Glycomimetics Targeting Glycosyltransferases: Synthetic, Computational and Structural Studies of Less-Polar Conjugates

Article abstract of DOI:10.1002/chem.201600467

The Leloir donors are nucleotide sugars essential for a variety of glycosyltransferases (GTs) involved in the transfer of a carbohydrate to an acceptor substrate, typically a protein or an oligosaccharide. A series of less-polar nucleotide sugar analogues derived from uridine have been prepared by replacing one phosphate unit with an alkyl chain. The methodology is based on the radical hydrophosphonylation of alkenes, which allows coupling of allyl glycosyl compounds with a phosphate unit suitable for conjugation to uridine. Two of these compounds, the GalNAc and galactose derivatives, were further tested on a model GT, such as GalNAc-T2 (an important GT widely distributed in human tissues), to probe that both compounds bound in the medium-high micromolar range. The crystal structure of GalNAc-T2 with the galactose derivative traps the enzyme in an inactive form; this suggests that compounds only containing the β-phosphate could be efficient ligands for the enzyme. Computational studies with GalNAc-T2 corroborate these findings and provide further insights into the mechanism of the catalytic cycle of this family of enzymes.

Full text of DOI:10.1002/chem.201600467

DOI: 10.1002/chem.201600467
Full Paper
&
Molecular Modeling
Glycomimetics Targeting Glycosyltransferases: Synthetic,
Computational and Structural Studies of Less-Polar Conjugates
Mattia Ghirardello,[a] Matilde de las Rivas,[b] Alessandra Lacetera,[c] Ignacio Delso,[a, d]
Erandi Lira-Navarrete,[a] Tomµs Tejero,[a] Sonsoles Martín-Santamaría,*[c] Ramón Hurtado-
Guerrero,*[b, e, f] and Pedro Merino*[a]
Abstract: The Leloir donors are nucleotide sugars essential
for a variety of glycosyltransferases (GTs) involved in the
transfer of a carbohydrate to an acceptor substrate, typically
a protein or an oligosaccharide. A series of less-polar nucleo-
tide sugar analogues derived from uridine have been pre-
pared by replacing one phosphate unit with an alkyl chain.
The methodology is based on the radical hydrophosphony-
lation of alkenes, which allows coupling of allyl glycosyl
compounds with a phosphate unit suitable for conjugation
to uridine. Two of these compounds, the GalNAc and galac-
tose derivatives, were further tested on a model GT, such as
GalNAc-T2 (an important GT widely distributed in human tis-
sues), to probe that both compounds bound in the
medium–high micromolar range. The crystal structure of
GalNAc-T2 with the galactose derivative traps the enzyme in
an inactive form; this suggests that compounds only con-
taining the b-phosphate could be efficient ligands for the
enzyme. Computational studies with GalNAc-T2 corroborate
these findings and provide further insights into the mecha-
nism of the catalytic cycle of this family of enzymes.
ferent metabolites.[1] The resulting glycoconjugates mediate
a wide range of functions from structure and storage to signal-
ling and, as a consequence, they are related to important dis-
eases. Therefore, the chemical manipulation of the activity of
GTs could lead to the development of useful therapeutic
drugs.[2] Thus, considerable synthetic efforts have been direct-
ed toward the preparation of efficient GT inhibitors.[3] Transfer
of the sugar residue occurs from an anionic nucleotide sugar
donor to the acceptor substrate; it can take place with reten-
tion or inversion of configuration at the anomeric centre of the
sugar residue.[4] In this context, elucidation of the mechanisms
used by GTs has been pursued with much interest.[5] Notably,
only nine sugar donors are known to be involved in protein
glycosylation, which is the most abundant post-translational
modification in nature, in mammals.[6] Six of these sugar
donors contain the uridine moiety (Figure 1), which is in agree-
ment with the existence of GTs employing uridine diphosphate
(UDP) sugars as the most predominant in nature.[7]
Introduction
Glycosyltransferases (GTs) are key enzymes responsible for the
incorporation of carbohydrates into a variety of acceptor bio-
molecules, including proteins, lipids, oligosaccharides and dif-
[a] M. Ghirardello, Dr. I. Delso, Dr. E. Lira-Navarrete, Prof. T. Tejero,
Prof. P. Merino
Departamento de Síntesis y Estructura de BiomolØculas
Instituto de Síntesis Química y Catµlisis HomogØnea (ISQCH)
Universidad de Zaragoza, CSIC, 50009 Zaragoza, Aragón (Spain)
E-mail: pmerino@unizar.es
[b] M. d. l. . Rivas, Dr. R. Hurtado-Guerrero
Instituto de Biocomputación y Fisica de Sistemas Complejos (BIFI)
BIFI-IQFR (CSIC) Joint Unit, Universidad de Zaragoza
50009 Zaragoza (Spain)
E-mail: rhurtado@bifi.es
[c] A. Lacetera, Dr. S. Martín-Santamaría
Departamento de Biología Físico-Química
Centro de Investigaciones Biológicas
Most of designed inhibitors mimic nucleotide phosphate
sugars by incorporating anionic groups to emulate binding of
the diphosphate bridge.[8] However, such compounds are not
capable of permeating into cells due to their high polarity.[9] To
overcome this problem, neutral inhibitors have also been pre-
pared,[10] and derivatives in which the phosphate group was re-
placed by a different apolar group showed enhanced cell inter-
nalisation,[10f,g] although some reduction in binding affinity oc-
curred. Vocadlo and co-workers demonstrated that UDP-
5SGlcNAc acted as an inhibitor of O-GT,[11] but it should be
generated inside the cell from precursor 5SGlcNAc by using
the salvage pathway. Nevertheless, most of the designed com-
pounds resulted in poor biological activity because they did
CIB-CSIC, Ramiro de Maeztu, 9, 28040 Madrid (Spain)
E-mail: smsantamaria@cib.csic.es
[d] Dr. I. Delso
Servicio de Resonancia MagnØtica Nuclear
Centro de Química y Materiales de Aragón (CEQMA)
Universidad de Zaragoza, CSIC, Campus San Francisco
50009 Zaragoza (Spain)
[e] Dr. R. Hurtado-Guerrero
Fundación ARAID, 50018 Zaragoza (Spain)
[f] Dr. R. Hurtado-Guerrero
Instituto de Investigaciones Sanitarias de Aragón (IIS-A)
Zaragoza 50009 (Spain)
Supporting information and the ORCID identification number(s) for the au-
thor(s) of this article can be found under http://dx.doi.org/10.1002/
chem.201600467.
Chem. Eur. J. 2016, 22, 7215 – 7224
7215
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 1. Uridine-based sugar donors.
not incorporate the nucleoside moiety, which diminished rec-
ognition and selectivity by the target enzymes.
Herein, we report the design, preparation and binding stud-
ies of less-polar uridyl–sugar analogues 6 as suitable GT bind-
ers in which the b-phosphate unit has been replaced by an
alkyl chain (Figure 2).
Insertion of methylene (2)[12] and replacement of the anome-
ric oxygen by methylene (3)[13] or propylene (4)[14] groups has
already been reported, but compounds 2–4 are still too polar
due to the presence of two phosphate units (Figure 2). Com-
pound 3a was evaluated against three bacterial UDP–galacto-
pyranose mutases and showed only moderate inhibition, al-
though crystallographic studies showed that it was bound to
the active site of the enzyme in a novel conformation not ob-
served previously.[13d] As a model enzyme, we have selected
GalNAc-T2; a member of the important group of N-acetylgalac-
tosaminyl transferases (GalNAcTs E.C. 2.4.1.41) involved in
Figure 2. Structures of the C-analogues of uridine dinucleotide sugars dis-
cussed herein.
group close to the sugar moiety could be enough for the
uridyl–sugar analogue to interact with the protein. These com-
bined studies also provide new evidence of the mechanism of
the catalytic cycle of GalNAc-T2.
mucin biosynthesis.[15] In particular, a previous computational Results and Discussion
study with GalNAc-T2 reported by Masgrau and co-workers an-
Synthesis of glycomimetics
ticipated that UDP-2’-deoxyGal would not be a good donor
substrate, whereas UDP-Gal could be a valid sugar donor in
some cases.[16]
The starting allyl sugars 7 used herein were prepared from d-
glucose for allylsugar 7a,[18] d-mannose for allylsugar 7b,[19] d-
glucosamine for allylsugar 7c[20] and d-galactose for allylsugars
7d[19,21] and 7e (Scheme 1).[22] Protected diacetyl uridine 11
was prepared from commercially available uridine via the trityl
derivative.[23]
Based on these precedents, we designed analogues 6, in
which one phosphate group was replaced by an ethylene
group, as suitable GT binders with GalNAc-T2 as a model
enzyme. The synthetic strategy is based on coupling between
uridine and phosphoalkyl sugars, which are prepared through
radical hydrophosphonylation[17] of allyl sugars.
Photoinduced free-radical hydrophosphonylation of 7a–e,
following the protocol developed by Dondoni and co-work-
ers,[24] afforded phosphonates 8a–e with complete regioselec-
tivity and good yields (Scheme 1). Removal of the protecting
groups of 8d under typical conditions[25] led to free phospho-
nate 9d in excellent yield. Any attempt to couple compound
9d with 10 in the presence of N,N’-dicyclohexylcarbodiimide
(DCC)[26] failed, as well as deprotection of both OMe groups.
Izumi and co-workers previously reported the condensation of
a phosphate monoester with a nucleoside in good yield.[27]
We also demonstrate through tryptophan fluorescence spec-
troscopy and X-ray studies that two of these compounds bind
moderately to the enzyme. In particular, trapping of a galactose
moiety containing compound bound to an inactive form of
GalNAc-T2 provides hints for improving their affinity through
the incorporation of b-phosphate or similar negatively charged
groups. Furthermore, in silico studies support the aforemen-
tioned findings and that the presence of only one phosphate
Chem. Eur. J. 2016, 22, 7215 – 7224
www.chemeurj.org
7216
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Scheme 1. Synthesis of partially free phosphonates. i) dimethyl phosphite, dieethyl phosphate acetophenone (DPAP), hn, 30 min, RT; ii) Me3SiCl, NaBr, MeCN,
1 h, 408C; then 2 h, RT; then 5:1 H2O/EtOAc, 2 h, RT; then NH4OH, 2 h, RT; iii) PhSH, Et3N, dioxane, 4 h, RT; iv) 10 (1.0 equiv), (benzotriazol-1-yloxy)tripyrrolidino-
phosphonium hexafluorophosphate (BOP), DMF, iPrEt2N, 4 h, RT; v) 25% NH4OH(aq), MeOH, 7 h, RT.
Therefore, we decided to prepare monoesters 10a–e through
selective deprotection of the phosphonate group by using thi-
ophenol and triethylamine.[28] Condensation of monoesters
10a–d and diacetyluridine 11 by using BOP and N,N-diisopro-
pylethylamine (DIEA) in DMF[29] yielded the fully protected nu-
cleotide sugar analogues 12a–e (as a mixture of diastereomers
due to phosphorous chirality) with good yields. Removal of the
methyl group of the phosphonate[28] gave 13a–e.
Finally, deacetylation of 13a–e with a 25% aqueous solution
of ammonium hydroxide afforded free sugar nucleotide ana-
logues 14a–e in good yields after purification through a Bio-
Gel P-2 column (eluted with NH4HCO3 in 4:1 H2O/MeOH), fol-
lowed by lyophilisation.
that interactions with Mn2+ for these ligands might not be cru-
cial for their binding. Contrary to the preference of GalNAc-T2
for UDP–GalNAc as the favourite donor substrate for catalysis,
the results also suggest that the compound with the galactose
moiety (14d) binds about 3.5-fold better than that with the
GalNAc moiety (14e).
Compound 14d traps the enzyme in an inactive state. To
understand the binding mode of these compounds to GalNAc-
T2, we solved the crystal structure of GalNAc-T2 in a complex
Crystallographic and binding studies
To validate whether our compounds might have a biological
effect, we selected the GT GalNAc-T2 as an example. This
enzyme is part of a large family of 20 members that transfer
a GalNAc moiety from UDP–GalNAc onto Ser or Thr residues of
proteins in the presence of manganese.[30] This enzyme pre-
dominantly uses UDP–GalNAc as the sugar donor, although, in
some cases, it might also use UDP–galactose.[16,31] Prompted
by this, we determined whether compounds 14d and 14e,
that contained a galactose and GalNAc moiety, respectively,
might bind to this enzyme. Tryptophan fluorescence spectros-
copy studies for compound 14d gave Kd values of (269Æ40)
and (254Æ20) mm, and those for compound 14e gave Kd
values of (800Æ10) and (915Æ180) mm, in the absence and
presence of 2 mm Mn2+ (Figure 3), respectively; this suggested
Figure 3. Quenching of intrinsic GalNAc-T2 tryptophan fluorescence mea-
sured with increasing concentrations of 14d in the absence and presence of
Mn2+. All data points represent the meansÆstandard deviations (S.D.) for
three measurements. The Kd value for 14d was determined by fitting fluo-
rescence intensity data against the concentration of 14d.
Chem. Eur. J. 2016, 22, 7215 – 7224
www.chemeurj.org
7217
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
with 14d. Despite numerous attempts with both compounds
(14d and 14e) in cocrystallisation and soaking experiments,
we only managed to obtain a complex with 14d through soak-
ing experiments for orthorhombic crystals previously grown
with UDP/Mn2+. The structure was solved at 2.07 Š, which al-
lowed us to solve and interpret the density maps (Table S1 in
the Supporting Information).
The asymmetric unit, as reported previously, displayed six
molecules of GalNAc-T2.[31] Three out of six GalNAc-T2 mole-
cules present in the asymmetric unit clearly showed density
maps for the presence of both UDP/Mn2+ and 14d in each
active site, whereas the three remaining GalNAc-T2 molecules
contained UDP (Figure S1 in the Supporting Information). The
crystal structure shows that the typical GT-A fold (catalytic
domain) is located in the N-terminal region and the lectin
domain is located in the C-terminal region (Figure 4).
Figure 4. Surface representation of GalNAc-T2 in complex with 14d/UDP,
UDP and UDP-5S-GalNAc. Protein, flexible loop and nucleotides/14d are col-
oured black, dark grey and light grey, respectively. The active and inactive
states correspond to closed (4D0Z) and open (5V9 and 2FFV) conformations,
respectively.
Strikingly, GalNAc-T2 in complex with 14d adopts an inac-
tive-state conformation, as reported previously for other binary
complexes containing either glycopeptides or UDP (Fig-
ure 4).[15a,31,32] Both active and inactive states are present
during the catalytic cycle of this enzyme and are associated
with the motion of a flexible loop that can oscillate between
closed and open conformations, respectively. In particular, UDP
and 14d are exposed to the solvent due to the open confor-
mation of the flexible loop, whereas UDP-5S-GalNAc is covered
by the flexible loop, which adopts a closed conformation, ren-
dering the enzyme in an active state (Figure 4; PDB en-
try 4D0Z).
Compound 14d induces an inverted uridine conformation
and an unusual 4-coordinate Mn2+ complex. Closer inspection
of the active site reveals that the uridine moiety of both 14d
and UDP adopts an unusual inverted conformation; this was
also found earlier in a structure of this enzyme in complex
with UDP (Figure 5 and PDB entry 2FFV).[32] This atypical con-
formation was proposed to represent a final step during the
catalytic cycle of GalNAc-T2 in which UDP was ready to exit
the enzyme.[31,32] One major difference is apparent between
both complexes: the metal is unusually four-coordinate in our
crystal structure and five-coordinate in the GalNAc-T2-UDP
Figure 5. Structural features of the sugar nucleotide binding site for GalNAc-T2 in complex with UDP/14d. A1) A close-up view of the active site in complex
with UDP-Mn2+/14d. A2) Detailed view of metal interactions. Two-dimensional schemes of the interactions between 14d (A3) and UDP (A4) and the residues
of GalNAc-T2. Close-up views of the active site in complex with UDP/Mn2+ (B1) and UDP-5S-GalNAc-Mn2+ (C1). Two-dimensional schemes of the interactions
between the ligands (B2) and residues of GalNAc-T2 (C2).
Chem. Eur. J. 2016, 22, 7215 – 7224
www.chemeurj.org
7218
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
complex (PDB entry 2FFV; Figure 5). The metal in the former
structure is coordinated by the a-phosphate, Asp224, His226
and His359 (Figure 5, A3), whereas in the latter complex the
metal is additionally coordinated by the UDP b-phosphate
(Figure 5, B1 and B2). However, six-coordinate Mn2+, which re-
quires the previous residues, pyrophosphate and an extra
water molecule to form an octahedral complex, is not only the
most common coordination in crystal structures of this
enzyme, but also the most abundant in proteins in gener-
al.[15a,31–34] These results suggest that these UDP–phosphonate
scaffolds, with only the a-phosphate, are weaker ligands in
terms of binding to GalNAc-T2, but they are still suitable bind-
ers for this enzyme.
Figure 6. Left: Superposition of docked poses in GalNAc-T2 (PDB-ID 2FFV) of
14d (cyan) and UDP-GalNAc (purple). The open loop is shown in yellow.
Right: Superposition of docked poses in GalNAc-T2 (PDB-ID 4D0Z) of 14d
(cyan) and UDP-GalNAc (purple). The closed loop is shown in magenta.
Previously, different crystal structures obtained from the
same type of orthorhombic crystals suggested that UDP might
be present in both active and inactive states of the enzyme.
On the contrary to our structure, UDP in these crystals always
adopted the typical conformation found for UDP-GalNAc,
which was required for efficient turnover.[31] Thus, it is tempt-
ing to speculate that 14d has allowed us to trap a conforma-
tion in the final step of the catalytic cycle of this enzyme, in
which UDP would be ready for immediate departure from the
enzyme.
poses for all ligands, although with some slight differences.
UDP, UDP-GalNAc and UDP-5S-GalNAc were also docked for
comparison purposes to validate the computational protocol.
Interestingly, all compounds were predicted to bind in both
conformations (open and closed) inside the active site of the
enzyme with binding poses similar to those found in the crys-
tallographic structures.
The crystal structure also supports the aforementioned Kd
values; this confirms that 14d does not rely on Mn2+ for bind-
ing to GalNAc-T2. With regard to direct interactions between
the enzyme and nucleotides/14d in the different complexes, it
is important to note that 14d shows fewer interactions
(Figure 5) than UDP (PDB entry 2FFV) and UDP-5S-GalNAc (PDB
entry 4D0Z). It is important to note that Trp331, a critical resi-
due in the catalytic cycle of these enzymes,[32] adopts an “out”
(out of the active site) conformation in the complex with 14d/
UDP (Figure 5, A1), which correlates well with the inactive
state of GalNAc-T2. On the contrary, Trp331 adopts an “in”
(inside the active site) conformation in the complex with UDP-
GalNAc that sets up the enzyme in an active state (Figure 5,
C1).
Predicted binding energies were higher than those of refer-
ence diphosphate compounds due to the consequently de-
creased ionic phosphate–manganese interactions caused by
the lack of the b-phosphate group. Nevertheless, docked bind-
ing energies were favourable; thus validating the phosphonate
scaffold of analogues 14 as a possible platform for the optimi-
sation of GT binders. The best binding energies were always
predicted for the active (closed) conformation, which was in
agreement with a binding pose with more GalNAc-T2/ligand
interactions, although the docking results indicated the ability
to bind also to the open conformation, which was preferred in
the crystal structure. Both possibilities might coexist in the
equilibrium of conformational ensembles. Docking calculations
are able to estimate both binding poses, whereas the limita-
tions of X-ray, in this case, have not allowed us to obtain the
closed 3D structure. No major differences in the predicted
binding energy were found among the phosphonate ana-
logues 14, which indicated a lack of preferred binding. For
compounds 14, the main contributions to binding are provid-
ed by the uridine–enzyme hydrogen bonds, as in the 4D0Z
complex, and, to a lesser extent, from ionic phosphonate–
Mn2+ interactions through one or two of the phosphonate
oxygen atoms.
Most of the direct interactions between the uridine moiety
of 14d and the enzyme are kept as the UDP from the reported
PDB entry 2FFV, whereas a larger number of direct interactions
are established between UDP-GalNAc and the enzyme, pre-
sumably because the active site is covered by the flexible loop
(Figures 4–6). Finally, the phosphonate of 14d does not inter-
act directly with any residue and the galactose moiety is recog-
nised exclusively by Glu334 (Figure 5, A2); this is likely to be
an explanation for the poor binding of this compound to the
enzyme. In the docking studies, we found that the Gal moiety
adopted several alternative binding poses within the sugar
site.
In the case of the active conformation of the GT, an interest-
ing difference was found with regard to binding of the sugar
moiety of analogues 14, which also showed alternative binding
poses in which the sugar moiety reached the Ala307 CO group
and Asn146 side chain, according to the highly flexible aliphat-
ic chain. For the inactive conformation of GalNAc-T2, the dock-
ing calculations placed the synthetic analogues in the equiva-
lent region to that of the active conformation (Figure 6).
Docking studies
To obtain a 3D picture of the putative binding ability of this
family of compounds, docking calculations were performed for
analogues 14 in both conformations of GalNAc-T2: closed
(active, PDB ID 4D0Z) and open (inactive, PDB ID 2FFV). The
calculations were able to predict reasonable docked binding
Chem. Eur. J. 2016, 22, 7215 – 7224
www.chemeurj.org
7219
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
The uridine moiety is observed to rotate in a similar way to
the crystallographic pose, and the overall conformation of the
ligand was found to adopt a more bended conformation, with
the sugar moiety placed in front of the uridine plane. For com-
parison purposes we performed docking calculations for 14d
to our crystallographic structure (PDB 5V9, chain “B ”) finding
similar results between the crystallographic pose for 14d and
the docked pose for UDP-GalNAc.
Several functional groups from GalNAc-T2 establish interac-
tions, especially those involving the uridine moiety. In particu-
lar, the NH in position 3 of the uracyl ring establishes a stable
hydrogen bond along the simulation with the Asp176 side
chain (Figure 7, right, and Figure S5 in the Supporting Informa-
tion). The CO groups at positions 2 and 4 of the uracyl ring es-
tablish alternating hydrogen bonds with Arg201 and Thr143
side chains (Figure 7, right, and Figure S6 in the Supporting In-
formation) during the rotation of the uridine. This rotation
allows the torsional change undergone by the diphosphate
linker. Also, stacking interactions between the His145 side
chain and the uracyl moiety are identified. Most of these inter-
actions are found in the crystallographic structures, in agree-
ment with a specific anchorage for the uracil moiety. On the
contrary, the flexibility of the diphosphate linker allows GalNAc
to rotate and move during the closing process, which changes
the interactions along the simulation to adapt to the new sur-
rounding environment.
We conclude, from the overall docking study, that com-
pounds 14 are putative GalNAc-T2 binders in which the uridine
moiety plays a major role in binding, in agreement with the
crystallographic binding poses, whereas the phosphonate
group and sugar moiety provide additional interactions; there
are no differences in the types of sugar.
Targeted molecular dynamics (tMD) simulations
To shed light on the conformational changes that the protein
needs to undergo between the inactive and active conforma-
tions (opening/closing process), and to analyse the required in-
teractions to host the substrate, targeted molecular dynamics
(tMD) simulations were undertaken for the GalNAc-T2/UDP-
GalNAc complex. The simulation was started from the complex
in the inactive open conformation (from the docked binding
pose) towards the complex in the active closed conformation
(from the docked binding pose of UDP-GalNAc, very similar to
the crystallographic complex with UDP-5S-GalNAc; Figure 7).
The analysis of the root-mean-square fluctuation (Figure S2 in
the Supporting Information) during the tMD simulation time
identified a main fluctuation that corresponded to the flexible
loop Arg362–Ser373 with a maximum value for Pro370. This
main change corresponds to the folding of the loop over the
substrate, sealing it in the active site, and leading to the
closed conformation. Additionally to the Asp224, His226 and
His359 side chains, the Mn2+ ion was found to coordinate to
two phosphate oxygen atoms (Figure S3 in the Supporting In-
formation). During the progression of the tMD simulation, one
water molecule entered into the coordination sphere, accord-
ing to a six-coordinate Mn2+ model (Figure S4 in the Support-
ing Information).
For example, in the inactive conformation, GalNAc is in-
volved in two hydrogen bonds: one between the endocyclic
GalNAc oxygen and the Arg362 backbone NH, and another
one between the N-acetylamido CO group and the Arg362
guanidinium group. After closing, these interactions with
Arg362 are lost and new hydrogen bonds are established: one
between GalNAc OH-3 and the backbone NH of Gly309, and
another one between OH-4 from the GalNAc moiety and the
Ala307 CO group. Moreover, new CH–p interactions form be-
tween the ribose and the Tyr367 indole ring.
In summary, tMD results may give us an approximation of
the binding pose changes that the UDP-GalNAc undergoes
during the closing/opening process when bound to the GT.
Specific interactions for GalNAc during the closing/opening
process may be useful to understand the mechanism of the
catalytic cycle of this enzyme, and to help in the design of se-
lective GT binders. In fact, we provide a plausible explanation
for the role of the N-acetylamido group in binding, which sup-
ports its higher affinity to the enzyme in comparison to UDP-
Gal.
Conclusion
GalNAc-T2 has been used as a prototype GT for studying the
binding mode of less-polar glycomimetics. Easily accessible
sugar nucleotides analogues, in which one phosphate group
has been replaced by an apolar alkyl chain, are weaker binders
of GalNAc-T2 than the natural substrate UDP-GalNAc, as dem-
onstrated by combined structural and computational studies.
However, compound 14d is capable of trapping the enzyme in
an inactive form. Interestingly, docking studies reveal that both
compounds can bind to the inactive and active forms of
GalNAc-T2, which suggests a more complex scenario. The crys-
tal structure provides additional useful information to design
better compounds that might inhibit GalNAc-T2. In particular,
it is inferred from the crystal structure that the b-phosphate is
required for binding to Mn2+, and thus, replacing the a-phos-
phate by a propylene group, while keeping the phosphate or
Figure 7. Superposition of five structures of the GalNAc-T2/UDP-GalNAc
complex along the tMD simulation of the transition from the open (black,
loop in yellow) to the closed (white, loop in magenta) conformation. Inter-
mediate protein structures are shown as a gradient from yellow (starting ge-
ometry) to orange (intermediate geometries) to magenta (final target geom-
etry). The ligand UDP-GalNAc is shown in yellow for the open complex and
in magenta for the closed complex. Left: Full perspective. Right: Detailed
image showing some representative residues.
Chem. Eur. J. 2016, 22, 7215 – 7224
www.chemeurj.org
7220
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
similar groups in the b position, should maintain or improve
the binding of these compounds.
Hydrophosphonylation of allylglycosides 7a–e: General pro-
cedure
Finally, tMD simulations of the complex of GalNAc-T2 with
UDP-GalNAc has provided new insights into substrate recogni-
tion events involved in the preliminary steps of the catalytic
cycle of GalNAc-T2. We have simulated the change from the
open, inactive conformation to the closed, active conformation
to identify the overall change, with regard to the conformation
and interactions with the protein, as experienced by the sub-
strate to finally lead to the active, closed complex. This transi-
tion involves the rotation of the uridine and GalNAc moieties
and a conformational change of a very dynamic flexible loop.
Our studies, combining the synthesis of UDP-GalNAc-based an-
alogues with structural and computational techniques, allowed
us a better understanding of the substrate recognition and cat-
alytic cycle of GalNAc-T2.
DPAP (37.8 mg, 0.148 mmol) was added to a well-stirred solution
of allyl glycoside 7 (0.295 mmol) in dimethyl phosphite (2.7 mL,
29.5 mmol). The solution was irradiated at room temperature
under stirring for 30 min and then concentrated. The residue was
purified by column chromatography on silica gel (EtOAc/MeOH,
1:0 to 95:5 containing 0.2% of Et3N) to give 8 as an oil.
Dimethyl
[(2,3,4,6-tetra-O-acetyl-a-d-glucopyranosyl)propyl]-
phosphonate (8a): Yield: 367 mg, 94%; ½aŠ27 = +53 (c=1.0,
D
1
CHCl3); H NMR (400 MHz, CDCl3): d=5.32 (dd, J=9.5, 8.7 Hz, 1H;
H-3), 5.09 (dd, J=9.5, 5.8 Hz, 1H; H-2), 4.99 (dd, J=9.6, 8.7 Hz, 1H;
H-4), 4.27 (dd, J=12.0, 5.7 Hz, 1H; H-6), 4.18–4.11 (m, 1H; H-1),
4.09 (dd, J=12.0, 2.6 Hz, 1H; H-6), 3.83 (ddd, J=9.6, 5.7, 2.6 Hz,
1H; H-5), 3.77 (d, J=11.0 Hz, 6H; OCH3), 2.08 (s, 3H; Ac), 2.04 (s,
3H; Ac), 2.01 (s, 3H; Ac), 2.00 (s, 3H; Ac), 1.94–1.51 ppm (m, 6H;
H-7, H-8, H-9); 13C NMR (100 MHz, CDCl3): d=170.1, 170.0, 169.9,
169.8, 72.5, 70.5 (2 C), 69.0 (2 C), 62.7, 52.6 (d, J=6.3 Hz), 26.6 (d,
J=15.5 Hz), 25.0 (d, J=140.5 Hz), 20.9 (2 C), 20.8 (2 C), 18.7 ppm
(d, J=5.0 Hz); 31P NMR (162 MHz, CDCl3): d=32.0 ppm; elemental
analysis calcd (%) for C19H31O12P: C 47.31, H 6.48, P 6.42; found: C
47.56, H 6.30, P 6.24.
Experimental Section
General methods
Dimethyl [(2,3,4,6-tetra-O-acetyl-a-d-mannopyranosyl)propyl]-
phosphonate (8b): Yield: 254 mg, 42%; ½aŠ28 = +7 (c=0.9, CHCl3);
The reaction flasks and other glass equipment were heated in an
oven at 1308C overnight and assembled in a stream of argon. All
reactions were monitored by TLC on silica gel 60 F254; the posi-
tions of the spots were detected by l=254 nm UV light or by
spraying with either a 5% solution of phosphomolybdic acid in
ethanol or Mostain solution. Column chromatography was carried
out in a Buchi 800 MPLC system or a Combiflash apparatus by
using silica gel 60 microns and with solvents that were distilled
prior to use. Melting points were uncorrected. Purification by semi-
preparative HPLC (column Atlantisꢁ DC18 5 mm, 19”100 mm, flow:
12.5 mLminÀ1) was carried out in a Waters 515 pump system with
photodiode array (PDA) detection. 1H and 13C NMR spectra were re-
corded on Bruker Avance 400 MHz or AVANCE II 300 MHz instru-
ments in the stated solvent. Chemical shifts are reported in ppm
(d) relative to CHCl3 (d=7.26) in CDCl3. NMR assignments were
made by using standard 2D experiments. For numbering of NMR
data, see Figure 8. Optical rotations were recorded on a JASCO
DIP-370 polarimeter. Elemental analyses were performed on a Perki-
nElmer 240B microanalyzer or with a PerkinElmer 2400 instrument.
High-resolution mass spectra were recorded on a QToF spectrome-
ter equipped with an ESI source (microTOF-Q, Bruker Daltonik) by
using sodium formate as an external reference. Hydrophosphonyla-
tion was carried out in a glass vial (diameter: 1 cm; wall thickness:
0.65 mm), closed with a natural rubber septum, located 2.5 cm
away from the household UVA lamp apparatus equipped with four
15 W tubes (1.5”27 cm each).
D
1H NMR (400 MHz, CDCl3): d=5.22 (dd, J=8.9, 3.3 Hz, 1H; H-3),
5.17 (dd, J=8.9, 8.5 Hz, 1H; H-4), 5.15 (dd, J=3.3, 3.2 Hz, 1H; H-2),
4.36 (dd, J=12.0, 6.5 Hz, 1H; H-6), 4.10 (dd, J=12.0, 4.0 Hz, 1H; H-
6), 3.95 (ddd, J=10.6, 3.8, 3.2 Hz, 1H; H-1), 3.87 (ddd, J=8.5, 6.5,
4.0 Hz, 1H; H-5), 3.74 (d, J=9.5 Hz, 6H; OCH3), 2.12 (s, 3H; Ac),
2.11 (s, 3H; Ac), 2.06 (s, 3H; Ac), 2.03 (s, 3H; Ac), 1.92–1.60 ppm
(m, 6H; H-7, H-8, H-9); 13C NMR (100 MHz, CDCl3): d=170.7, 170.5,
170.0, 169.8, 73.8, 70.6 (2 C), 68.8, 66.9, 62.3, 52.2 (d, J=9.8 Hz),
29.2 (d, J=15.2 Hz), 24.0 (d, J=140.5 Hz), 20.8, 20.7, 20.6, 20.5,
18.5 ppm (d, J=4.5 Hz); 31P NMR (162 MHz, CDCl3): d=31.8 ppm;
elemental analysis calcd (%) for C19H31O12P: C 47.31, H 6.48, P 6.42;
found: C 47.22, H 6.38, P 6.51.
Dimethyl [(2-acetamido-2-deoxy-3,4,6-tri-O-acetyl-a-d-glucopyr-
anosyl)propyl]phosphonate (8c): Yield: 213 mg, 73%; ½aŠ28 = +29
D
(c=0.9, CHCl3); 1H NMR (400 MHz, CDCl3): d=6.00 (d, J=8.5 Hz,
1H; NHAc), 4.97 (dd, J=8.5, 7.5 Hz, 1H; H-3), 4.87 (dd, J=7.5,
7.0 Hz, 1H; H-4), 4.26 (dd, J=12.0, 5.6 Hz, 1H; H-6), 4.19 (dt, J=8.5,
4.7 Hz, 1H; H-2), 4.09–4.01 (m, 2H; H-1, H-6), 3.79 (ddd, J=7.0, 5.6,
3.5 Hz, 1H; H-5), 3.67 (d, J=10.6 Hz, 6H; OCH3), 2.04 (s, 3H; Ac),
2.01 (s, 3H; Ac), 2.00 (s, 3H; Ac), 1.91 (s, 3H; Ac), 1.79–1.44 ppm
(m, 6H; H-7, H-8, H-9); 13C NMR (100 MHz, CDCl3): d=170.9, 170.6,
169.9, 169.0, 71.3, 70.2, 70.1, 68.5, 61.7, 52.3 (d, J=6.7 Hz), 52.3 (d,
J=6.7 Hz), 50.7, 27.3 (d, J=14.3 Hz), 24.0 (d, J=141.5 Hz), 23.1,
20.8, 20.7, 20.6, 18.5 ppm (d, J=4.9 Hz); 31P NMR (162 MHz, CDCl3):
d=34.2 ppm; elemental analysis calcd (%) for C19H32NO11P: C
47.40, H 6.70, N 2.91, P 6.43; found: C 47.31, H 6.55, N 3.31, P 6.35.
Dimethyl [(2,3,4,6-tetra-O-acetyl-a-d-galactopyranosyl)propyl]-
phosphonate (8d): Yield: 561 mg, 78%; ½aŠ28 = +60 (c=1.0,
D
1
CHCl3); H NMR (400 MHz, CDCl3): d=5.36 (dd, J=3.4, 2.1 Hz, 1H;
H-4), 5.21 (dd, J=9.5, 5.1 Hz, 1H; H-2), 5.14 (dd, J=9.5, 3.4 Hz, 1H;
H-3), 4.22–4.13 (m, 2H; H-1, H-6), 4.10–3.99 (m, 2H; H-5, H-6), 3.70
(d, J=10.7 Hz, 6H; OCH3), 2.08 (s, 3H; Ac), 2.04 (s, 3H; Ac), 2.03 (s,
3H; Ac), 1.98 (s, 3H; Ac), 1.81–1.45 ppm (m, 6H; H-7, H-8, H-9);
13C NMR (100 MHz, CDCl3): d=170.3, 170.1, 170.0, 169.9, 71.5, 68.4,
68.3, 68.1, 68.7, 61.5, 52.3 (d, J=11.1 Hz), 52.2 (d, J=11.1 Hz), 26.3
(d, J=15.0 Hz), 24.2 (d, J=141.0 Hz), 20.7, 20.7, 20.6, 20.6,
19.1 ppm (d, J=4.4 Hz); 31P NMR (162 MHz, CDCl3): d=34.0 ppm;
Figure 8. Numbering used for NMR spectroscopy assignments of com-
pounds.
Chem. Eur. J. 2016, 22, 7215 – 7224
www.chemeurj.org
7221
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
elemental analysis calcd (%) for C19H31O12P: C 47.31, H 6.48, P 6.42;
found: C 47.45, H 6.37, P 6.29.
11.8, 4.3, 2.5 Hz, 1H; H-5’), 3.98–3.89 (m, 2H; H-5’, H-1), 3.84 (dd,
J=10.7, 5.8 Hz, 1H; H-2), 3.69 (dd, J=12.2, 2.2 Hz, 1H; H-6), 3.64–
3.57 (m, 2H; H-3, H-6), 3.43–3.36 (ddd, J=9.8, 5.5, 2.2 Hz, 1H; H-5),
3.29 (dd, J=9.8, 8.8 Hz, 1H; H-4), 1.92 (s, 3H; NAc), 1.79–1.31 ppm
(m, 6H; H-7, H-8, H-9); 13C NMR (100 MHz, D2O, 258C): d=174.3,
166.3, 151.8, 141.5, 102.3, 88.8, 83.2 (d, J=7.9 Hz), 73.9, 73.3, 72.4,
70.8, 70.7, 69.5, 62.5 (d, J=5.2 Hz), 61.0, 53.4, 25.7 (d, J=16.8 Hz),
25.6 (d, J=134.7 Hz), 21.8, 19.1 ppm (d, J=4.3 Hz); 31P NMR
(162 MHz, D2O, 258C): d=28.3 ppm; elemental analysis calcd (%)
for C20H35N4O13P: C 42.11, H 6.18, N 9.82, P 5.43; found: C 42.28, H
6.23, N 9.71, P 5.65.
Dimethyl [(2-acetamido-2-deoxy-3,4,6-tri-O-acetyl-a-D galacto-
28
D
pyranosyl)propyl]phosphonate (8e): Yield: 82.8 mg, 58%; ½aŠ
=
1
31 (c=1.0, CHCl3); H NMR (400 MHz, CDCl3): d=6.14 (d, J=8.3 Hz,
1H; NHAc), 5.30 (t, J=3.0 Hz, 1H; H-4), 5.10 (dd, J=9.7, 3.0 Hz, 1H;
H-3), 4.52–4.42 (m, 1H; H-2), 4.29–4.14 (m, 2H; H-1, H-6), 4.07 (dd,
J=11.6, 4.9 Hz, 1H; H-6), 3.99 (m, 1H; H-5), 3.72 (d, J=10.8 Hz, 6H;
OCH3), 2.10 (s, 3H; Ac), 1.05 (s, 3H; Ac), 2.03 (s, 3H; Ac), 1.96 (s,
3H; Ac), 1.79–1.44 ppm (m, 6H; H-7, H-8, H-9); 13C NMR (100 MHz,
CDCl3): d=170.7 (2 C), 170.5, 170.2, 71.8, 68.5, 68.3, 67.0, 61.6, 52.3
(d, J=6.7 Hz), 48.6, 26.7, 24.0 (d, J=140.5 Hz), 23.0, 20.9, 20.7, 20.7,
18.8 ppm (d, J=5.2 Hz); 31P NMR (121 MHz, CDCl3): d=27.0 ppm;
elemental analysis calcd (%) for C19H32NO11P: C 47.40, H 6.70, N
2.91, P 6.43; found: C 47.58, H 6.82, N 3.05, P 6.34.
Ammonium uridin-5’-yl [(a-d-galactopyranosyl)propyl]phospho-
nate (14d): Yield: 11.3 mg, 63%; m.p. 152–1558C; ½aŠ29 = +34 (c=
D
1
0.9, H2O); H NMR (400 MHz, D2O, 258C): d=7.93 (d, J=8.1 Hz, 1H;
H-6u), 5.94 (d, J=8.1 Hz, 1H; H-5u), 5.94 (d, J=4.6 Hz, 1H; H-1’),
4.27 (t, J=4.6 Hz, 1H; H-2’), 4.23 (t, J=4.6 Hz, 1H; H-3’), 4.20–4.16
(m, 1H; H-4’), 4.05 (ddd, J=11.8, 4.0, 2.5 Hz, 1H; H-5’), 3.97 (ddd,
J=11.8, 5.2, 2.8 Hz, 1H; H-5’), 3.93 (dd, J=6.1, 2.8 Hz, 1H; H-1),
3.87 (dd, J=9.7, 5.8 Hz, 1H; H-3), 3.84 (dd, J=3.5, 0.7 Hz, 1H; H-2),
3.68 (dd, J=9.9, 3.5 Hz, 1H; H-4), 3.54–3.63 (m, 3H; H-5, H-6), 1.76–
1.35 ppm (m, 6H; H-7, H-8, H-9); 13C NMR (100 MHz, D2O, 258C):
d=166.2, 151.7, 141.6, 102.2, 88.9, 88.2 (d, J=8.0 Hz), 75.1, 73.9,
71.5, 69.7, 69.4, 69.1, 68.3, 62.3 (d, J=5.2 Hz), 61.1, 25.7 (d, J=
134.8 Hz), 24.9 (d, J=17.1 Hz), 19.3 ppm (d, J=4.3 Hz); 31P NMR
(162 MHz, D2O, 258C): d=28.6 ppm; elemental analysis calcd (%)
for C18H32N3O13P: C 40.84, H 6.09, N 7.94, P 5.85; found: C 40.82, H
5.94, N 7.90, P 5.93.
Deprotection of compounds 13a–e: General procedure
A stirred solution of the corresponding derivative 13 (0.034 mmol)
in MeOH (1.7 mL) was treated with a 25% solution of NH4OH
(1.7 mL) and stirred for 7 h at room temperature. The solution was
concentrated in vacuo, and the residue was purified on a Bio-Gel
P-2 column (1”70 cm), eluted with 100 mmol NH4HCO3 in 4:1
H2O/MeOH, and lyophilised to give pure 14 as a foam.
Ammonium uridin-5’-yl [(a-d-glucopyranosyl)-propyl]phospho-
nate (14a): Yield: 13.1 mg, 65%; m.p. 153–1558C; ½aŠ27 = +31 (c=
D
1
0.7, H2O); H NMR (400 MHz, D2O, 258C): d=8.00 (d, J=8.1 Hz, 1H;
Ammonium uridin-5’-yl [(2-acetamido-2-deoxy-a-d-galactopyra-
nosyl)propyl]phosphonate (14e): Yield: 12.6 mg, 65%; m.p. 159–
H-6u), 6.00 (d, J=4.7 Hz, 1H; H-1’), 5.99 (d, J=8.1 Hz, 1H; H-5u),
4.39 (dd, J=5.0, 4.7 Hz, 1H; H-2’), 4.36 (t, J=5.0 Hz, 1H; H-3’),
4.32–4.29 (m, 1H; H-4’), 4.18 (ddd, J=11.8, 4.2, 2.5 Hz, 1H; H-5’),
4.10 (ddd, J=11.8, 5.3, 2.9 Hz, 1H; H-5), 4.03 (ddd, J=11.4, 5.9,
2.8 Hz, 1H; H-1), 3.84 (dd, J=12.2, 2.3 Hz, 1H; H-6), 3.75–3.63 (m,
3H; H-6, H-2, H-3), 3.53 (ddd, J=9.8, 5.6, 2.3 Hz, 1H; H-5), 3.37 (dd,
J=9.8, 8.9 Hz, 1H; H-4), 1.88–1.50 ppm (m, 6H; H-7, H-8, H-9);
13C NMR (100 MHz, D2O, 258C): d=166.6, 152.0, 141.6, 102.4, 88.8,
83.2 (d, J=16.2 Hz), 75.4, 73.9, 73.3, 72.4, 71.3, 70.4, 69.6, 62.6 (d,
J=5.1 Hz), 61.1, 25.7 (d, J=134.5 Hz), 25.0 (d, J=16.8 Hz),
19.1 ppm (d, J=4.3 Hz); 31P NMR (121 MHz, D2O, 258C): d=
28.4 ppm; elemental analysis calcd (%) for C18H32N3O13P: C 40.84, H
6.09, N 7.94, P 5.85; found: C 40.77, H 6.18, N 8.14, P 5.97
1618C; ½aŠ26 = +57 (c=1.0, H2O); 1H NMR (400 MHz, D2O, 258C):
D
d=7.88 (d, J=8.1 Hz, 1H; H-6u), 5.83 (d, J=8.1 Hz, 1H; H-5u), 5.82
(d, J=5.1 Hz, 1H; H-1’), 4.25 (t, J=5.1 Hz, 1H; H-2’), 4.20 (t, J=
5.1 Hz, 1H; H-3’), 4.17–4.14 (m, 1H; H-4’), 4.09 (dd, J=11.0, 5.9 Hz,
1H; H-2), 4.04 (ddd, J=11.7, 4.0, 2.4 Hz, 1H; H-5’), 4.01–3.92 (m,
2H; H-5’, H-1), 3.83–3.81 (m, 1H; H-4), 3.77 (dd, J=11.0, 3.3 Hz, 1H;
H-3), 3.62–3.53 (m, 3H; H-5, H-6), 1.91 (s, 3H; NAc), 1.78–1.32 ppm
(m, 6H; H-7, H-8, H-9); 13C NMR (100 MHz, D2O, 258C): d=174.5,
166.3, 151.7, 141.6, 102.1, 88.9, 83.1 (d, J=7.9 Hz), 73.9, 73.2, 71.4,
69.3, 68.4, 67.3, 62.3 (d, J=5.2 Hz), 61.2, 49.7, 25.7 (d, J=134.9 Hz),
25.5 (d, J=17.3 Hz), 21.9, 19.2 ppm (d, J=4.2 Hz); 31P NMR
(162 MHz, D2O, 258C): d=28.4 ppm; elemental analysis calcd (%)
for C20H35N4O13P: C 42.11, H 6.18, N 9.82, P 5.43; found: C 42.02, H
6.21, N 9.90, P 5.52.
Ammonium uridin-5’-yl [(a-d-mannopyranosyl)propyl]phospho-
nate (14b): Yield: 16.6 mg, 92%; m.p. 134–1368C; ½aŠ28 = +9 (c=
D
1
0.9, H2O); H NMR (400 MHz, D2O, 258C): d=7.96 (d, J=8.1 Hz, 1H;
H-6u), 5.96 (d, J=4.4 Hz, 1H; H-1’), 5.94 (d, J=8.1 Hz, 1H; H-5u),
4.35 (dd, J=5.0, 4.4 Hz, 1H; H-2’), 4.32 (t, J=5.0 Hz, 1H; H-3’),
4.28–4.24 (m, 1H; H-4’), 4.14 (ddd, J=11.7, 4.1, 2.4 Hz, 1H; H-5’),
4.06 (ddd, J=11.7, 5.3, 2.9 Hz, 1H; H-5’), 3.95–3.89 (m, 1H; H-1),
3.87 (dd, J=3.3, 1.8 Hz, 1H; H-2), 3.82 (dd, J=12.2, 2.3 Hz, 1H; H-
6), 3.81 (dd, J=9.4, 4.3 Hz, 1H; H-3), 3.70 (dd, J=12.2, 6.0 Hz, 1H;
H-6), 3.62 (t, J=9.4 Hz, 1H; H-4), 3.50 (ddd, J=9.4, 6.0, 2.3 Hz, 1H;
H-5), 1.94–1.46 ppm (m, 6H; H-7, H-8, H-9); 13C NMR (100 MHz,
D2O, 258C): d=166.3, 151.8, 141.6, 102.3, 88.8, 83.2 (d, J=7.9 Hz),
78.0, 73.9, 73.4, 71.5, 70.8, 69.5, 67.3, 62.6 (d, J=5.2 Hz), 61.3, 28.5
(d, J=16.6 Hz), 25.5 (d, J=134.8 Hz), 19.5 ppm (d, J=4.4 Hz);
31P NMR (121 MHz, D2O, 258C): d=28.4 ppm; elemental analysis
calcd (%) for C18H32N3O13P: C 40.84, H 6.09, N 7.94, P 5.85; found: C
40.86, H 6.23, N 7.89, P 5.74.
Expression and purification of HsGalNAc-T2
Homo sapiens GalNAc-T2 (HsGalNAc-T2) was expressed in Pichia
pastoris and purified as described previously.[15a,32]
Crystallisation
Protein drops were prepared by mixing protein solution (2 mL; the
mix contained 7 mgmLÀ1 GalNAc-T2, 5 mm UDP, 5 mm MnCl2 in
25 mm Tris-HCl pH 7.5) and reservoir solution (2 mL; 8–10% poly-
ethylene glycol (PEG) 8000, 8% ethylene glycol, 100 mm Hepes
pH 7.0–7.5), as reported previously.[32] Crystals were grown by
hanging drop diffusion at 188C for 2 days. Then, compound 13d,
as a powder, was soaked onto these crystals for 25 min. The crys-
tals were cryoprotected in the precipitant solution containing an
additional 18% ethylene glycol. All crystals were frozen in a nitro-
gen gas stream cooled to 100 K.
Ammonium uridin-5’-yl [(2-acetamido-2-deoxy-a-d-glucopyrano-
syl)propyl]phosphonate (14c): Yield: 12.2 mg, 63%; m.p. 161–
1638C; ½aŠ25 = +38 (c=1.1, H2O); 1H NMR (400 MHz, D2O, 258C):
D
d=7.85 (d, J=8.1 Hz, 1H; H-6u), 5.84 (d, J=4.4 Hz, 1H; H-1’), 5.83
(d, J=8.1 Hz, 1H; H-5u), 4.24 (dd, J=4.9, 4.4 Hz, 1H; H-2’), 4.20 (t,
J=4.9 Hz, 1H; H-3’), 4.17–4.13 (m, 1H; H-4’), 4.06–3.99 (ddd, J=
Chem. Eur. J. 2016, 22, 7215 – 7224
www.chemeurj.org
7222
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
3
253 Š was generated for each protein, with the centre defined as
the centre of mass among Asp176, His145, Arg362 and His226 resi-
dues. For all ligands, the torsional angles were allowed to rotate
with no restrictions. A standard precision docking was performed
for both structures. In the case of AutoDock, different conformers
of the starting geometries of the ligands were docked by using the
Lamarckian genetic algorithm, by randomly changing the torsion
angles and overall orientation of the molecule. A volume for explo-
ration was defined with a grid spacing of 0.375 Š, centred on the
centre of mass among Asp176, His145, Arg362 and His226 resi-
Structure determination and refinement
All data were processed and scaled by using the XDS packag-
e[35]and CCP4[36] software; relevant statistics are given in Table S1 in
the Supporting Information. The crystal structure was solved by
molecular replacement with Phaser[37] and by using PDB entry 2FFV
as a template. Initial phases were further improved by cycles of
manual model building in Coot[38] and refinement with
REFMAC5:[39] The final model was validated with PROCHECK
(model statistics are given in Table S1 in the Supporting Informa-
tion). The asymmetric units of these crystals showed six molecules
of GalNAc-T2, forming three independent dimers, as confirmed by
the PISA server.[31] Only chain D showed disordered density to-
wards the end of the lectin domain, and thus, we did not include
residues with numbering 555 to 569 in the final model. Coordi-
nates and structure factors have been deposited in the Worldwide
Protein Data Bank (wwPDB; see Table S1 in the Supporting Infor-
mation for the pdb codes).
3
3
dues, and a 3D grid of 21”18”19 Š for 4D0Z and 21”21”18 Š
for 2FFV. After docking, the 200 solutions were clustered in groups
with root-mean-square deviations of less than 2.0 Š. The clusters
were ranked by the lowest energy representative of each cluster.
Docking results from both Glide and AutoDock methods were ana-
lysed and taken together for the discussion.
tMD simulations: The tMD simulation was performed with the
AMBER12 software[43] by using the ff10, gaff and Glycam force
fields. The Ca atoms of the starting structure (GalNAc-T2 in com-
plex with UDP-GalNAc in the open conformation) were restrained
to the corresponding positions of the target structure (GalNAc-T2
in complex with UDP-GalNAc in the closed conformation). Explicit
water solvent (TIP3BOX model) was considered by adding the
same number of water molecules to both complexes. Also, the
same number of counterions (4 chlorine ions) was added. Energy
minimisation of all initial structures (closed and open states) was
carried out by using the steepest descent method in AMBER12 for
500 steps and the non-bonded cutoff was 9 Š. For the tMD simula-
tion, the SHAKE algorithm was used, the time step was 0.002 ps,
and the non-bonded cutoff was 15 Š. The temperature coupling
method was used to keep the temperature constant at 300 K and
Berendsen temperature coupling scheme was used. No positional
restraints were applied. The total simulation time was 5 ns.
Tryptophan fluorescence spectroscopy
Fluorescence spectroscopy was used to determine the dissociation
constant of GalNAc-T2 against the compounds. All experiments
were carried out in a Cary Eclipse spectrofluorometer (Varian) at
258C with GalNAc-T2 at 1 mm, and concentrations of the com-
pound varying from 1 to 1000 mm in 25 mm Tris, 150 mm NaCl,
pH 7.5. Fluorescence emission spectra were recorded in the l=
300–400 nm range with an excitation wavelength of l=280 nm
and a slit width of 5 nm. Data analysis was performed in Prism
(GraphPad software) by considering a model with a single binding
site [see Eq. (1), in which F0 is the intrinsic fluorescence of the
enzyme in the absence of quencher [Q], F1 is the observed fluores-
cence at a given quencher concentration, fa is the fractional degree
of fluorescence, and Kd is the dissociation constant].
F1
F0
fa½QŠ
ð1Þ
1 À
¼
Kd þ ½QŠ
Acknowledgements
This work was supported by the Spanish MINECO contracts
CTQ2013-44367-C2-1-P (to P.M.), CTQ2013-44367-C2-2-P (to
R.H.G.), and CTQ2011-22724 and CTQ2014-57141-R (to S.M.S.).
We also acknowledge the Government of Aragón (Spain) (Bio-
organic Chemistry group E-10 and Protein Targets group B-89)
for financial support. We acknowledge the Institute of Biocom-
putation and Physics of Complex Systems (BIFI) at the Universi-
ty of Zaragoza (Spain) for computer time at clusters Terminus
and Memento. We thank synchrotron radiation sources DLS
(Oxford) and, in particular, beamline I04 (experiment number
MX8035-26). M.G. thanks the Spanish Ministry of Education
(MEC) for a predoctoral grant (FPU program). The European
Commission is gratefully acknowledged (Grant GLYCOPHARM
FP7-PITN-GA-2012-317297 and BioStruct-X grant agreement
no. 283570 and BIOSTRUCTX_5186).
Computational studies
Macromolecule preparation: Two X-ray crystal structures of
GalNAc-T2 were selected: one in the closed active conformation
(PDB entry 4D0Z, in complex with UDP-5S-GalNAc, EA2 peptide
and Mn2+), and the second one in the open inactive conformation
(PDB entry 2FFV, in complex with UDP and Mn2+). Chain A was se-
lected in both PDB structures. All water molecules, ethylene glycol
molecules and ligands were removed, but Mn2+ ions were re-
tained. By using the Maestro package,[40] missing residues were
added and modelled, and neutral terminal N and C groups and hy-
drogen atoms were added. Charges were assigned to all atoms
with the OPLS_2005 force field, and the final protein structure was
finally minimised with the same force field.
Ligand preparation: The 3D structure for UDP was extracted from
PDB 2FFV. The 3D structure for UDP-GalNAc was built from the 3D
structure of UDP-5S-GalNAc from PDB 4D0Z. UDP-5S-GalNAc was
also prepared for docking as a reference compound. The 3D struc-
tures of 16 compounds were built from their SMILES codes. All li-
gands were subjected to geometry optimisation by using the
MMFFs force field implemented in Maestro. This method was
chosen by comparing the docking results for reference compounds
by applying different force fields.
Keywords: carbohydrates
·
glycomimetics
·
glycosyltransferases · molecular modeling · proteins
[1] a) J. Voglmeir, S. L. Flitsch, in Science of Synthesis, Biocatalysis in Organic
Synthesis (Eds.: K. Faber,W.-D. Fessner, N. J. Turner), Thieme, Stuttgart,
2015, Vol. 1, pp. 507–542; b) T. M. Gloster, Curr. Opin. Struct. Biol. 2014,
28, 131–141; c) R. Roychoudhury, N. L. B. Pohl, Curr. Opin. Chem. Biol.
Docking calculations: Docking was performed by means of the
Glide[41] and AutoDock4 programs.[42] In Glide, a cubic grid box of
Chem. Eur. J. 2016, 22, 7215 – 7224
www.chemeurj.org
7223
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
2010, 14, 168–173; d) L. L. Lairson, B. Henrissat, G. J. Davies, S. G. With-
ers Ann. Rev. Biochem. 2008, 77, 521–555.
[2] X. Chen, ACS Chem. Biol. 2011, 6, 14–17.
nado, P. Bruscolini, H. Clausen, A. Lostao, F. Corzana, R. Hurtado-Guer-
rero, Nat. Commun. 2015, 6, 6937; b) B. L. Cantarel, P. M. Coutinho, C.
Rancurel, T. Bernard, V. Lombard, B. Henrissat, Nucleic Acids Res. 2009,
37, D233–D238; c) K. G. T. Hagen, T. A. Fritz, L. A. Tabak, Glycobiology
2003, 13, 1R–16R.
[3] a) S. Wang, S. Vidal, Carbohydr. Chem. 2013, 368, 78–101; b) P. Merino,
T. Tejero, I. Delso, R. Hurtado-Guerrero, A. Gomez-SanJuan, D. Sadaba,
Mini-Rev. Med. Chem. 2012, 12, 1455–1464; c) T. Kajimoto, M. Node, Syn-
thesis 2009, 3179–3210; d) L. J. Whalen, W. A. Greenberg, M. L. Mitchell,
C.-H. Wong in Iminosugars (Eds.: P. Compain, O. R. Martin), Wiley, Hobo-
ken, 2007, pp. 153–176; e) K.-H. Jung, R. R. Schmidt in Carbohydrate-
Based Drug Discovery, Vol. 2 (Ed.: C.-H. Wong), Wiley-VCH, Weinheim,
2003, pp. 609–659; f) X. Qian, M. M. Palcic, Carbohydr. Chem. Biol. 2000,
3, 293–312; g) L. Tedaldi, G. K. Wagner, MedChemComm 2014, 5, 1106–
1125.
[16] H. Gómez, R. Rojas, D. Patel, L. A. Tabak, J. M. Lluch, L. Masgrau, Org.
Biomol. Chem. 2014, 12, 2645–2655.
[17] A. Dondoni, A. Marra, Org. Biomol. Chem. 2015, 13, 2212–2215.
[18] a) D. Horton, T. Miyake, Carbohydr. Res. 1988, 184, 221–229; b) J. R.
Kramer, T. J. Deming, J. Am. Chem. Soc. 2010, 132, 15068–15071.
[19] R. Y. Tam, S. S. Ferreira, P. Czechura, J. L. Chaytor, R. N. Ben, J. Am. Chem.
Soc. 2008, 130, 17494–17501.
ˇ
ˇ
[20] a) A. Babic, S. Gobec, C. Gravier-Pelletier, Y. Le Merrer, S. Pecar, Tetrahe-
dron 2008, 64, 9093–9100; b) J. Cui, D. Horton, Carbohydr. Res. 1998,
309, 319–330.
[4] D.-M. Liang, J.-H. Liu, H. Wu, B.-B. Wang, H.-J. Zhu, J.-J. Qiao, Chem. Soc.
Rev. 2015, 44, 8350–8374.
[5] a) D. Albesa-JovØ, M. F. Mendoza, A. Rodrigo-Unzueta, F. Gomollón-Bel,
J. Cifuente, S. Urresti, N. Comino, H. Gómez, J. Romero-García, J. M.
Lluch, E. Sancho-Vaello, X. BiarnØs, A. Planas, P. Merino, L. Masgrau, M. E.
Guerin, Angew. Chem. Int. Ed. 2015, 54, 9898–9902; Angew. Chem.
2015, 127, 10036–10040; b) A. Ard›vol, C. Rovira, J. Am. Chem. Soc.
2015, 137, 7528–7547; c) H. Gómez, I. Polyak, W. Thiel, J. M. Lluch, L.
Masgrau, J. Am. Chem. Soc. 2012, 134, 4743–4752; d) R. Hurtado-Guer-
rero, G. J. Davies, Curr. Opin. Chem. Biol. 2012, 16, 479–487; e) I. Tvaros-
ka, Mini-Rev. Org. Chem. 2011, 8, 263–269; f) D. L. Jakeman, ChemBio-
Chem 2011, 12, 2540–2542.
[6] J. Rini, J. Esko, A. Varki in Essentials of Glicobiology. 2nd ed. (Eds.: A.
Varki, R. D. Cummings, J. D. Esko, H. H. Freeze, P. Stanley, C. R. Bertozzi,
G. W. Hart, M. E. Etzler), Cold Spring Harbor Laboratory Press, Cold
Spring Harbor, NY, 2009, Available from: http://www.ncbi.nlm.nih.gov/
books/NBK1921/.
[21] S. Liu, R. N. Ben, Org. Lett. 2005, 7, 2385–2388.
[22] a) V. R. Bouvet, R. N. Ben, J. Org. Chem. 2006, 71, 3619–3622; b) C. H.
Rçhrig, M. Takhi, R. R. Schmidt, Synlett 2001, 1170–1172.
[23] a) M. J. Damha, N. Usman, K. K. Ogilvie, Can. J. Chem. 1989, 67, 831–
839; b) P. Barthelemy, C. A. H. Prata, S. F. Filocamo, C. E. Immoos, B. W.
Maynor, S. A. N. Hashmi, S. J. Lee, M. W. Grinstaff, Chem. Commun. 2005,
1261–1263.
[24] A. Dondoni, S. Staderini, A. Marra, Eur. J. Org. Chem. 2013, 5370–5375.
[25] H. Pizova, P. Bobal, Tetrahedron Lett. 2015, 56, 2014–2017.
[26] P. J. Casara, K. C. Jund, A. Clauss, J.-F. Nave, R. D. Snyder, Bioorg. Med.
Chem. Lett. 1992, 2, 145–148.
[27] M. Izumi, K. Wada, H. Yuasa, H. Hashimoto, J. Org. Chem. 2005, 70,
8817–8824.
[28] G. W. Daub, E. E. J. Van Tameken, J. Am. Chem. Soc. 1977, 99, 3526–
3528.
[7] V. Malik, W. Black Gary, Adv. Protein Chem. Struct. Biol. 2012, 87, 87–115.
[8] a) K. Descroix, G. K. Wagner, Org. Biomol. Chem. 2011, 9, 1855–1863;
b) M. S. M. Timmer, M. V. Chumillas, W. E. Donker-Koopman, J. Alerts,
G. A. van der Marel, H. S. Overkleeft, J. H. van Boom, J. Carbohydr. Chem.
2005, 24, 335–351; c) H. C. Dorfmueller, V. S. Borodkin, D. E. Blair, S.
Pathak, I. Navratilova, D. M. F. van Aalten, Amino Acids 2011, 40, 781–
792; d) F. Stolz, M. Reiner, A. Blume, W. Reutter, R. R. Schmidt, J. Org.
Chem. 2004, 69, 665–679.
[29] J.-M. Campagne, J. Coste, P. Jouin, J. Org. Chem. 1995, 60, 5214–5223.
[30] E. P. Bennett, U. Mandel, H. Clausen, T. A. Gerken, T. A. Fritz, L. A. Tabak,
Glycobiology 2012, 22, 736–756.
[31] E. Lira-Navarrete, J. Iglesias-Fernandez, W. F. Zandberg, I. Companon, Y.
Kong, F. Corzana, B. M. Pinto, H. Clausen, J. M. Peregrina, D. J. Vocadlo,
C. Rovira, R. Hurtado-Guerrero, Angew. Chem. Int. Ed. 2014, 53, 8206–
8210; Angew. Chem. 2014, 126, 8345–8349.
[32] T. A. Fritz, J. Raman, L. A. Tabak, J. Biol. Chem. 2006, 281, 8613–8619.
[33] C. Andreini, I. Bertini, G. Cavallaro, G. L. Holliday, J. M. Thornton, J. Biol.
Inorg. Chem. 2008, 13, 1205–1218.
[9] J. Jiang, M. B. Lazarus, L. Pasquina, P. Sliz, S. Walker, Nat. Chem. Biol.
2012, 8, 72–77.
[10] a) R. Jørgensen, L. L. Grimm, N. Sindhuwinata, T. Peters, M. M. Palcic,
Angew. Chem. Int. Ed. 2012, 51, 4171–4175; Angew. Chem. 2012, 124,
4247–4251; b) K. Schaefer, J. Albers, N. Sindhuwinata, T. Peters, B.
Meyer, ChemBioChem 2012, 13, 443–450; c) J. C. Errey, S. S. Lee, R. P.
Gibson, C. M. Fleites, C. S. Barry, P. M. J. Jung, A. C. O’Sullivan, B. G.
Davis, G. J. Davies, Angew. Chem. Int. Ed. 2010, 49, 1234–1237; Angew.
Chem. 2010, 122, 1256–1259; d) S. Wang, J. A. Cuesta-Seijo, D. Lafont,
M. M. Palcic, S. Vidal, Chem. Eur. J. 2013, 19, 15346–15357; e) S. Wang,
J. A. Cuesta-Seijo, A. Striebeck, D. Lafont, M. M. Palcic, S. Vidal, ChemPlu-
sChem 2015, 80, 1525–1532; f) S. Wang, D. L. Shen, D. Lafont, A.-S. Ver-
coutter-Edouart, M. Mortuaire, Y. Shi, O. Maniti, A. Girard-Egrot, T. Lefeb-
vre, B. M. Pinto, D. Vocadlo, S. Vidal, MedChemComm 2014, 5, 1172–
1178; g) J. Jiang, V. Kanabar, B. Padilla, F. Man, S. C. Pitchford, C. P. Page,
G. K. Wagner Chem. Commun 2016, 52, 3955–3958.
[34] C. Andreini, I. Bertini, G. Cavallaro, G. L. Holliday, J. M. Thornton, J. Biol.
Inorg. Chem. 2008, 13, 1205–1218.
[35] W. Kabsch, Acta Crystallogr. Sect. D 2010, 66, 125–132.
[36] M. D. Winn, C. C. Ballard, K. D. Cowtan, E. J. Dodson, P. Emsley, P. R.
Evans, R. M. Keegan, E. B. Krissinel, A. G. Leslie, A. McCoy, S. J. McNicho-
las, G. N. Murshudov, N. S. Pannu, E. A. Potterton, H. R. Powell, R. J.
Read, A. Vagin, K. S. Wilson, Acta Crystallogr. Sect. D 2011, 67, 235–242.
[37] Collaborative Computational Project, Number 4, Acta Crystallogr. Sect. D
1994, 50, 760–763.
[38] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D 2004, 60, 2126–2132.
[39] G. N. Murshudov, A. A. P. Skubµk, N. S. Lebedev, R. A. Pannu, R. A. Stein-
er, M. D. Nicholls, F. L. Winn, A. A. Vagin, Acta Crystallogr. Sect. D 2011,
67, 355–367.
[40] Schrçdinger Release 2014-4, Maestro, version 10.0, Schrçdinger, LLC,
New York, NY, 2014.
[11] T. M. Gloster, W. F. Zandberg, J. E. Heinonen, D. L. Shen, L.-H. Deng, D. J.
Vocadlo, Nat. Chem. Biol. 2011, 7, 174–181.
[12] A. Schäfer, J. Thiem, J. Org. Chem. 2000, 65, 24–29.
[41] Glide, version 6.5, Schrçdinger, LLC, New York, NY, 2014.
[42] G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart, R. K.
Belew, A. J. Olson, J. Comput. Chem. 1998, 19, 1639–1662.
[43] D. A. Case, T. A. Darden, T. E. Cheatham, C. L. Simmerling, J. Wang, R. E.
Duke, R. Luo, R. C. Walker, W. Zhang, K. M. Merz, B. Roberts, S. Hayik, A.
Roitberg, G. Seabra, J. Swails, A. W. Gçtz, I. Kolossvµry, K. F.Wong, F. Pae-
sani, J. Vanicek, R. M.Wolf, J. Liu, X. Wu, S. R. Brozell, T. Steinbrecher, H.
Gohlke, Q. Cai, X. Ye, J. Wang, M.-J. Hsieh, G. Cui, D. R. Roe, D. H. Math-
ews, M. G. Seetin, R. Salomon-Ferrer, C. Sagui, V. Babin, T. Luchko, S. Gu-
sarov, A. Kovalenko, P. A. Kollman, AMBER 12, 2012, University of Califor-
nia, San Francisco.
[13] a) R. Chang, T.-T. Vo, N. S. Finney, Carbohydr. Res. 2006, 341, 1998–2004;
b) J. Hajduch, G. Nam, E. J. Kim, R. Froehlich, J. A. Hanover, K. L. Kirk,
Carbohydr. Res. 2008, 343, 189–195; c) A. J. Clarke, R. Hurtado-Guerrero,
S. Pathak, A. W. Schuttelkopf, V. Borodkin, S. M. Shepherd, A. F. M. Ibra-
him, D. M. F. van Aalten, Embo J. 2008, 27, 2780–2788; d) S. K. Partha,
A. Sadeghi-Khomami, K. Slowski, T. Kotake, N. R. Thomas, D. L. Jakeman,
D. A. R. Sanders, J. Mol. Biol. 2010, 403, 578–590; e) S. M. Forget, A. Jee,
D. A. Smithen, R. Jagdhane, S. Anjum, S. A. Beaton, D. R. J. Palmer, R. T.
Syvitskia, D. L. Jakeman, Org. Biomol. Chem. 2015, 13, 866–875.
[14] S. Vidal, I. Bruyere, A. Malleron, C. Auge, J.-P. Praly, Bioorg. Med. Chem.
2006, 14, 7293–7301.
[15] a) E. Lira-Navarrete, M. d. L. Rivas, I. Companon, M. C. Pallares, Y. Kong, J.
Iglesias-Fernandez, G. a. J. L. Bernardes, J. M. Peregrina, C. Rovira, P. Ber-
Received: January 31, 2016
Published online on April 13, 2016
Chem. Eur. J. 2016, 22, 7215 – 7224
www.chemeurj.org
7224
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim