New fluorinated fructose analogs as selective probes of the hexose transporter protein GLUT5

Article abstract of DOI:10.1039/c5ob00314h

Facilitated hexose transporters (GLUTs) mediate the transport of hexoses and other substrates across the membranes of numerous cell types, and while some are expressed ubiquitously (e.g., GLUT1), others are more tissue specific (e.g., GLUT5). These properties have been exploited for the imaging of cancer cells by the use of hexose based probes, including fluorinated hexose derivatives for use with positron emission tomography (PET). However, design of new probes has been hampered by a limited understanding of how GLUT transporters interact with their substrates at the molecular level. Two fluorinated fructose surrogates designed for uptake by the GLUT5 transporter are described here: 3-deoxy-3-fluoro-d-fructose (3-FDF) and 1-deoxy-1-fluoro-2,5-anhydromannitol (1-FDAM). Synthesis (both cold and radiolabeled) and in vitro analysis of their transport characteristics in two breast cancer cell lines (EMT-6 and MCF-7) expressing GLUT5 are detailed. Both analogues are readily taken up into both cancer cell lines, with uptake mediated primarily by GLUT5. They also have low IC₅₀ values, indicating a high affinity for the transporter, suggesting that the uptake of these probes would be unaffected by endogenously circulating fructose. Selective uptake by GLUT5 was also demonstrated in Xenopus oocytes. Finally, these results are the first demonstration that a hexose existing predominantly in the pyranose ring structure (3-FDF) is transported by GLUT5, strongly suggesting that this transporter can handle both furanose and pyranose forms of fructose.

Full text of DOI:10.1039/c5ob00314h

Organic &
Biomolecular Chemistry
View Article Online
View Journal
PAPER
New fluorinated fructose analogs as selective
probes of the hexose transporter protein GLUT5†
Cite this: DOI: 10.1039/c5ob00314h
Olivier-Mohamad Soueidan,a,b Brendan J. Trayner,b Tina N. Grant,a,b
Jeff R. Henderson,a Frank Wuest,c F. G. West*a and Chris I. Cheeseman*b
Facilitated hexose transporters (GLUTs) mediate the transport of hexoses and other substrates across the
membranes of numerous cell types, and while some are expressed ubiquitously (e.g., GLUT1), others are
more tissue specific (e.g., GLUT5). These properties have been exploited for the imaging of cancer cells
by the use of hexose based probes, including fluorinated hexose derivatives for use with positron emission
tomography (PET). However, design of new probes has been hampered by a limited understanding of
how GLUT transporters interact with their substrates at the molecular level. Two fluorinated fructose sur-
rogates designed for uptake by the GLUT5 transporter are described here: 3-deoxy-3-fluoro-D-fructose
(3-FDF) and 1-deoxy-1-fluoro-2,5-anhydromannitol (1-FDAM). Synthesis (both cold and radiolabeled) and
in vitro analysis of their transport characteristics in two breast cancer cell lines (EMT-6 and MCF-7)
expressing GLUT5 are detailed. Both analogues are readily taken up into both cancer cell lines, with
uptake mediated primarily by GLUT5. They also have low IC50 values, indicating a high affinity for the
transporter, suggesting that the uptake of these probes would be unaffected by endogenously circulating
fructose. Selective uptake by GLUT5 was also demonstrated in Xenopus oocytes. Finally, these results are
the first demonstration that a hexose existing predominantly in the pyranose ring structure (3-FDF) is
transported by GLUT5, strongly suggesting that this transporter can handle both furanose and pyranose
forms of fructose.
Received 13th February 2015,
Accepted 11th May 2015
DOI: 10.1039/c5ob00314h
www.rsc.org/obc
current understanding of how they bind and transport their
substrates is very limited. All of the GLUTs appear to have a
Introduction
Facilitated hexose transporters (GLUTs), belonging to the gene basic structure of twelve transmembrane helices (TM’s)
family SLC2A, are responsible for the entry and exit of hexoses arranged in two bundles of six connected by a long intracellu-
and other substrates into and out of numerous cell types. lar loop between TM’s 6 and 7.3 Cysteine scanning studies
Some of these proteins are considered to be essential for basic suggest that four TM’s form an outer scaffold supporting the
cell metabolism and to that end mammalian cells express one remaining eight, regions of which come together to form an
or more of them in their plasma membranes at quite high aqueous pore containing the substrate binding site(s).4 Com-
levels. In addition, because of the differing metabolic needs of parison with other similar transport protein families, some of
specific cell and tissue types the various members of this gene which have been crystallized, suggests that transport is
family have different kinetic properties, are regulated in achieved by structural rearrangement of the protein folding
different ways and are expressed at varying levels.1 Thus, they such that the binding site is sequentially exposed to the
have the potential to be used as specific portals into target outside of the cell, then occluded and finally exposed to the
cells for imaging or therapeutic approaches.2 However, the inside.5–8 However, despite a significant number of studies
having been performed on this family of mammalian proteins,
it is only very recently that the crystal structure of GLUT1
aDepartment of Chemistry, 11227 Saskatchewan Drive University of Alberta,
(SLC2A1) itself been determined for the inward facing confor-
Edmonton, AB, Canada T6G 2G2. E-mail: frederick.west@ualberta.ca;
mation.9 The arrangement of the TM’s is almost identical to a
Fax: +1 780 492 8231; Tel: +1 780 492 8187
bDepartment of Physiology, 7-55 Medical Sciences Building, University of Alberta,
Edmonton, AB, Canada T6G 2H7. E-mail: chris.cheeseman@ualberta.ca;
Fax: +1 780 492 8915; Tel: +1 780 492 4955
previously reported structure for XLe, a bacterial homologue of
the human GLUT family and together these structures will
allow for a much better understanding of how substrates are
bound and translocated by these proteins. A number of key
amino acid motifs, contributed by several of the pore lining
TM’s, have been proposed to contribute to the binding site for
cDepartment of Oncology, Cross Cancer Institute, University of Alberta, Edmonton,
AB, Canada T6G 1Z2
†Electronic supplementary information (ESI) available: 1H, 13C and 19F NMR
spectra for 4–5 and 7–12. See DOI: 10.1039/c5ob00314h
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
the substrates. These include a QQLS sequence in TM7 which ability of 3-deoxy-3-fluoro-D-fructose (3-FDF) 12 to bind and be
is conserved almost completely throughout the GLUT family. transported by the previously characterized GLUT5 expressing
In addition, the pairing of hydrophobic residues across the cell lines EMT-6 and MCF-7 was of great interest. These experi-
exterior vestibule has been suggested to contribute to substrate ments were designed to establish the tolerance of GLUT5 to
selectivity.6 However, in no case has the full substrate binding fructose substitution at C-3, using breast cancer cell lines
pocket(s) of a GLUT yet been defined.
known to express high levels of this hexose transporter.
Competition experiments employing substrate analogues
The work of Yang et al. was also intriguing, as it suggested
have been successful in identifying key features of the hexose the possibility of using fluorinated derivatives of 2,5-anhydro-
structures involved in binding to the exo- and endofacial mannitol (2,5-AM) 1 as GLUT5 substrates.10 2,5-AM 1 is a C2-
elements of the pore. Pioneering studies done by Holman and symmetric fructose mimic permanently held in the furanose
colleagues examined the requirements for substrate binding to form and which can inhibit the transport of fructose. However,
the fructose transporter GLUT5 at both sides of the mem- its transport has never been directly measured, although,
brane. Using the ability of their analogues to inhibit fructose McQuade and co-workers recently showed that a 2,5-AM deriva-
transport they found that the fructofuranose ring structure, tive labeled with a nitrobenzoxadiazole fluorescent tag is trans-
usually assumed for this hexose, was not essential for binding ported by GLUT5 and/or GLUT2.15 A fluorinated 2,5-AM
and that related structures locked in the fructopyranose form derivative could show key advantages over a fructose analogue
could also bind. This led them to propose that the hydroxyls at for imaging of GLUT5 expressing tumors. 2,5-AM 1 possesses
position C1, C2, C3 and C4 approach the exofacial binding an affinity for GLUT5 comparable to that of fructose,
site, while the C5 and C6 hydroxyls are less involved in sub- suggesting that the furanose form might be a preferred struc-
strate recognition. This in turn suggested that most of the ture for the binding pocket [6-FDF also exists primarily in
interaction for recognition and binding involved hydrogen the furanose ring form13]; and being C2 symmetric, positions
bonding between the hydroxyls on the hexose and specific 1 and 6 are equivalent allowing for labeling at either position
amino acid residues lining the pore. They also proposed that while preserving affinity for GLUT512,15 and potentially permit-
C6 may lie within the pore and not normally interact with the ting phosphorylation by either KHK or hexokinase.10,16
exofacial binding site of GLUT5, and suggested that labeling
The 2,5-AM 1 and the 2,5-AM derivative 1-deoxy-1-fluoro-
fructose at position 6 would provide analogues that could be 2,5-anhydro-D-mannitol (1-FDAM) 5 were both synthesized, as
readily handled by GLUT5,6,10,11 and explicitly raised the possi- well as the fructose derivative 3-deoxy-3-fluoro-D-fructose (3-
bility fluorine substitution at position 6 could increase the FDF) 12 and herein the in vitro analysis of their transport
binding affinity of the compound. Importantly, though, this characteristics in two breast cancer cell lines expressing
seminal work focused only on binding but not transport.
GLUT5 is described. These data demonstrate that these fluori-
Using Holman’s work as a starting point, our initial efforts nated probes were transported into breast cancer cell lines pri-
focused on the synthesis and characterization of the fructose marily by the fructose transporter GLUT5 and provide the first
analogue 6-deoxy-6-fluoro-D-fructose (6-FDF) as a possible sub- demonstration that a hexose existing primarily in the pyranose
strate for GLUT5 with the goal of employing the 18F labeled form can be transported by GLUT5. In addition, this work pro-
compound as an imaging probe for use with PET. Sub- vides valuable information on the GLUT5 structural demands
sequently, we determined this compound to be handled and suggests potential routes for the development of new
readily by GLUT5 in two breast cell lines. We also demon- molecular imaging probes for breast cancer.
strated that it could be utilized for PET imaging of two murine
xenograft models of breast cancer, and that it had a favorable
clearance and dosimetry profile in a rat model.12,13 However,
although fluorine substitution at the 6 position allowed the
fructose derivative to enter the cells rapidly, this substitution
also blocked the ability of hexokinase to phosphorylate this ana-
Results and discussion
Synthesis of 1-deoxy-1-fluoro-2,5-anhydro-D-mannitol
(1-FDAM) 5
logue at the 6 position. Moreover, the lack of expression of keto- The synthesis of 2,5-anhydro-D-mannitol (2,5-AM) 1 was modi-
hexokinase (KHK) in these cells precluded phosphorylation at fied from the procedure reported by Horton and co-workers.17
the 1 position and hence no metabolic trapping occured.13
Briefly, treatment of D-glucosamine with sodium nitrite and
Given the favourable uptake and lack of metabolic trapping acidic acid in water generated 2,5-anhydro-D-mannose which
seen with 6-FDF, we chose to explore labeling at another posi- was subsequently reduced in situ with sodium borohydride to
tion on the ring structure, leaving C6 open for phosphoryl- afford 2,5-AM in 71% over two steps.
ation. Holman and coworkers suggested that replacement of
The synthesis of 1-deoxy-1-fluoro-2,5-anhydro-D-mannitol
the hydroxyl at the 3 position with a bulky group was not well (1-FDAM) 5 was accomplished in four steps starting from 1
tolerated by the transporter – and while this may be the case (Scheme 1).18 Selective protection of the primary C-1 hydroxyl
for the allyl derivatives which they studied, we reasoned that a with trityl chloride (TrCl) in pyridine at 90 °C for 3 h, followed
fluorine at position 3 may satisfy the size and hydrogen by global acetylation afforded the product 2 in moderate yield
bonding requirements for proper recognition and possible over the two steps. Subsequently, the trityl protecting group
translocation by GLUT5.10,11,14 Therefore, determining the was removed selectively by treatment with trifluoroacetic acid
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
Paper
Scheme 1 Synthesis of 1-deoxy-1-fluoro-2,5-anhydro-D-mannitol
(1-FDAM). Reagents and conditions: (a) TrCl, Pyr, 90 °C, 3 h then Ac2O, pyr,
0 °C-r.t., 12 h, 40%; (b) TFA (4–5% in CH2Cl2), Et3SiH, r.t., 20 min, 84%;
(c) Tf2O, pyr, CH2Cl2, −10 °C, 30 min then CsF, t-AmOH, 90 °C, 25 min,
80%; (d) NaOMe, MeOH, 30 min, 95%.
Scheme 2 Synthesis of 3-deoxy-3-fluoro-D-fructose (3-FDF). Reagents
and conditions: (a) Tf2O, pyr, CH2Cl2, −10 °C, 1 h then CsF, t-AmOH,
90 °C, 12 h, 77%; (b) 80% HCO2H, 60–70 °C, 24 h, 65%; (c) TPAP, NMO,
3 Å MS, CH2Cl2, r.t., 12 h, 94%; (d) 2-methanesulfonylbenzothiazole,
LiHMDS, THF, −78 °C, 45 min then DBU, THF, r.t., 1 h, 78%; (e) OsO4
(cat.), NMO, acetone : water (10 : 1), r.t., 12 h, 85%; (f) Pd(OH)2/C, H2,
MeOH, r.t., 12 h, 93%.
and triethylsilane in dichloromethane to afford primary
alcohol 3 in 84%. Triflation under standard conditions and
treatment with cesium fluoride in tert-amyl alcohol at 90 °C
gave the fluorinated compound 4 in 80% isolated yield.19
Finally, deprotection of the acetates using sodium methoxide
provided 1-FDAM 5 in 95% isolated yield.
[14C]-5, [14C]-D-glucosamine was used as starting material to
prepare first [14C]AM, then [14C]-5 (in 14% overall yield). For
[14C]-12, lactone
9
was olefinated using [14C]-2-(methyl-
Synthesis of 3-deoxy-3-fluoro-D-fructose (3-FDF) 12
sulfonyl)benzothiazole labeled specifically at the methyl posi-
tion, and the resulting [14C]-10 was carried through the
remaining steps to afford [14C]-12 in 70% overall from 9.
The synthesis of 3-deoxy-3-fluoro-D-fructose (3-FDF) 12 was
achieved in six steps starting from the known intermediate
methyl 3,5-di-O-benzyl-α-D-ribofuranoside
6
(Scheme 2).20
Inhibition of [14C]D-fructose transport with 2,5-AM 1
Briefly, the compound 6 provided the fluorinated compound 7
in 77% via triflation under standard conditions and treatment
with cesium fluoride in tert-amyl alcohol at 90 °C.19 Glycoside
hydrolysis with formic acid (80%) provided the compound 8 in
65%, which was subsequently oxidized in the presence of
TPAP and NMO in CH2Cl2 to give the compound 9 in 94%.
The latter was then converted to its corresponding olefin 10
via Julia–Lythgoe olefination in 78% yield over two steps.
Upjohn dihydroxylation in the presence of catalytic osmium
tetroxide and NMO gave compound 11 in 85%. Finally, de-
protection of the remaining benzyl protecting groups under
hydrogenolysis conditions afforded 3-FDF 12 in 93% yield.
Notably, NMR analysis indicated that 12 exists predominantly
in the pyranose form, raising the interesting question of
whether it would be transported effectively by GLUT5.
In order to understand the ability of 2,5-AM 1 and its deriva-
tives to bind to and be transported by GLUT5, 2,5-AM 1 was
used as a competitive inhibitor against [14C]D-fructose trans-
port (100 μM) and its IC50 determined. Fig. 1 shows inhibition
of [14C]D-fructose transport measured over 60 min in both of
the GLUT5 expressing EMT-6 and MCF-7 cell lines.12,13,21–23 In
both cell types 0.1–10 mM 2,5-AM 1 showed significant inhi-
bition of fructose with an IC50 of 1.06 0.58 µM in EMT-6 (n =
3) and 0.16 0.94 µM in MCF-7 cells (n = 3). At a concentration
of 10 mM 2,5-AM 1, transport of [14C]D-fructose was inhibited
by 62% in both EMT-6 and MCF-7 cells.
Inhibition of [14C]D-fructose transport by 1-FDAM 5
Fig. 2 shows initial experiments ascertaining the viability of
1-FDAM 5 as a substrate for GLUT5. The previously characterized
GLUT5 expressing breast cancer cell lines EMT-6 and MCF-7
Radiochemical synthesis
Syntheses of [14C]-labeled 1-FDAM 5 and 3-FDF 12 were carried were utilized for dose-dependent competitive inhibition of
out via variations on the routes described above. In the case of [14C]D-fructose uptake by 1-FDAM 5 and to determine the relative
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
Fig. 1 2,5-AM inhibition of [14C]D-fructose transport after a 60 min
incubation at 37 °C with both EMT-6 and MCF-7 using increasing con-
centrations of 2,5-AM. Fructose transport was inhibited by increasing
Fig. 3 3-FDF inhibition of [14C]D-fructose transport after a 60 min incu-
bation at 37 °C with both EMT-6 and MCF-7 using increasing concen-
trations of 3-FDF. Fructose transport was inhibited by increasing
●
concentrations of 2,5-AM, and the IC50 obtained for EMT-6 ( ) was
●
concentrations of 3-FDF, and the IC50 obtained for EMT-6 ( ) was 1.16
○
1.06 0.58 µM and 0.16 0.09 µM in MCF-7 ( ). Error bars represent
○
0.67 µM and 2.37 1.5 µM in MCF-7 ( ). Error bars represent the SEM.
the SEM (see ESI† for semi-log plots of the same data).
and with a IC50 of 1.16 0.67 µM in EMT-6 and 2.37 1.5 µM
in MCF-7. These values are comparable to those for 1-FDAM 5
and its parent compound 2,5-AM 1, providing preliminary
indication that it may be a high affinity substrate for GLUT5.
Uptake of [14C]1-FDAM ([14C]-5)
While it was observed that 1-FDAM 5 was able to competitively
inhibit the entry of [14C]D-fructose into the cell in a dose
dependent manner, it was not clear whether it was being trans-
located across the membrane, or simply binding to the extra-
cellular binding site of GLUT5 and thus blocking [14C]D-
fructose from the binding site and preventing its translocation.
To ascertain the ability of 1-FDAM 5 to be transported, uptake
of tracer concentrations (∼0.001 mM) of newly synthesized
[14C]1-FDAM (0.3 µCi ml−1) [14C]-5 was studied over time
(Fig. 4). Cells were incubated for 0, 1, 5, 10, 20, 30 and
60 minutes, and then lysed to determine how much com-
pound was internalized in each breast cancer cell line. After
60 min it was observed that EMT-6 had approximately 1.7
times more uptake per mg of protein than MCF-7 (0.48 nor-
malized CPM per mg protein, n = 3 vs. 0.29 normalized CPM
per mg protein, n = 3). Studies with EMT-6 cells elicited an
almost linear uptake while MCF-7 on the other hand began to
plateau after 30 minutes.
Fig. 2 1-FDAM inhibition of [14C]D-fructose transport after a 60 min
incubation at 37 °C with both EMT-6 and MCF-7 using increasing con-
centrations of 1-FDAM. Fructose transport was inhibited by increasing
●
concentrations of 1-FDAM, and the IC50 obtained for EMT-6 ( ) was
○
6.82 3.0 µM and 3.96 2.60 µM in MCF-7 ( ). Error bars represent
the SEM.
affinity for the transporter. After 60 min incubation with 0.1 to
10 mM non-radiolabeled 1-FDAM 5 in the extracellular media,
dose dependent inhibition of [14C]D-fructose uptake was
observed with an IC50 of 6.9 3.0 μM (n = 4) in EMT-6 and
3.96 2.60 μM (n = 4) in MCF-7 cells. At the highest examined
concentration of 10 mM 1-FDAM 5, there was 68% inhibition
of total [14C]D-fructose transport in EMT-6, and 76% in MCF-7 cells.
Uptake of [14C]3-FDF ([14C]-12)
Cell uptake experiments using 0.3 µCi ml−1 the [14C] labelled
3-FDF ([14C]-12) indicates that it is transported into both cell
lines, and that EMT-6 has higher uptake than that of MCF-7
which is agreement with previous findings of relative uptake
levels of [14C] fructose in both cell lines (Fig. 5). After
60 minutes EMT-6 had approximately 1.3 times greater uptake
Inhibition of [14C]D-fructose transport by 3-FDF 12
Fig. 3 illustrates that 3-FDF 12 is also able to inhibit [14C]D- of [14C]3-FDF ([14C]-12) than that of MCF-7 (0.30 normalized
fructose transport in a dose dependent manner (0.1–10 mM) CPM per mg protein, n = 4 vs. 0.23 normalized CPM per mg
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
Paper
14
○
[ C]1-FDAM 60 minute time course in both MCF-7 ( ) and
Fig. 4
●
EMT-6 ( ) at 37 °C. Uptake is observed in both cell types after a 60 min
incubation. Uptakes were corrected for non-carrier mediated fluxes.
Error bars represent the SEM.
Fig. 6 Effect of 50 µM cytochalasin B (CB) and 500 mM D-fructose on
the uptake of [14C]1-FDAM 5 into EMT-6 and MCF-7 cells. 37 °C incu-
bations lasted 60 min and uptakes were corrected for non-mediated
fluxes. Error bars represent the SEM. An unpaired t-test between was
found to be extremely significant between the uptake under control and
fructose treated conditions in both EMT6 and MCF-7 cell lines (n = 3,
unpaired t-test, p = 0.0002 (***)).
60 minute incubation, the uptake of [14C]1-FDAM [14C]-5 into
EMT-6 was inhibited 59% (n = 3) and into MCF-7 by 60% (n =
3) of control values. The relative ineffectiveness of D-fructose to
inhibit transport of this substrate suggests that like previously
described fluorinated GLUT5 substrates; they appear to have a
much higher affinity for the transporter than the natural sub-
strate D-fructose.
14
○
[ C]3-FDF 60 minute time course in both MCF-7 ( ) and EMT-6
Fig. 5
●
(
) at 37 °C. Uptake is observed in both cell types after a 60 min incu-
bation. Uptakes were corrected for non-carrier mediated fluxes. Error
bars represent the SEM.
protein, n = 4). Both uptake curves began to display plateau be-
havior within 30 minutes.
Inhibition of [14C]3-FDF ([14C]-12) uptake by cytochalasin B
and D-fructose
EMT-6 and MCF-7 cells were incubated in the presence of
0.3 µCi mL−1 of [14C]3-FDF ([14C]-12) and a 50 µM concentration
of the Class I GLUT inhibitor cytochalasin B (CB) for
Inhibition of [14C]1-FDAM ([14C]-5) uptake by cytochalasin B
and D-fructose
Cells were incubated in the presence of 0.3 µCi mL−1 of [14C]1- 60 minutes to measure the GLUT2 mediated component of
FDAM [14C]-5 and a 50 µM concentration of the Class I [14C]3-FDF flux (Fig. 6B). After 60 min incubation, uptake into
(GLUTs1–4) GLUT inhibitor cytochalasin B (CB) for 60 minutes EMT-6 cells in the presence of CB was inhibited 26% com-
to measure the GLUT2 mediated [14C]1-FDAM flux (Fig. 6). pared to the control (n = 3), and 28% in MCF-7 cells (n = 3). A
After 60 min incubation, uptake into EMT-6 cells in the pres- student’s t-test found no significant difference between the
ence of CB was inhibited only 12% compared to the control (n two conditions suggesting that the majority of the flux is
= 3), and uptake into MCF-7 cells was inhibited 11% (n = 3) mediated by GLUT5 and not by any of the class I GLUTs (i.e.
suggesting that GLUT5 mediates the majority of the flux GLUTs 1, 2, 3 or 4). After 60 minutes incubation, 500 mM
(88–89%), with GLUT2 playing at most a minor role. A Stu- D-fructose only inhibited uptake in both EMT-6 and MCF-7 cells
dent’s t-test found no significant difference between the by approximately 19% and 22% respectively, suggesting that
uptake under control and CB treated conditions in either cell 3-FDF 12 appears to have a much higher affinity for binding to
line indicating that uptake of [14C]1-FDAM [14C]-5 is primarily GLUT5 than D-fructose (Fig. 7).
mediated by GLUT5.
The objective of this study was to examine the ability of the
In order to further examine the characteristics of the ability fluorinated 2,5-AM derivative 1-FDAM 5 and the fructose ana-
of [14C]1-FDAM [14C]-5 to bind to GLUT5, transport was logue 3-FDF 12 to be transported into two models of breast
measured in the presence of 500 mM D-fructose to compare cancer, and characterize and compare the profile of 1-FDAM 5
the relative affinity of both substrates (Fig. 6). After a to that of 6-FDF in vitro. This work has identified that (i) both
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
and 3-FDF 12) suggesting that other fructose uptake mecha-
nisms might be present in these cells. The breast cancer cell
lines EMT6 and MCF-7 have been characterized to express
GLUT2 (CB sensitive) and GLUT7, 9 and 11 (CB insensitive), all
of which, in addition to GLUT5, are able to transport fructose
into the cells.6,13 As illustrated in Fig. 6 and 7, the cellular
influx of [14C]1-FDAM [14C]-5 and [14C]3-FDF [14C]-12 was mini-
mally influenced by the co-incubation with CB (specific inhibi-
tor for Class I GLUTs), suggesting that GLUT-2 is minimally
contributing to the 1-FDAM 5 and 3-FDF 12 transport. There-
fore, the remaining [14C]D-fructose (30–40%) could be
mediated by other fructose transporters such as GLUT7, 9
and 11.
Transport studies of the [14C] labelled substrates indicate
that 1-FDAM 5 is taken up into murine EMT-6 cells (0.48
0.06 corrected CPM per mg protein), as well as the human
MCF-7 cells (0.25 0.03 corrected CPM per mg protein). The
higher level of uptake in EMT-6 compared to MCF-7 cells is
consistent to what has been observed in previous studies on
the transport of 6-FDF and D-fructose.11,12
To verify the hypothesis that 1-FDAM 5 and 3-FDF 12
uptake is mediated by GLUT5, co-incubation of the Class I
GLUT inhibitor cytochalasin B (CB) with [14C]1-FDAM [14C]-5
and [14C]3-FDF [14C]-12 indicates that the majority of cellular
influx is for the most part CB insensitive, suggesting GLUT5 is
the major mediator of 1-FDAM and 3-FDF’s transport. This cor-
Fig. 7 Effect of 50 µM cytochalasin B (CB) and 500 mM D-fructose on
the uptake of [14C]3-FDF 12 into EMT-6 and MCF-7 cells. 37 °C incu-
bations lasted 60 min and uptakes were corrected for non-mediated
fluxes. Error bars represent the SEM. An unpaired t-test between was
found to be significant between the uptake under control and fructose
treated conditions in both EMT6 and MCF-7 cell lines (n = 3, unpaired
t-test, p = 0.03 (*)).
are transported into human MCF-7 and murine EMT-6 cells, relates with data for 6-FDF as described previously.13 Interest-
(ii) both 1-FDAM 5 and 3-FDF 12, like 2,5-AM 1, are high ingly, and in agreement with previous data looking at 6-[18F]
affinity substrates for the facilitative hexose transporter GLUT5 FDF transport inhibition with D-fructose,12 inhibiting [14C]1-
or other Class II GLUTs, (iii) a hexose existing predominantly FDAM [14C]-5 and [14C]3-FDF [14C]-12 transport with high con-
in the pyranose ring structure (3FDF) is readily transported by centrations of D-fructose is ineffective compared to 1-FDAM 5
GLUT5.
or 3-FDF 12 inhibition of [14C]D-fructose transport. In fact, it
2,5-AM 1 was synthesized to provide a benchmark to was necessary to use a very high concentration (0.5 M) of
1-FDAM 5 and compare how replacement of hydroxyl by fluorine D-fructose to observe significant inhibition of [14C]1-FDAM [14C]-
at position C-1 impacts transport and handling of 2,5-AM ana- 5 transport (see ESI†). This suggests that the binding affinity
logues via GLUT5. Inhibition studies show a marked decrease of both substrates for the transporter is much higher than the
in the affinity of 1-FDAM 5 versus that of 2,5-AM 1. 2,5-AM naturally occurring substrate.
showed an ability to inhibit fructose uptake at the very low
micromolar range (EMT-6: 1.06 0.58 µM and MCF-7: 0.16
To gain further insight into the uptake mechanism of these
probes into cells, we performed the relevant uptake experiment
0.09 µM) compared to an almost one order of magnitude of [14C]3-FDF [14C]-12 and [14C]D-fructose into Xenopus laevis
larger IC50 for 1-FDAM 5 (EMT-6: 7.11 3.2 µM and MCF-7: oocytes expressing GLUT5 (after pretreatment with
3.96 2.6 µM).
GLUT5 mRNA) and control oocytes (Fig. 8, see ESI† for more
The variability of the IC50 values between cell lines may details). The data clearly show significantly accelerated rate of
result from structural differences between murine and human uptake for both substrates in the eggs expressing the transpor-
GLUT5 (and hence their binding to the probe molecule), as ter protein (red and blue curves). These results indicate that a
well as the possible expression of other fructose transporting major portion of the uptake is GLUT5 dependent.
GLUTs that may also show some variable affinity for the sub-
1-FDAM 5 is very effective in binding to the extracellular ves-
strate. Due to the lack of specific inhibitors for the Class II and tibule in the initial steps of transport, as indicated by its very
III GLUTs (GLUTs 6, 8, 12 & HMIT), it is not possible to fully low IC50, but despite this favorable characteristic it is not
assess their contribution to overall transport. The superior being transported at the levels we previously observed with
ability of 2,5-AM 1 to inhibit fructose transport compared to 6-[14C]FDF.13 These observations are in agreement with the pre-
1-FDAM 5 suggests that the presence of hydroxyl groups at both liminary evaluation of [18F]1-FDAM [14C]-5 as a PET radiotracer
positions 1 and 6 increases the relative affinity for binding.
for breast cancer imaging, recently reported by Sun and co-
As shown in Fig. 1–3, a significant proportion of [14C]D-fruc- workers.18 This result offers an important reminder that high
tose (30–40%) remained even in the presence of high concen- affinity extracellular binding to the transporter is not in itself a
trations of all of the probe compounds (2,5-AM 1, 1-FDAM 5 good indicator of how rapidly a substrate will be translocated
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
Paper
anhydrous solvents and reagents was accomplished with oven-
dried syringes or cannula. Solvents were distilled before use:
methylene chloride (CH2Cl2) and tert-amyl alcohol from
calcium hydride, and pyridine from KOH. Thin layer chromato-
graphy was performed on glass plates precoated with 0.25 mm
Kieselgel 60 F254. Flash chromatography columns were packed
with 230–400 mesh silica gel. Optical rotations were measured
at 22
2 °C. Proton nuclear magnetic resonance spectra
(1H NMR) were recorded at 300 MHz, 400 MHz or 500 MHz
and coupling constants (J) are reported in hertz (Hz). Standard
notation was used to describe the multiplicity of signals
Fig. 8
[ [
14C]3-FDF and 14C]D-fructose 30 minute time courses in
1
observed in H NMR spectra: broad (br), multiplet (m), singlet
oocytes injected with GLUT5 mRNA or with water as a control experi-
ment. Each data point represents the average of 10 oocytes. Error bars
indicate SEM.
(s), doublet (d), triplet (t), etc. Carbon nuclear magnetic reson-
ance spectra (13C NMR) were recorded at 100 MHz or 125 MHz
and are reported (ppm) relative to the centre line of the
triplet from chloroform-d (77.00 ppm). Fluorine nuclear mag-
netic resonance spectra (19F NMR) were recorded at 377 MHz
and are reported (ppm) relative to trifluoroacetic acid
(−76.55 ppm). Infrared (IR) spectra were measured with a
Mattson Galaxy Series FT-IR 3000 spectrophotometer. Mass
spectra were determined on a PerSeptive Biosystems Mariner
high-resolution electrospray positive ion mode spectrometer.
into the cell. 6-FDF on the other hand may have less affinity
for the first steps of translocation, but is seems to be more
amenable to transport in the latter stages during confor-
mational change and facilitative transport via GLUT5. These
in vitro results may indicate that 1-FDAM 5 would have somewhat
lower uptake in GLUT5 expressing tumors in vivo compared to
6-FDF, but its uptake would not be impeded by the presence of
fructose in the plasma.
Procedure for the synthesis of 1-trityl-3,4,6-tri-O-acetyl-
2,5-anhydro-D-mannitol (2)
2,5-Anhydro-D-mannitol 1 (0.57 g, 3.4 mmol) was dissolved in
pyridine (6 mL) and trityl chloride (0.98 g, 3.5 mmol) was
added. Subsequently, the solution was equipped with a reflux
condenser and heated at 90 °C for 3 hours. The resulting
yellow solution was cooled to 0 °C and acetic anhydride (5 mL)
was added dropwise. The reaction was slowly warmed to room
temperature overnight and was quenched by the addition of
water (15 mL) and extracted with CH2Cl2 (3 × 20 mL). The com-
bined organic layers were washed with 5% aqueous H2SO4
(20 mL) and water (20 mL) and dried over anhydrous MgSO4.
Filtration and concentration in vacuo afforded a oil residue
that was purified via column chromatography (9 : 1 to 1 :
1 hexane : EtOAc) to afford 2 as a clear colourless oil that
matched previously reported characterization data2 (0.75 g,
1.4 mmol, 40% yield): 1H NMR (500 MHz, CDCl3): δ 7.48 (d, J =
7.5 Hz, 6H), 7.32 (t, J = 7.8 Hz, 6H), 7.25 (t, J = 7.3 Hz, 3H), 5.14
(t, J = 3.2 Hz, 1H), 5.14 (dd, J = 2.7, 4.3 Hz, 1H), 4.37–4.21
(m, 4H), 3.34 (dd, J = 4.4, 9.7 Hz, 1H), 3.30 (dd, J = 5.6, 9.8 Hz,
1H), 2.12 (s, 3H), 2.09 (s, 3H), 2.02 (s, 3H); 13C NMR (125 MHz,
CDCl3): δ 170.7, 169.9, 169.9, 143.7, 128.7, 127.8, 127.1, 86.9,
82.5, 80.9, 78.7, 78.5, 63.5, 63.4, 20.8, 20.8, 20.7.
Conclusion
In conclusion, 1-FDAM 5 and 3-FDF 12 are novel compounds
with the potential for imaging of GLUT5 expressing breast
cancer tissue using PET. GLUT5 mediated uptake with very
high affinity into both human and murine tissues has been
observed, suggesting their ability to be potential radiotracers
for imaging of these tissues. Their kinetic characteristics
suggest that low concentrations of these probes would be
readily taken up into tumors. Further work needs to be done
to analyze their transport characteristics in a human model of
breast cancer and their metabolic fate both in vitro and in vivo
to determine how effective their use with PET could be.12
Fructose transport via GLUT5 and its metabolism
appear to be important players in breast cancer growth and
proliferation.12,13,21–23 The characterization of 1-FDAM 5 and
3-FDF 12 has shown that fructose based probes can be readily
taken up into breast cancer cells via this and possibly other
fructose transporters, and in the case of 3-FDF 12 may not
require the presence of a furanose ring structure to be a suit-
able substrate. The effects of further structural modifications
of the fructose scaffold, based upon these results, are under
current study and will be reported in due course.
Procedure for the synthesis of 3,4,6-tri-O-acetyl-2,5-anhydro-
D-mannitol (3)
Trifluoroacetic acid (40 mL, 5% solution in CH2Cl2) was added
to a mixture of compound 2 (2.00 g, 3.74 mmol) and triethyl-
silane (0.72 mL, 4.5 mmol). The solution was stirred for
30 min and quenched with aqueous saturated NaHCO3. The
resulting mixture was extracted with CH2Cl2 (3 × 25 mL) and
Experimental section
General information
Reactions were carried out in flame-dried glassware under a the combined organic layers were dried over anhydrous
positive argon atmosphere unless otherwise stated. Transfer of MgSO4. Filtration and concentration in vacuo, followed by
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
flash column chromatography (9 : 1 to 1 : 1 hexane : EtOAc) in vacuo to afford 5 as a clear colourless oil that crystallized
afforded alcohol 3 as a clear colourless oil that matched pre- over time (0.27 g, 1.6 mmol, 95% yield): mp 84–86 °C; [α]D20
=
1
viously reported data24 (0.90 g, 3.1 mmol, 84% yield): H NMR +39.99 (c 0.78, MeOH); IR (film) 3365, 2924, 1456, 1119,
(500 MHz, CDCl3): δ 5.25–5.18 (m, 2H), 4.25 (s, 3H), 4.10 1058 cm−1 1H NMR (400 MHz, D2O): δ 4.71–4.50 (m, 2H),
;
(app q, J = 4.9 Hz, 1H), 3.85–3.76 (m, 2H), 2.28 (t, J = 6.1 Hz, 4.14–3.99 (m, 3H), 3.92 (m, 1H), 3.78 (dd, J = 2.7, 12.6 Hz, 1H),
1H), 2.11 (s, 6H), 2.10 (s, 3H); 13C NMR (125 MHz, CDCl3): 3.69 (dd, J = 5.5, 12.8 Hz, 1H); 13C NMR (100 MHz, CDCl3):
δ 170.6, 170.3, 170.0, 83.4, 80.7, 78.4, 77.8, 63.1, 62.0, 20.8, 20.7. δ 83.5 (d, JC–F = 168.9 Hz), 83.3, 81.4 (d, JC–F = 14.6 Hz), 76.8,
76.2 (d, JC–F = 8.2 Hz), 61.7; 19F NMR (400 MHz, CDCl3):
Procedure for the synthesis of 3,4,6-tri-O-acetyl-1-deoxy-
1-fluoro-2,5-anhydro-D-mannitol (4)
δ −229.8 (ddd, J = 24.5, 47.6, 47.6 Hz); HRMS (ESI, [M + Na+])
for C6H11O4FNa calcd 189.0534, found: m/z 189.0532.
3,4,6-Tri-O-acetyl-2,5-anhydro-D-mannitol 3 (0.61 g, 2.1 mmol)
was dissolved in freshly distilled CH2Cl2 (21 mL) and cooled to
Radiochemical synthesis of [14C]FDAM 5
−10 °C. Pyridine (0.24 mL, 2.9 mmol) and triflic anhydride The procedures described above were repeated using [14C]-
(0.39 mL, 2.3 mmol) were added dropwise and the resulting D-glucosamine as starting material. Final product [14C]-5
solution was mixed at −10 °C. After one hour, water (10 mL) (SA ∼0.3 μCi mL−1) was obtained in 14% yield overall, and was
was added and the solution was extracted with CH2Cl2 found to be identical by TLC to the cold material and
(3 × 20 mL). The combined organic layers were washed with homogeneous.
10% aqueous H2SO4 (2 × 20 mL), dried over anhydrous MgSO4,
filtered and concentrated to afford the triflate intermediate as
Procedure for the synthesis of methyl 3,5-di-O-benzyl-2-deoxy-
2-fluoro-α-D-arabinofuranoside (7)
a viscous yellow oil that was used without further purification
in the next reaction: [α]2D0 = +15.81 (c 0.95, CHCl3); IR (film) Methyl 3,5-di-O-benzyl-α-D-ribofuranoside19
6
(0.28 g,
1746, 1415, 1373, 1227, 1147, 1049, 956 cm−1
;
1H NMR 0.81 mmol) was dissolved in CH2Cl2 (8 mL) at room tempera-
(500 MHz, CDCl3): δ 5.20 (t, J = 2.5 Hz, 1H), 5.12 (dd, J = 2.5, ture. The temperature of the reaction mixture was dropped to
4.1 Hz, 1H), 4.68 (d, J = 4.2 Hz, 2H), 4.36–4.16 (m, 4H), 2.13 −10 °C. Pyridine (0.10 mL, 1.20 mmol) and trifluoromethane-
(s, 3H), 2.11 (s, 6H); 13C NMR (125 MHz, CDCl3): δ 170.5, sulfonic anhydride (0.15 mL, 0.89 mmol) were subsequently
170.1, 169.9, 118.4 (q, JC–F = 319.6 Hz), 81.6, 80.9, 78.0, added at low temperature via syringe. After 1 hour stirring at
77.7, 74.2, 62.6, 20.7, 20.6, 20.6; HRMS (ESI, [M + Na+]) for −10 °C, water was added to quench the reaction. After dilution
C13H17O10SF3Na calcd 445.0387, found: m/z 445.0380.
with CH2Cl2, the organic and aqueous layers were separated.
Crude triflate (≤2.09 mmol) was dissolved in distilled tert- The organic layer was washed with 10% H2SO4 aq. (2 × 20 mL)
amyl alcohol (6.3 mL) and cesium fluoride (0.93 g, 6.2 mmol) and saturated aqueous NaHCO3. The organic layer was then
was added as a single portion. The reaction was heated at dried over anhydrous MgSO4 and filtered before removing
95 °C for 25 minutes and then cooled to room temperature. the solvent in vacuo. The 2-O-triflyl product was obtained as
Water (10 mL) was added and the mixture was extracted with a yellow oil and used in the next step without further
CH2Cl2 (3 × 10 mL). The combined organic layers were then purification.
washed with water (10 mL), dried over anhydrous MgSO4, fil-
The crude material was re-dissolved in tert-amyl alcohol
tered and concentrated. Flash column chromatography (9 : 1 to (2.5 mL).18 Cesium fluoride (0.37 g, 2.4 mmol) was added in a
1 : 1 hexane : EtOAc) afforded the fluorinated product 4 as a single portion. The reaction mixture was then heated to 90 °C
clear, colourless oil (0.49 g, 1.7 mmol, 80% yield over two and allowed to stir overnight. After cooling to room tempera-
steps): [α]2D0 = +18.39 (c 0.82, CHCl3); IR (film) 1744, 1436, ture, water was added to quench the reaction. CH2Cl2 was
1232, 1041 cm−1 1H NMR (400 MHz, CDCl3): δ 5.20 (t, J = added to dilute the reaction mixture. The organic layer was
;
3.8 Hz, 1H), 5.15 (t, J = 3.4 Hz, 1H), 4.55 (dd, J = 3.8, 46.8 Hz, washed with water (3 × 10 mL), dried (MgSO4) and filtered
2H), 4.27–4.12 (m, 4H), 2.07 (s, 3H), 2.06 (s, 3H), 2.05 (s, 3H); before removing the solvent in vacuo. The crude material was
13C NMR (100 MHz, CDCl3): δ 170.5, 170.0, 169.9, 82.3 purified by column chromatography (8 : 2 hexane : EtOAc) to
(d, JC–F = 174.8 Hz), 81.8 (d, JC–F = 15.3 Hz), 81.0, 77.9, 77.4 provide pure methyl 3,5-di-O-benzyl-2-deoxy-2-fluoro-α-D-arabino-
(d, JC–F = 6.5 Hz), 63.0, 20.7, 20.6; 19F NMR (400 MHz, CDCl3): furanoside 7 (0.21 g, 0.62 mmol, 77% over two steps) as a
δ −230.1 (ddd, J = 25.1, 46.9, 46.9 Hz); HRMS (ESI, [M + Na+]) pale yellow oil. [α]D20 = +96.78 (c 0.8, CHCl3); IR (film) 3031,
for C12H17O7FNa calcd 315.0851, found: m/z 315.0851.
2918, 2865, 1454, 1365, 1103, 1056, 990 cm−1 1H NMR
;
(500 MHz, CDCl3): δ 7.38–7.28 (m, 10H), 5.07 (d, J = 12.0 Hz,
1H), 4.96 (dd, J = 2.0, 51.5 Hz, 1H), 4.68 (d, J = 12.0 Hz, 1H),
4.57 (ABq, ΔνAB = 14.7 Hz, JAB = 12.0 Hz, 2H), 4.55 (d, J =
12.0 Hz, 1H), 4.23 (app dt, J = 4.0, 5.5 Hz, 1H), 4.00 (br dd,
Procedure for the synthesis of 1-deoxy-1-fluoro-2,5-anhydro-
D-mannitol (5)
3,4,6-Tri-O-acetyl-1-deoxy-1-fluoro-2,5-anhydro-D-mannitol
4
(0.49 g, 1.7 mmol) was dissolved in anhydrous MeOH (15 mL) J = 5.5, 24.0 Hz, 1H), 3.64 (dd, J = 3.5, 10.5 Hz, 1H), 3.60 (dd,
and NaOMe in MeOH (1.5 M, 0.35 mL) was added dropwise. J = 5.5, 11.0 Hz, 1H), 3.42 (s, 3H); 13C NMR (125 MHz, CDCl3):
The solution was mixed at room temperature for 30 minutes δ 137.9, 137.2, 128.4, 128.3, 127.9 (2C), 127.7, 127.6, 106.5
and neutralized with the addition of Amberlite IR-120 (H+). (d, JC–F = 35.9 Hz), 99.6 (d, JC–F = 180.1 Hz), 82.9 (d, JC–F
=
The resin was filtered off and the filtrate was concentrated 25.4 Hz), 81.3 (d, JC–F = 3.8 Hz), 73.5, 72.5, 69.3, 54.9; 19F NMR
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
Paper
(469 MHz, CDCl3): δ −188.8 (ddd, J = 12.2, 24.4, 51.1 Hz); silica gel to remove particulate material (EtOAc). After rotary
HRMS (ESI, [M + Na]+) for C20H23FNaO4 calcd 369.1473, evaporation, a pale yellow oil was obtained. The material was
found: m/z 369.1481.
purified by column chromatography (silica gel, 2 : 1 hexane :
EtOAc) to afford 3,5-di-O-benzyl-2-deoxy-2-fluoro-D-arabino-
lactone 9 (2.23 g, 6.76 mmol, 94%) as a white solid: m.p.
53–56 °C; [α]2D0 = +82.83 (c 0.83, CHCl3); IR (film) 3064, 3021,
Procedure for the synthesis of 3,5-di-O-benzyl-2-deoxy-2-fluoro-
α/β-D-arabinose (8)
Methyl 3,5-di-O-benzyl-2-deoxy-2-fluoro-α-D-arabinofuranoside 2922, 2869, 1802, 1454, 1378, 1323, 1177, 1115, 1028 cm−1
;
7 (0.50 g, 1.4 mmol) was dissolved in 80% aqueous formic acid 1H NMR (400 MHz, CDCl3): δ 7.40–7.26 (m, 10H), 5.31 (dd, J =
(31 mL) at room temperature. The reaction flask was equipped 7.2, 51.6 Hz, 1H), 4.78 (d, J = 11.6 Hz, 1H), 4.60 (d, J = 11.2 Hz,
with a reflux condenser and heated to 60–70 °C. After over- 1H), 4.57 (d, J = 11.2 Hz, 1H), 4.53 (ddd, J = 7.2, 7.6, 17.6 Hz,
night stirring, the reaction was cooled to room temperature 1H), 4.50 (d, J = 12.0 Hz, 1H), 4.37 (ddd, J = 2.4, 3.6, 8.0 Hz,
and neutralized with the addition of aqueous NaOH (2N). The 1H), 3.76 (app td, J = 2.0, 11.6 Hz, 1H), 3.61 (dd, J = 4.0,
reaction was then diluted with CH2Cl2 and the organic/ 11.6 Hz, 1H); 13C NMR (100 MHz, CDCl3): δ 168.2 (d, JC–F
aqueous layers separated. The aqueous layer was extracted 22.3 Hz), 137.2, 136.3, 128.4, 128.3, 128.2, 127.9, 127.7, 127.6,
=
with CH2Cl2 and the combined organic layers washed with 91.4 (d, JC–F = 197.5 Hz), 78.1 (d, JC–F = 9.8 Hz), 77.6 (d, JC–F
=
saturated aqueous NaHCO3. The organic layer was dried 19.9 Hz), 73.3, 72.7, 67.0; HRMS (ESI, [M + Na]+) for
(MgSO4) and filtered before removing the solvent in vacuo. C19H19FNaO4 calcd 353.1160, found: m/z 353.1160.
The crude material was purified by column chromatography
Procedure for the synthesis of 4,6-di-O-benzyl-1,2,3-trideoxy-
3-fluoro-2-methylene-D-arabinofuranose (10)
(9 : 1 to 8 : 2 hexane : EtOAc), providing a 2 : 1 mixture of
α/β-anomers of 3,5-di-O-benzyl-2-deoxy-2-fluoro-D-arabinose 8
(0.31 g, 0.91 mmol, 65%) as a clear, colorless oil. [α]2D0 = +42.7 3,5-Di-O-benzyl-2-deoxy-2-fluoro-D-arabinolactone
9 (1.20 g,
(c 0.96, CHCl3); IR (film) 3412, 3031, 2924, 2869, 1454, 1367, 3.60 mmol) and 2-(methylsulfonyl)benzothiazole25,26 (0.94 g,
1097, 1028 cm−1; HRMS (ESI, [M + Na]+) for C19H21FNaO4 4.4 mmol) were dissolved in THF (15 mL). The temperature of
calcd 355.1316, found: m/z 355.1316.
the reaction mixture was then dropped to −78 °C. LiHMDS in
α-Anomer. 1H NMR (500 MHz, CDCl3): δ 7.40–7.28 (m, THF (14.7 mL, 8.70 mmol, 0.590 M) was subsequently added
10H), 5.48 (d, J = 10.5 Hz, 1H), 4.96 (dd, J = 1.0, 50.5 Hz, 1H), to the cold reaction via cannula over 10 minutes. The reaction
4.66 (d, J = 12.0 Hz, 1H), 4.60 (d, J = 12.0 Hz, 1H), 4.56 (s, 2H), was stirred for 45 minutes before the addition of acetic acid
4.47 (ddd, J = 4.5, 5.5, 5.5 Hz, 1H), 4.03 (dd, J = 4.5, 20.0 Hz, (0.65 mL). Upon warming to room temperature, the mixture
1H), 3.61 (dd, J = 5.5, 10.5 Hz, 1H), 3.53 (dd, J = 5.5, 10.5 Hz, was diluted with EtOAc and water. The organic and aqueous
1H), 3.10 (br s, 1H); Partial 13C NMR (125 MHz, CDCl3): layers were separated and the aqueous layer extracted with
δ 100.4 (d, JC–F = 34 Hz), 97.8 (d, JC–F = 182.9 Hz), 82.3 (d, JC–F
25.7 Hz), 82.1 (d, JC–F = 1.9 Hz), 73.4, 72.4, 69.7 (d, JC–F
=
=
EtOAc (3 × 20 mL). The combined organic layers were dried
(MgSO4) and filtered before removing the solvent in vacuo. The
0.75 Hz); 19F NMR (376 MHz, CDCl3): δ −189.7 (ddd, J = 10.3, cloudy yellow oil was then re-dissolved in THF (75 mL) and
19.7, 50.6 Hz).
DBU (1.08 mL, 7.24 mmol) was added via syringe. The reaction
7.40–7.28 was stirred for 1 hour before removing the solvent in vacuo to
β-Anomer. 1H NMR (500 MHz, CDCl3):
δ
(m, 10H), 5.31 (br s, 1H), 4.97 (dd, J = 4.5, 53.0 Hz, 1H), 4.66 provide a dark yellow oil. The crude material was purified by
(d, J = 12.0 Hz, 1H), 4.56 (s, 2H), 4.54 (dd, J = 12.0, 2.5 Hz, 1H), column chromatography (silica gel, 4 : 1 hexane : EtOAc) to
3.32 (app td, J = 5.0, 17.5 Hz, 1H), 4.12 (app td, J = 3.5, 5.0 Hz, afford the product 10 (0.92 g, 2.8 mmol, 78%) as a yellow oil:
1H), 3.92 (br s, 1H), 3.62–3.58 (m, 1H, obscured by α-anomer), IR (film) 3031, 2925, 2865, 1685, 1665, 1454, 1367, 1092,
1
3.50 (dd, J = 3.5, 10.5 Hz, 1H); Partial 13C NMR (125 MHz, 1028 cm−1; H NMR (300 MHz, CDCl3): δ 7.41–7.28 (m, 10H),
CDCl3): δ 96.0 (d, JC–F = 172.7 Hz), 95.3 (d, JC–F = 39.4 Hz), 80.3 5.36 (tdd, J = 1.5, 3.9, 54.9 Hz, 1H), 4.63 (AB quartet, ΔνAB
=
(d, JC–F = 23.5 Hz), 80.2 (d, JC–F = 7.5 Hz), 73.7, 72.1, 69.7; 30.0 Hz, JAB = 12.0 Hz, 2H), 4.64–4.58 (m, 3H), 4.39 (app q, J =
19F NMR (376 MHz, CDCl3): δ −202.8 (dd, J = 17.8, 52.5 Hz).
5.1 Hz, 1H), 4.35 (ddd, J = 1.2, 2.4, 3.6 Hz, 1H), 4.23 (ddd, J =
3.9, 5.1, 17.1 Hz, 1H), 3.63 (dd, J = 1.0, 5.1 Hz, 2H); 13C NMR
(125 MHz, CDCl3): δ 158.0 (d, JC–F = 18.3 Hz), 137.7, 136.9,
128.4, 128.3, 127.9, 127.7, 127.6, 127.6, 94.1 (d, JC–F = 185.3
Procedure for the synthesis of 3,5-di-O-benzyl-2-deoxy-2-fluoro-
D-arabinolactone (9)
3,5-Di-O-benzyl-2-deoxy-2-fluoro-α/β-D-arabinose
8
(2.40 g, Hz), 86.7 (d, JC–F = 6.1 Hz), 82.0 (d, JC–F = 5.1 Hz), 80.8 (d, JC–F =
7.20 mmol) was dissolved in CH2Cl2 (60 mL) at room tempera- 22.4 Hz), 73.3, 72.0, 68.9 (d, JC–F = 1.9 Hz); 19F NMR (376 MHz,
ture. 3 Å molecular sieves (1.8 g) were added, followed by CDCl3): δ −183.2 (dd, J = 54.5, 16.5 Hz); HRMS (ESI, [M + Na]+)
N-methylmorpholine N-oxide (1.30 g, 11.0 mmol) at room for C20H21FNaO3 calcd 351.1370, found: m/z 351.1373.
temperature. The reaction mixture was stirred for 10 minutes
Procedure for the synthesis of 4,6-di-O-benzyl-3-deoxy-
3-fluoro-D-fructose (11)
prior to the addition of tetrapropylammonium perruthenate
(0.13 g, 0.36 mmol) in a single portion. The black reaction
mixture was stirred at room temperature and monitored by 4,6-Di-O-benzyl-1,2,3-trideoxy-3-fluoro-D-arabino-hex-1-enofura-
TLC (silica gel, 2 : 1 hex : EtOAc). After overnight stirring, the nose 10 (0.92 g, 2.8 mmol) was dissolved in acetone (19 mL)
crude reaction mixture was filtered through a short plug of and water (1.9 mL). N-Methylmorpholine N-oxide (0.48 g,
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
4.1 mmol) was added in a single portion at room temperature. 50.6 Hz, ∼90%); HRMS (ESI, [M + Na]+) for C6H11FNaO5 calcd
Osmium tetroxide solution (0. 35 mL, 4 wt% solution) was 205.04827, found: m/z 205.04807.
then added via syringe and the reaction was allowed to stir at
room temperature for 24 hours. Water (10 mL) and EtOAc
(10 mL) were then added and the organic/aqueous layers sep-
Radiochemical synthesis of [14C]3-FDF 12
arated. The aqueous layer was extracted with EtOAc (2 × [14C]-2-(methylsulfonyl)benzothiazole was prepared by treat-
20 mL) and the combined organic layers were subsequently ment of 2-mercaptobenzothiazole with [14C]-methyl nosylate
washed with water (2 × 20 mL). The organic layer was dried via a variation of the published procedures.25,26 This methy-
(MgSO4) and filtered before removing the solvent in vacuo to lenating reagent was then used as per the procedures above to
afford a yellow oil. The crude material was purified by column provide [14C]-12 (SA ∼0.3 µCi mL−1) in 70% over 3 steps, found
chromatography (silica gel, 2 : 3 hexane : EtOAc) to provide a to be identical by TLC to the cold material and homogeneous.
1 : 1.1 mixture of α/β-anomers of 4,6-di-O-benzyl-3-deoxy-
3-fluoro-D-fructose 11 (0.86 g, 2.4 mmol, 85%) as a clear,
colorless oil: [α]2D0 = +44.32 (c 0.67, CHCl3); IR (film) 3403,
Cell culture and transport experiments
3032, 2926, 2870, 1454, 1366, 1072, 1028 cm−1; HRMS EMT-6 and MCF-7 cells were grown in a CO2 incubator at 37 °C
(ESI, [M + Na]+) for C20H23FNaO5 calcd 385.1421, found: m/z in Gibco DMEM/F12 media supplemented with 15 mM
385.1420.
HEPES, L-glutamine, 10% fetal bovine serum and 1% penicil-
α-Anomer. Partial 1H NMR (500 MHz, CDCl3): δ 7.38–7.27 lin/streptomycin with media renewal every 2 days. Uptake
(m, 10H), 5.09 (dd, J = 5.5, 53.5 Hz, 1H), 4.44 (td, J = 5.5, 18.0 studies were performed once allowing the cells to reach conflu-
Hz, 1H), 4.14 (ddd, J = 2.5, 2.5, 5.5 Hz, 1H), 3.62 (ddd, J = 1.5, ence in 12 well plates. One hour before the initiation of an
3.0, 10.5 Hz, 1H), 3.48 (dd, J = 2.5, 10.5 Hz, 1H), 2.09 (dd, J = experiment, the media was removed, the plate rinsed twice
5.5, 8.5 Hz, 1H); Partial 13C NMR (125 MHz, CDCl3): δ 104.5 with PBS and then replaced with glucose-free Krebs–Ringer
(d, JC–F = 25.8 Hz), 98.3 (d, JC–F = 187.1 Hz), 82.5 (d, JC–F
25.5 Hz), 80.7, 73.4, 72.4, 69.3, 63.5 (d, JC–F = 6.5 Hz); 19F NMR MgSO4, 25 mM NaHCO3, 70 μM CaCl2, pH 7.4). After the hour
long incubation period, the glucose-free Krebs–Ringer solution
=
solution (120 mM NaCl, 4 mM KCl, 1.2 mM KH2PO4, 2.5 mM,
(469 MHz, CDCl3): δ −195.2 (dd, J = 21.1, 51.6 Hz).
β-Anomer. Partial 1H NMR (500 MHz, CDCl3): δ 7.38–7.27 was removed, and 300 μL of a radiotracer containing Krebs–
(m, 10H), 4.99 (dd, J = 2.0, 51.5 Hz, 1H), 4.38 (app q, J = 5.0 Hz, Ringer solution was added to each well with a specific activity
1H), 4.10 (ddd, J = 2.0, 5.0, 21.0 Hz, 1H), 3.58 (dd, J = 5.5, 10.5 of 0.3 µCi mL−1 of 14C-labeled D-fructose (Moravek Biochemi-
Hz, 1H), 3.54 (ddd, J = 1.0, 5.5, 10.5 Hz, 1H), 1.91 (dd, J = 5.5, cals), 3-FDF (proprietary) or 1-FDAM (proprietary). This was
8.0 Hz, 1H); Partial 13C NMR (125 MHz, CDCl3): δ 101.6 left to incubate within the wells for specific periods of time in
(d, JC–F = 18.3 Hz), 95.3 (d, JC–F = 195.5 Hz), 80.6 (d, JC–F
=
a 37 °C incubator until the media was aspirated and rinsed
19.3 Hz), 80.0 (d, JC–F = 9.3 Hz), 73.7, 72.1, 69.7, 63.8; 19F NMR with ice-cold Krebs–Ringer to stop further transport. 500 μL of
(469 MHz, CDCl3): δ −201.3 (dd, J = 17.8, 53.4 Hz).
5% trichloroacetic acid was then added to each well lysing the
cells on a rotating rocker for an hour. The cell lysate was then
transferred into scintillation vials containing 4 mL of Scinti-
g, Safe™ liquid scintillation fluid for counting in a liquid scintil-
Procedure for the synthesis of 3-deoxy-3-fluoro-D-fructose (12)
4,6-Di-O-benzyl-3-deoxy-3-fluoro-D-fructose
11
(0.71
2.0 mmol) was dissolved in MeOH (50 mL). Palladium hydrox- lation counter (Beckman LS 6500 multi-purpose liquid
ide on carbon (0.15 g, 20% wt Pd(OH)2) was added in a single scintillation counter). Protein quantification was performed by
portion at room temperature. The round bottom flask was lysing the cells using Cellytic™ M (Sigma) and then perform-
then equipped with a hydrogen-filled balloon and the reaction ing a BCA protein assay (Pierce, Rockford, IL, U.S.A.) according
progress monitored by TLC (9 : 1 EtOAc : MeOH). After to the manufacturer’s specifications. Inhibition studies were
24 hours stirring at room temperature, the reaction mixture performed as described previously.12,13
was filtered through a short Celite pad to remove particulates.
The crude material was purified by column chromatography
(9 : 5 EtOAc : MeOH) to afford 3-deoxy-3-fluoro-D-fructose 12
Data analysis
(0.34 g, 1.86 mmol, 93%) as a white solid. In aqueous solution In the inhibition experiments performed, counts per minute
3-FDF exists as a mixture of 3 isomers in a ratio of 17 : 1 : 1. (CPM) were normalized to standards, background levels of
[α]2D0 = −91.88 (c 0.61, MeOH); IR (film) 3387, 2945, 1646, 1420, substrate subtracted and then plotted against the maximum
1163, 1052 cm−1 1H NMR (500 MHz, D2O) for the major uptake of the radiolabeled tracer (i.e. [14C]1-FDAM, [14C]3-FDF
;
isomer: δ 4.67 (dd, J = 10.0, 50.5 Hz, 1H), 4.15 (ddd, J = 3.5, or [14C]D-fructose). For time courses, values were corrected
9.5, 13.0 Hz, 1H), 4.05–4.03 (m, 2H), 3.71 (d, J = 13.0 Hz, 1H), with standards and to the protein levels present per well. IC50
3.69 (d, J = 12.0 Hz, 1H), 3.58 (d, J = 12.0 Hz, 1H); 13C NMR values (concentration at which half maximum inhibition of
(125 MHz, D2O): δ 97.2 (d, JC–F = 19.0 Hz), 89.3 (d, JC–F = 183.1 cellular uptake of a radiotracer was observed) were determined
Hz), 70.6 (d, JC–F = 8.5 Hz), 68.9 (d, JC–F = 17.5 Hz), 64.4, 64.3; using non-linear regression analysis in GraphPad Prism 5
19F NMR (469 MHz, D2O): δ −195.1 (dd, J = 24.4, 51.6 Hz, (GraphPad Software, San Diego, CA, USA), and significance
∼5%), −205.9 (dd, J = 18.8, 53.5 Hz, ∼5%), −209.1 (dd, J = 13.0, was determined at p < 0.05 using a Student’s t-test.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
Paper
10 J. Yang, J. Dowden, A. Tatibouët, Y. Hatanaka and
Author contributions
G. D. Holman, Biochem. J., 2002, 367, 533.
The manuscript was written with contributions from all 11 J. Girniene, A. Tatibouët, A. Sackus, J. Yang, G. D. Holman
authors. All authors have given approval to the final version of
the manuscript.
and P. Rollin, Carbohydr. Res., 2003, 338, 711.
12 M. Wuest, B. J. Trayner, T. N. Grant, H.-S. Jans,
J. R. Mercer, D. Murray, F. G. West, A. J. McEwan, F. Wuest
and C. I. Cheeseman, Nucl. Med. Biol., 2011, 38, 461.
13 B. J. Trayner, T. N. Grant, F. G. West and C. I. Cheeseman,
Bioorg. Med. Chem., 2009, 17, 5488.
Funding sources
This work was funded by a Collaborative Health Research 14 A. Tatibouët, J. Yang, C. Morin and G. D. Holman, Bioorg.
Project grant from NSERC (Canada) and Canadian Institutes of Med. Chem., 2000, 8, 1825–1833.
Health Research, and by and an operating grant from the 15 M. Tanasova, M. Plutschack, M. E. Muroski, S. J. Sturla,
Canadian Breast Cancer Foundation.
G. F. Strouse and D. T. McQuade, ChemBioChem, 2013, 14,
1263.
16 W. L. Dills, J. I. Geller, K. M. Gianola and G. M. McDonough,
Anal. Biochem., 1983, 133, 344.
17 D. Horton and K. D. Philips, Carbohydr. Res., 1973, 30, 367.
Acknowledgements
The authors wish to thank Kenneth Wong for technical 18 For a recent synthesis of this compound, see: B. Niu,
support.
X. Wen, Z. Jia, X. Wu, W. Guo and H. Sun, Chin. J. Chem.,
2013, 31, 1159.
19 D. W. Kim, D. S. Ahn, Y. H. Oh, S. Lee, H. S. Kil, S. J. Oh,
S. J. Lee, J. S. Kim, J. S. Ryu, D. H. Moon and D. Y. Chi,
J. Am. Chem. Soc., 2006, 128, 16394.
Notes and references
1 M. Mueckler and B. Thorens, Mol. Aspects Med., 2013, 34, 20 N.-S. Li, J. Lu and J. A. Piccirilli, Org. Lett., 2007, 9, 3009–
121. 3012.
2 D. T. McQuade, M. B. Plutschack and P. H. Seeberger, Org. 21 S. P. Zamora-León, D. W. Golde, I. I. Concha, C. I. Rivas,
Biomol. Chem., 2013, 11, 4909.
3 M. Mueckler, C. Caruso, S. A. Baldwin, M. Panico,
F. Delgado-López, J. Baselga, F. Nualart and J. C. Vera, Proc.
Natl. Acad. Sci. U. S. A., 1996, 93, 1847.
I. Blench, H. R. Morris, W. J. Allard, G. E. Lienhard and 22 J. Levi, Z. Cheng, O. Gheysens, M. Patel, C. T. Chan,
H. F. Lodish, Science, 1985, 229, 941.
4 M. Mueckler and C. Makepeace, Biochemistry, 2009, 48,
5934.
5 A. Salas-Burgos, P. Iserovich, F. Zuniga, J. C. Vera and
J. Fischbarg, Biophys. J., 2004, 87, 2990.
6 A. R. Manolescu, R. Augustin, K. Moley and C. Cheeseman,
Mol. Membr. Biol., 2007, 24, 455.
Y. Wang, M. Namavari and S. S. Gambhir, Bioconjugate
Chem., 2007, 18, 628.
23 A. Godoy, V. Ulloa, F. Rodríguez, K. Reinicke, A. J. Yañez,
L. García Mde, R. A. Medina, M. Carrasco, S. Barberis,
T. Castro, F. Martínez, X. Koch, J. C. Vera, M. T. Poblete,
C. D. Figueroa, B. Peruzzo, F. Pérez and F. Nualart, J. Cell.
Physiol., 2006, 207, 614.
7 V. Iancu, J. Zamoon, S. B. Woo, A. Aleshin and J. Y. Choe, 24 R. D. Guthrie, I. D. Jenkins, J. J. Watters, M. W. Wright and
Proc. Natl. Acad. Sci. U. S. A., 2013, 110, 17862. R. Yamasaki, Aust. J. Chem., 1982, 35, 2169.
8 M. G. Madeja, L. Sunb, N. Yanb and H. R. Kabacka, Proc. 25 P. R. Blakemore, D. K. H. Ho and W. M. Nap, Org. Biomol.
Natl. Acad. Sci. U. S. A., 2014, 111, E719–E727. Chem., 2005, 3, 1365.
9 D. Deng, C. Xu, P. Sun, J. Wu, C. Yan, M. Hu and N. Yan, 26 D. Gueyrard, R. Haddoub, A. Salem, N. S. Bacar and
Nature, 2014, 110, 121.
P. G. Goekjian, Synlett, 2005, 3, 52.
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem.