. Schulze-Adams, B. Bernet, D.
Manuscript received: April 9, 2018
Revised manuscript received: May 10, 2018
Version of record online: && &&, 0000
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
11
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
FULL PAPER
Sugar acetonides in the protein
&
Drug Design
pocket: In the search for new agents
against Shigella bacteria, lin-benzogua-
nine ligands functionalized with aceto-
nide-protected sugars were prepared
and shown to have high inhibitory ac-
tivity against the target Z. mobilis TGT
(tRNA-guanine transglycosylase). The
acetonide motif provides good water
solubility and high metabolic stability.
The optimal space-filling in the ribose-
33 pocket of the enzyme is achieved by
stepwise changing the structure of
sugar acetonides and through incorpo-
ration of a tricyclic, [5+5+4-ring] water
cluster into the pocket.
L. D. Movsisyan, E. Schꢀfer, A. Nguyen,
F. R. Ehrmann, A. Schwab, T. Rossolini,
D. Zimmerli, B. Wagner, H. Daff, A. Heine,
G. Klebe,* F. Diederich*
&& – &&
Sugar Acetonides are a Superior Motif
for Addressing the Large, Solvent-
Exposed Ribose-33 Pocket of tRNA-
Guanine Transglycosylase
&
&
Chem. Eur. J. 2018, 24, 1 – 12
www.chemeurj.org
12
ꢁ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!