Identification and Optimization of the First Highly Selective GLUT1 Inhibitor BAY-876

Article abstract of DOI:10.1002/cmdc.201600276

Despite the long-known fact that the facilitative glucose transporter GLUT1 is one of the key players safeguarding the increase in glucose consumption of many tumor entities even under conditions of normal oxygen supply (known as the Warburg effect), only

Full text of DOI:10.1002/cmdc.201600276

DOI: 10.1002/cmdc.201600276
Full Papers
Identification and Optimization of the First Highly
Selective GLUT1 Inhibitor BAY-876
Holger Siebeneicher,[a] Arwed Cleve,[a] Hartmut Rehwinkel,[a] Roland Neuhaus,[a]
Iring Heisler,[b] Thomas Mꢀller,[b] Marcus Bauser,[a] and Bernd Buchmann*[a]
Despite the long-known fact that the facilitative glucose trans-
porter GLUT1 is one of the key players safeguarding the in-
crease in glucose consumption of many tumor entities even
under conditions of normal oxygen supply (known as the War-
burg effect), only few endeavors have been undertaken to find
this challenging potency and selectivity profile. The N-(1H-pyra-
zol-4-yl)quinoline-4-carboxamides were identified as an excel-
lent starting point for further compound optimization. After
extensive structure–activity relationship explorations, single-
digit nanomolar inhibitors with a selectivity factor of >100
against GLUT2, GLUT3, and GLUT4 were obtained. The most
promising compound, BAY-876 [N4-[1-(4-cyanobenzyl)-5-
methyl-3-(trifluoromethyl)-1H-pyrazol-4-yl]-7-fluoroquinoline-
2,4-dicarboxamide], showed good metabolic stability in vitro
and high oral bioavailability in vivo.
a
GLUT1-selective small-molecule inhibitor. Because other
transporters of the GLUT1 family are involved in crucial pro-
cesses, these transporters should not be addressed by such an
inhibitor. A high-throughput screen against a library of ~3 mil-
lion compounds was performed to find a small molecule with
Introduction
In 1930 Otto Warburg observed a phenomenon in which
cancer cells often metabolically switch from oxidative phos-
phorylation to glycolysis, even under normal oxygen supply.[1]
At first sight, the contradictory behavior of cancer cells with
regard to their increased demand on energy supply is of cru-
cial importance to feed the essential biochemical anabolic
pathways with higher amounts of central intermediates.[1f] To
fulfill the increased demand on these central intermediates,
the glycolytic rate is often up-regulated in tumor entities.[2] As
the cellular uptake of glucose is the first rate-limiting step in
the glycolytic process it is not surprising that the transporters
responsible for this uptake were found to be up-regulated in
both solid and hematological malignancies.[3]
1a,[16] can induce the GLUT1 overexpression.[17] Additionally,
there is a widely clinically applied diagnostic modality PET
imaging, which makes use of increased glucose uptake in
some types of cancer with [18F]fluoro-2-deoxyglucose (FDG).[18]
All these factors demonstrate the importance of GLUT1 func-
tion for cancer cell viability suggesting that an inhibition of
this transporter might be therapeutically beneficial in the treat-
ment of tumors with high glucose turnover.
Whereas GLUT1 is nearly ubiquitously expressed in all
normal tissues to maintain the basal glucose supply[19] the ex-
pression of some members of the GLUT family is more specific.
Those GLUTs can be involved in central processes like insulin
secretion in pancreas (GLUT2),[20] neuronal glucose uptake
(GLUT3),[21] and insulin-regulated transport of glucose in
muscle and fat cells (GLUT4).[22] To enable a therapeutic
window with a potential GLUT1 inhibitor selectivity within the
GLUT family is decisive for any possible cancer treatment with
this approach.
Of the two different classes of hexose transporters (SGLT:
sodium-dependent glucose transporter, GLUT: facilitative glu-
cose transporter[4]) the GLUTs were found to be overexpressed
in many tumors.[5] In particular, GLUT1 overexpression has
been reported in many types of human cancers, including
those of brain,[6] breast,[7] colon,[8] kidney,[9] lung,[10] ovary,[11]
and prostate,[12] and is correlated with advanced cancer stages
and poor clinical outcomes. It was demonstrated that the acti-
vation of certain oncogenes like c-myc,[13] KRAS,[10] BRAF,[14] and
p53,[15] and transcription factors like hypoxia inducible factor-
Several small-molecule GLUT1 inhibitors have already been
described in literature including resveratrol,[23] naringenin,[24]
phloretin,[25] WZB117,[26] salicylketoximes,[27] thiazolidinedione,[28]
STF-31,[29] pyrazolopyrimidines,[30] and phenylalanine amides.[31]
With their thiazolidinediones Wang et al. demonstrated the in-
hibition of [3H]-2-deoxy-d-glucose uptake in LNCaP cells and
the suppressive effects on the viability of LNCaP cells in MTT
assays.[28] Chan et al. used the principle of chemical synthetic
lethality to demonstrate the sensitivity of VHL deficient renal
cancer cells to glucose uptake inhibition by STF-31.[29] The
compound WZB117 was able to inhibit the glucose uptake in
A549 cancer cells and their cell proliferation in a dose-depen-
dent manner.[26] All these results underline the feasibility of
GLUT1 inhibition as cancer treatment.
[a] Dr. H. Siebeneicher, Dr. A. Cleve, Dr. H. Rehwinkel, Dr. R. Neuhaus,
Dr. M. Bauser, Dr. B. Buchmann
Bayer AG, Drug Discovery, 13353 Berlin (Germany)
E-mail: bernd.buchmann@bayer.com
[b] Dr. I. Heisler, Dr. T. Mꢀller
Bayer AG, Drug Discovery, 42096 Wuppertal (Germany)
Supporting information for this article can be found under http://
dx.doi.org/10.1002/cmdc.201600276.
ChemMedChem 2016, 11, 1 – 12
1
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
We decided to continue SAR investigation on the benzylic
moiety attached to the pyrazole keeping the quinoline frag-
ment constant (Table 1). Directly attaching the para-fluoro-
phenyl ring to the pyrazole led to a significant GLUT1 potency
loss (4). The same was observed with an elongated spacer (6)
so that a methylene link between the pyrazole and the para-
fluorophenyl ring was considered optimal. Ramification at the
methylene position (5) also decreased the potency. Substitu-
tion of the para-fluoro group at the phenyl was, however, fea-
sible and later on the para-cyano group revealed the most
promising data (7). With this moiety in hand further substitu-
tions were possible while maintaining single-digit nanomolar
potency. For instance, a pyridine nitrogen (8) or a fluorine in
ortho (9) or meta position (10) of the cyano function could be
introduced. Efforts to exchange the aromatic benzyl head at
the pyrazole for fully saturated systems only gave less potent
Results and Discussion
Therefore, an HTS screen was performed using a cell-based
assay with CellTiter-Gloꢂ[32] readout for ATP production. In
a
pairwise chemical screen, DLD1 (for human GLUT1),
DLD1Glut1ꢀ/ꢀ (Horizon discovery, for human GLUT3), CHO-
hGLUT2 and CHO-hGLUT4 cell lines were co-incubated with
the potential GLUT1 inhibiting test compounds and with rote-
none as inhibitor for the oxidative phosphorylation.[33] With the
rotenone co-incubation the cells could only produce ATP via
glycolysis and the amount of produced ATP could be linked to
the amount of glucose taken up. Subsequent assays regarding
cellular uptake and consumption of glucose in the presence of
the test compounds confirmed their competitive GLUT1 inhibi-
tory behavior. In total, ~3 million compounds were tested in
this HTS and 285 hits were identified inhibiting the GLUT1
transporter and showing a good selectivity against GLUT2. In
addition, from these 285 hits two clusters and one singleton
demonstrated very good selectivity against GLUT3. The com-
parison of the two clusters and the singleton revealed the N-
(1H-pyrazol-4-yl)quinoline-4-carboxamide 1 (Figure 1) having
most promising primary data. To have a more lead-like starting
point for structure-activity relationship (SAR) optimization, we
first envisioned to decrease lipophilicity and possible metabolic
liabilities. Quickly, we found out that the 2-furanyl moiety was
not necessary for good potency and even a small methyl
group in position 2 was acceptable (2). Complete omission of
the quinoline methyl substituents in 2, however, deteriorated
the GLUT1 potency (IC50: 0.96 mm, not shown). First in vitro
pharmacokinetics (PK) data of 2 still revealed remaining meta-
bolic liabilities in this compound. Our first interest was then to
alter the methyl groups of the quinoline. The CF3 substituent
in position 2 and exchanging the 6,8-dimethyl pattern for a 6-
bromine significantly enhanced GLUT1 potency reaching the
single-digit nanomolar range (3). Furthermore, no detrimental
effect regarding selectivity toward GLUT2 and GLUT3 could be
observed and a selectivity factor of 35 toward GLUT4 was also
a promising starting point to explore the SAR of this com-
pound class in more detail.
compounds, for example, 11.
Compound 3 displayed relatively high clearance of about =
2
3
of liver blood flow in rats. Metabolite identification experi-
ments revealed remaining metabolic soft spots at the pyrazole
and quinoline parts (see Table 6 below). Therefore, we aimed
at modifying at least one of the pyrazole methyl groups and
abandoning the 6-bromine. Replacing a methyl with a CF3
group at pyrazole position 3 led to compound 12 already
showing slightly better clearance with 2.6 Lhꢀ1 kgꢀ1 in rat hep-
atocytes (see Figure 2 below). To be able to omit the 6-bromo
substituent completely in the final inhibitor avoiding in vivo
debromination or hepatotoxicity, we had to change the sub-
stituent in position 2 of the quinoline again, because within
the bromine-free 2-CF3-quinoline series we were not able to
find single-digit nanomolar inhibitors of GLUT1. For instance,
the debromo compound of 3 had only an IC50 of 0.093 mm in
the GLUT1 assay. Exchanging the 2-CF3 group in 12 for a pri-
mary carboxamide already resulted in a single-digit nanomolar
inhibitor with a des-bromo substance. Upon additional ex-
change of the para-fluoro group for a cyano moiety an IC50 of
0.006 mm was attained for 13. In addition, the metabolic stabili-
ty in rat hepatocytes was also increased by lowering the clear-
ance to 1.81 Lhꢀ1 kgꢀ1, however, the excellent selectivity of 12
Figure 1. Initial SAR beyond HTS hit 1.
&
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
2
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
lectivity (selectivity factor 134) than in positions 5 and 6. Intro-
duction of a methyl (20) decreased GLUT4 selectivity and a me-
thoxy group (21) resulted in a less potent compound. A halo-
gen at position 8 (22, 23) was also feasible, but the selectivity
profile was not better than in position 5 and 6.
Table 1. SAR investigations of the benzylic moiety at the pyrazole group
of compound 3.
Introduction of an additional ring nitrogen at position 6
(24), 7 (25), or 8 (26) gave double-digit nanomolar inhibitors
with less selectivity toward GLUT4 (selectivity factors: 6–28) as
compared to compound 19. Highly potent compounds could
be obtained upon double fluorine substitution (27, 28), but
again the GLUT4 selectivity of these inhibitors was impaired.
An exchange of the phenyl part of the quinoline for a five-
membered heterocycle (29) or complete omission of the
phenyl moiety (30) resulted in less potent compounds.
Next to the evaluation of a second substitution at the quino-
line ring the position of the carboxamide was also thoroughly
explored. On the basis of the original HTS hit 1 different aro-
matic (hetero)cycles at position 2 were tested (Table 3). An un-
substituted 2-thiazole (31) could be introduced leading to
a 4 nm inhibitor.
With the additional bulk of two methyl groups (32) the po-
tency was diminished significantly whereas a 1,2,4-triazole 33
with only one methyl substituent showed an overall good po-
tency and selectivity profile. Adding pyridyl (34) or substituted
phenyl (35) to position 2 only gave less potent compounds.
Explorations at the carboxamide nitrogen (36–38) showed that
increase of steric demand in that position deteriorated potency
significantly so that overall the unsubstituted carboxamide at
position 2 turned out to be the best option.
Compd
R1
IC50 [mm][a]
GLUT1[b] GLUT2[c] GLUT3[d] GLUT4[e]
4
0.47
n.d.[f]
20.5
59.5
n.d.
3
5
0.006
0.043
18.4
0.22
2.02
7.11
5.63
6.09
6
0.087
2.39
0.34
7
8
0.004
0.002
12.8
10.4
n.d.
36.1
26.4
2.74
1.72
2.02
9
0.003
n.d.
Next, we investigated the substitution pattern at the central
pyrazole (Table 4). The corresponding dimethyl analogue 39 of
19 showed an improvement in GLUT1 potency and an excel-
lent GLUT4 selectivity (selectivity factor 721). However, the
second pyrazole methyl group exposed an additional metabol-
ic soft spot as depicted in Figure 2. Regarding mono-methyla-
tion, a methyl group at pyrazole position 5 (41) was leading
more efficiently to good GLUT3 selectivity than the corre-
sponding regioisomer 40. Leaving the CF3 group as sole sub-
stituent at position 3 (42) yielded a double-digit nanomolar in-
hibitor, whereas the combination of an isopropyl group at po-
sition 3 with a methyl group at position 5 (43) resulted in
a very potent inhibitor with an excellent selectivity profile. The
drawback of such an excellent selectivity is the well-known
metabolic liability of the isopropyl group.
10
11
0.008
0.059
n.d.
2.86
0.87
n.d.
8.50
15.6
[a] Cell-based assay with cells constitutively expressing luciferase; 1 mm
roteneone and 300 mm glucose were co-incubated with inhibitor 15 min
before CellTiter-Gloꢂ readout. Results were normalized to the control cy-
tochalasin B (IC50 GLUT1: 0.1 mm, GLUT2: 2.8 mm, GLUT3: 0.12 mm, GLUT4:
0.28 mm); assay variance: 9%, IC50 calculation R2 >0.9. [b] DLD1 cells.
[c] CHO-hGLUT2 cells. [d] DLD1GLUT1ꢀ/ꢀ cells from Horizon discovery.
[e] CHO-hGLUT4 cells. [f] n.d.: not determined.
against GLUT3 significantly dropped for compound 13, making
this a parameter that had to be regained in later SAR explora-
tions.
Keeping the CF3 group at position 3 and the methyl group
at position 5 of the pyrazole constant, we investigated varia-
tions of the benzyl residue in more detail (Table 5). A cyano
group in ortho position (44) led to a 3 nm inhibitor with excel-
lent selectivity toward GLUT3 but modest toward GLUT4. With
a methyl group (45) a GLUT4 selectivity increase could be ob-
served, whereas with an ortho OCF3 (46) the GLUT1 potency
eroded. A functional group in meta position resulted in double
digit-nanomolar inhibitors for the cyano (47) and methyl (48)
while meta OCF3 (49) only led to a 540 nm inhibitor.
Next, the 2-carboxamide was kept constant and we tested
the influence of a second quinoline substituent. At position 5,
fluorine (14) is superior to a methyl group (15) with regard to
both, potency and selectivity profile. Comparing fluorine (16)
and methyl (17) at position 6 revealed an advantage for the
methyl compound with regard to the selectivity profile. A me-
thoxy substituent at the same position (18) pushed GLUT1 po-
tency below the nanomolar range, however, at the expense of
also hitting the counter-targets GLUT3 and GLUT4 (Table 2).
In position 7, a fluorine substituent (19) is clearly better tol-
erated regarding GLUT3 (selectivity factor 770) and GLUT4 se-
Compared with ortho and meta substituents the para posi-
tion turned out to be a bit more flexible regarding steric
demand. Not only a cyano (19) and CF3 group (50) were well
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
3
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 2. SAR explorations at quinolone ring A.
Compd
A
R1
IC50 [mm]
GLUT1
GLUT2
GLUT3
GLUT4
13
14
15
ꢀH
ꢀF
0.006
0.003
0.018
69.0
n.d.
n.d.
0.58
0.23
0.086
0.17
0.094
0.017
ꢀMe
16
17
18
ꢀF
0.005
0.007
0.0003
n.d.
36.1
n.d.
0.29
3.82
0.11
0.044
0.12
0.013
ꢀMe
ꢀOMe
19 (BAY-876)
ꢀF
0.002
0.008
0.024
10.8
n.d.
9.28
1.67
1.67
2.48
0.29
0.095
1.67
20
21
ꢀMe
ꢀOMe
22
23
ꢀF
0.005
0.003
2.15
n.d.
0.61
0.76
0.19
0.029
ꢀCl
24
25
26
27
28
29
30
–
–
–
–
–
–
–
0.014
0.010
0.045
0.0005
0.002
0.035
0.20
51.2
51.2
51.2
22.2
n.d.
n.d.
n.d.
1.19
23.8
0.30
0.47
6.07
49.9
5.65
0.38
0.061
1.00
0.016
0.18
0.28
2.48
tolerated but also the OCF3 (51) and ethyl (52) groups resulted
in excellent GLUT1 inhibitors with very good selectivity pro-
files. Only the sterically more demanding tert-butyl group (53)
revealed to be sterically too demanding to give a potent com-
pound.
Keeping the para-cyano group constant, the additional in-
troduction of a pyridine nitrogen was viable. Here, especially
the position ortho to the core connection (54) yielded a very
potent and highly selective compound. This substitution pat-
tern revealed to be more attractive than the corresponding re-
&
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
4
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Figure 2. Advancing SAR.
gioisomeric pyridine (55), only resulting in a double-digit nano-
molar inhibitor. In the series of aromatic systems with two ring
nitrogen atoms the pyrazine pattern (57) gave a more potent
derivative than the pyridazine (56) and pyrimidine (58). How-
ever, the fee for this increase in potency was a nearly
complete deprivation of selectivity of compound 57
toward GLUT4. Also five-membered ring heteroaro-
matic systems were investigated at the benzylic posi-
tion. Keeping the cyano group, thiophene 59, thia-
zole 60 and isoxazole 61 led to double-digit nano-
molar inhibitors, partially with low selectivity toward
GLUT4. As could be expected from bioisosteric ap-
proaches employed in literature,[34] the thiophene
represented the best mimic for the benzene ring re-
garding GLUT1 potency. Comparing the potency and
Table 3. SAR explorations of the substituent at position 2 of quinolone.
Compd
R1
R2
IC50 [mm]
GLUT1
0.004
GLUT2
GLUT3
GLUT4
selectivity profiles of all 61 compounds and taking
metabolic stability into account, we identified the
31
32
ꢀH
ꢀH
n.d.
n.d.
2.03
0.09
0.91
compounds 19 (BAY-876) and 54 as very promising
candidates for further characterizations.
The straightforward synthesis of such selective
GLUT1 inhibitors is exemplified for BAY-876 (19) in
Scheme 1. Starting from the commercially available
0.76
1.55
28.4
33
34
35
ꢀH
ꢀH
0.007
51.2
n.d.
0.54
2.05
5-methyl-4-nitro-3-(trifluoromethyl)-1H-pyrazole 62
[35]
benzylation with Cs2CO3
yielded regioselectively
nitropyrazole 64 that was subsequently reduced
using zinc under acidic conditions.[36] With this ami-
nopyrazole in hand the quinoline part of BAY-876
(19) was derived from inexpensive 6-fluoroisatine 66
that was converted into the 7-fluoroquinoline-2,4-di-
carboxylic acid 67 with pyruvic acid under basic Pfit-
zinger conditions.[37] After transforming 67 into the
diester 68, reaction with ammonia in methanol gave
0.074
1.99
ꢀH
ꢀF
ꢀF
0.92
n.d.
10.8
51.2
7.96
1.67
28.4
2.77
0.29
3.12
19
(BAY-876)
0.002
0.082
the 2-quinolinecarboxamide 69 regioselectively.[38]
A
36
37
38
final amide bond formation between 65 and 69
using HATU[39] as coupling reagent yielded the de-
sired compound BAY-876 (19).
Regarding further characterization we first exam-
ined BAY-876 (19) and other candidates in metabolic
stability assays using liver microsomes and hepato-
cytes of different species. Furthermore, the most
promising compounds were checked in the Caco-2
ꢀF
ꢀF
0.34
0.34
n.d.
n.d.
10.1
28.4
1.78
1.13
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
5
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 4. SAR explorations at pyrazole ring B.
Compd
B
IC50 [mm]
GLUT1
0.0009
GLUT2
GLUT3
GLUT4
0.68
39
n.d.
n.d.
n.d.
0.55
40
41
42
0.003
0.007
0.035
0.46
0.40
0.42
0.019
0.002
n.d.
7.96
1.67
0.41
0.29
19
(BAY-876)
10.8
Scheme 1. Reagents and conditions: a) Cs2CO3, MeCN, 608C, 92%; b) Zn,
HOAc, EtOH, H2O, 608C, 92%; c) 33% KOH(aq), pyruvic acid, 408C, 85%;
d) 1. SOCl2, reflux, 2. MeOH, reflux, 46%; e) 7n NH3 in MeOH, 508C, 81%;
f) NaOH(aq), MeOH, RT, 84%; g) 65, HATU, Et2NiPr, DMSO, RT, 46%.
43
0.007
51.2
22.2
5.05
propylpyrazole 43 showed double the clearance in human mi-
crosomes. This metabolic liability was not only observed for 43
but also for other isopropylpyrazole compounds within the
project. Due to this finding no isopropylpyrazole was selected
for advanced in vivo studies.
permeability assay. Starting from compounds 2, 3, and 12 with
moderate to high metabolic clearance in rat hepatocytes (Fig-
ures 1 and 2), installation of a 2-carboxamide at the quinoline
and a CF3 at the pyrazole, like in compound 13, already result-
ed in compounds with low metabolic clearance in human, dog
and mouse liver microsomes as well as dog hepatocytes. How-
ever, rat hepatocyte clearance was higher giving only average
maximal bioavailability according to the well-stirred model.
Comparing the Caco-2 data of 2, 3, 12, and 13, the permeabili-
ty from low (3.5 nmsꢀ1) to high (200 nmsꢀ1) (apical to basolat-
eral) showed the high variation possible within the N-(1H-pyra-
zol-4-yl)quinoline compound class.
Despite the sub-nanomolar potency of compound 52 its
metabolic liability discouraged further in vivo characterization.
Cyanopyridine 54 did not only show promising potency and
selectivity in the GLUT assays but also very good metabolic sta-
bility in liver microsomes and hepatocytes across all tested
species. However, the strong efflux ratio in the Caco-2 assay
appeared to be a major drawback of this compound relative to
BAY-876 (19). The pyrazine 57 and the pyrimidine 58 showed
a two- to threefold higher clearance in rat hepatocytes than
BAY-876 (19).
Taking into account GLUT inhibition, metabolic stability, and
Caco-2 performance we selected BAY-876 (19) as candidate for
in vivo pharmacokinetic studies that were conducted in two
different species (Table 7). In good agreement with the in vitro
hepatocyte data BAY-876 displayed low clearance also in vivo
in rat and in dog. The volume of distribution in steady state
(Vss) was moderate in both species. Terminal half life was inter-
mediate in rat and long in dog due to the very low clearance.
BAY-876 (19) showed low metabolic in vitro clearance in all
tested species except for monkey hepatocytes displaying mod-
erate clearance (Table 6). The Caco-2 permeability was high
and the efflux ratio not considered critical. Compared with the
moderate stability of triazole 33 in mouse liver microsomes
and in rat hepatocytes, dimethylpyrazole 39 revealed good
bioavailability of 88% in mouse liver microsomes and 75% in
rat hepatocytes. Its stability in human microsomes was slightly
lower with 66%. In comparison with 39 the corresponding iso-
&
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
6
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
ene to be the optimal spacer between the phenyl
and pyrazole ring. After successful replacement of
one of the pyrazole methyl groups for the metabol-
ically more stable CF3 group (12), further SAR explo-
ration of the quinoline core demonstrated an unsub-
stituted amide at position 2 and a fluorine at position
7 (19, BAY-876) to be very beneficial regarding GLUT1
potency and selectivity against the other GLUTs.
At the pyrazole core a double substitution at posi-
tions 3 and 5 was found to be crucial for the excel-
lent potency/selectivity profile especially against
GLUT3. Regarding a substituent at the benzyl ring,
the para position yielded more promising com-
pounds than the ortho or meta position. With a para-
cyano group and an additional ring nitrogen, com-
pound 54 demonstrated also an interesting GLUT
profile.
Table 5. SAR investigations of the benzylic moiety at the pyrazole of compound 19.
Compd
R1
R2
IC50 [mm]
GLUT1
GLUT2
GLUT3
GLUT4
44
45
46
ꢀCN
ꢀMe
0.003
0.007
0.39
n.d.
n.d.
n.d.
5.95
10.3
4.98
0.088
1.14
2.04
ꢀOCF3
47
48
49
ꢀCN
ꢀMe
0.025
0.011
0.54
n.d.
n.d.
n.d.
3.15
14.7
35.5
0.95
0.61
1.07
ꢀOCF3
The synthesis of the N-(1H-pyrazol-4-yl)quinoline-4-
carboxamides was straightforward as exemplified for
BAY-876 (19). In vitro PK data showed that both BAY-
876 (19) and 54 were very stable in liver microsomes
and hepatocytes, although 54 had a strong efflux
ratio of around 16. Preliminary in vivo PK studies of
BAY-876 (19) demonstrated that a good oral bioavail-
ability and long terminal half-life is attainable making
it an excellent chemical probe to further evaluate the
hypothesis of cancer treatment with a very selective
GLUT1 inhibitor.
19
(BAY-876)
50
ꢀCN
ꢀCF3
ꢀOCF3
ꢀEt
0.002
0.009
0.005
0.0009
1.18
10.8
n.d.
n.d.
20.5
1.67
6.55
3.27
2.85
5.47
0.29
0.91
0.37
0.37
0.50
51
52
53
ꢀtBu
>10
54
55
56
57
–
–
–
–
0.004
0.032
0.078
0.005
20.2
n.d.
n.d.
n.d.
2.92
9.89
5.02
1.11
0.62
2.24
2.24
0.016
Experimental Section
Chemistry
58
59
60
–
–
–
0.026
0.012
0.067
n.d.
n.d.
n.d.
2.89
2.88
4.32
1.33
0.16
0.89
Materials and methods: All commercially available start-
ing materials and solvents were used without further pu-
rification. Microwave irradiation was applied with a Biot-
age Initiator 60. Flash column chromatography was per-
formed on prepacked flash chromatography columns PF-
15-SIHP purchased from Interchim or KP-Sil purchased
from Biotage using a Biotage Isolera separation system.
1H and 13C NMR spectra were recorded at room tempera-
ture on Bruker Avance spectrometers operating at 300
or 400 MHz for 1H NMR and at 75 or 100 MHz for
13C NMR acquisitions. NMR signal multiplicities are re-
ported as they appear, without considering higher-order
effects. Chemical shifts (d) are given in ppm with the re-
61
–
0.089
25.6
9.09
1.35
sidual solvent signal used as reference (CDCl3: s, 7.26 ppm (1H) and
t, 77.1 ppm (13C); [D6]DMSO: quint, 2.50 ppm (1H) and quint,
40.1 ppm (13C)). LC–MS spectra were recorded on a Waters Acquity
UPLC–MS SQD 3001 spectrometer, using an Acquity UPLC BEH C18
1.7 50ꢃ2.1 mm column, with acetonitrile and water+0.1% formic
acid as eluents at 608C, a flow of 0.8 mLminꢀ1, an injection volume
of 2 mL, with DAD scan at 210–400 nm, ELSD. All tested com-
pounds were at least 95% pure as determined by 1H NMR spec-
troscopy.
As would be expected from the low blood clearance oral bio-
availability was high at the given doses and formulations.
Overall, the preliminary data of BAY-876 demonstrate a favora-
ble in vivo PK profile.
Conclusions
Starting from moderately potent and metabolically labile HTS
hit 1 carrying a furanyl moiety at position 2 of the quinoline
and four methyl groups in total, we quickly improved both,
potency and metabolic stability by substituting the furan and
taking out the two quinoline methyl groups. Using the quino-
line core of compound 3 we found an unsubstituted methyl-
4-{[5-Methyl-4-nitro-3-(trifluoro-
methyl)-1H-pyrazol-1-yl]methyl}-
benzonitrile (64): 3-Methyl-4-nitro-5-
(trifluoromethyl)-1H-pyrazole (4.30 g,
22.0 mmol, 62, CAS-RN 27116–80–9)
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
7
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 6. In vitro pharmacokinetics and Caco-2 permeability data.
[e]
Compd
LM[a] stability CLb,LM[b] [Lhꢀ1 kgꢀ1] (Fmax [%][c])
Hep[d] stability CLb,Hep [Lhꢀ1 kgꢀ1] (Fmax [%])
Caco-2 permeability Papp A!B [nmsꢀ1
]
(ER)[f]
2
1.2 (11) (h)[g]
4.0 (25) (m)[h]
3.7 (12) (r)[i]
200 (0.84)
3
7
12
4.5 (16) (m)
3.4 (38) (m)
2.2 (59) (m)
2.9 (31) (r)
n.d.[j]
2.6 (39) (r)
72 (0.35)
n.d.
3.5 (0.73)
13
17
0.21 (84) (h)
0.06 (97) (d)[k]
1.4 (74) (m)
0.11 (95) (d)
1.8 (57) (r)
74 (3.7)
2.3 (57) (m)
2.0 (53) (r)
n.d.
19
(BAY-876)
0.32 (76) (h)
0.76 (70) (mo)[l]
stable (100) (d)
1.1 (74) (r)
0.37 (72) (h)
1.4 (46) (mo)
0.20 (90) (d)
0.46 (89) (r)
78 (2.5)
0.33 (94) (m)
33
39
2.1 (61) (m)
1.9 (54) (r)
1.0 (75) (r)
n.d.
n.d.
0.45 (66) (h)
0.67 (88) (m)
43
52
0.87 (34) (h)
0.99 (25) (h)
n.d.
2.9 (31) (r)
n.d.
n.d.
54
0.19 (86) (h)
0.40 (84) (mo)
0.001 (100) (d)
0.37 (91) (r)
0.03 (98) (h)
0.32 (88) (mo)
0.001 (100)(d)
0.04 (99) (r)
16 (16)
0.36 (93) (m)
57
58
0.53 (60) (h)
0.22 (83) (h)
1.2 (72) (r)
0.89 (79) (r)
n.d.
n.d.
[a] Liver microsomes (LM). [b] Blood clearance (CLb): CLb =QHꢁCL ; QH: liver blood flow (human: 1.32 Lhꢀ1 kgꢀ1, monkey: 2.6 Lhꢀ1 kgꢀ1, dog: 2.1 Lhꢀ1 kgꢀ1
,
int
QHþCLint
rat: 4.2 Lhꢀ1 kgꢀ1, mouse: 5.4 Lhꢀ1 kgꢀ1), CLint: intrinsic clearance, (well-stirred model of hepatic clearance). [c] Maximal bioavailability after per oral adminis-
ꢀ
ꢁ
CLb
tration (Fmax): Fmax
=
1ꢀQH . [d] Hepatocytes (Hep). [e] The apparent permeability (Papp) values are derived from the transport of the compounds (2 mm)
over a 2 h period from the apical (A) to the basolateral (B) compartment and vice versa. [f] The letters ER refer to the efflux ratios and are calculated by di-
viding the Papp B!A values by the Papp A!B values. [g] Human (h). [h] Mouse (m). [i] Rat (r). [j] n.d.: not determined. [k] Dog (d). [l] Monkey (mo).
was evaporated, and the crude product was purified by flash chro-
Table 7. In vivo pharmacokinetics data of BAY-876 (19).
matography to obtain the desired compound 64 (6.30 g, 92%):
1H NMR (300 MHz, [D6]DMSO): d=2.63 (s, 3H), 5.67 (s, 2H), 7.41 (d,
J=8.5 Hz, 2H), 7.86 ppm (d, J=8.5 Hz, 2H); LC–MS (ESI+): tR =
1.21 min, m/z 311.1 [M+H]+.
Parameter
male Wistar rat[a]
female Beagle dog[b]
Dose i.v. [mgkgꢀ1
]
0.3
0.23
0.33
0.79
2.5
0.6
0.33
85
0.1
0.033
0.059
1.0
22
0.2
CLplasma [Lhꢀ1 kgꢀ1
]
CLblood [Lhꢀ1 kgꢀ1
Vss [Lkgꢀ1
terminal t1/2 [h]
]
4-{[4-Amino-5-methyl-3-(trifluoromethyl)-1H-pyrazol-1-yl]me-
]
thyl}benzonitrile (65): To
a
solution of 4-{[5-methyl-4-nitro-
(3.20 g,
3-(trifluoromethyl)-1H-pyrazol-1-yl]methyl}benzonitrile
10.3 mmol, 64) in ethanol (160 mL)
was added water (80 mL), acetic acid
(16 mL), and zinc dust (3.20 g,
49.0 mmol). This reaction mixture
was stirred at 608C for 1.5 h. After
cooling to 258C the suspension was
filtered through Celite, washed with
Dose p.o. [mgkgꢀ1
]
Cmax,norm,p.o. [kgLꢀ1
F [%]
]
0.93
79
[c]
[a] Formulation: PEG400/water/EtOH (60/30/10). [b] Formulation: PEG400/
water/EtOH (50/40/10). [c] Cmax,norm,p.o. =Cmax,p.o. [mgLꢀ1]/dose p.o.
[mgkgꢀ1].
ethyl acetate, and the complete filtrate was evaporated. To the res-
idue was added water (100 mL) and conc. aq. sodium carbonate
(50 mL). This aqueous phase was extracted three times with ethyl
acetate (150 mL). The combined organic layer was washed with
brine, dried over sodium sulfate, filtered and evaporated to obtain
was dissolved in acetonitrile (65 mL), and 4-(bromomethyl)benzoni-
trile (5.19 g, 26.5 mmol, 63, CAS-RN 17201-43-3) and cesium car-
bonate (8.62 g, 26.5 mmol) were added. The suspension was stirred
at 608C for 2 h. Then, the reaction mixture was filtered, the filtrate
&
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
8
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
a crude product that was purified by flash chromatography to
obtain compound 65 (2.66 g, 92%): H NMR (300 MHz, [D6]DMSO):
d=2.05 (s, 3H), 4.06 (s, 2H), 5.38 (s, 2H), 7.22 (d, J=8.5 Hz, 2H),
7.79–7.85 ppm (m, 2H); LC–MS (ESI+): tR =0.96 min, m/z 281.1
[M+H]+.
with water; 10% aq. sulfuric acid was then added until pH 5 was
reached. After stirring for additional 15 min the solid was isolated
by filtration and dried in vacuo to obtain the desired compound
70 (2.38 g, 84%), which was used without further purification:
1H NMR (300 MHz, [D6]DMSO): d=7.76 (ddd, J=9.4, 8.4, 2.8 Hz,
1H), 7.89 (dd, J=9.9, 2.7 Hz, 1H), 7.92 (brs, 1H), 8.35 (brs, 1H),
8.46 (s, 1H), 8.89 ppm (dd, J=9.4, 6.2 Hz, 1H); LC–MS (ESI+): tR =
0.70 min, m/z 235.1 [M+H]+.
1
7-Fluoroquinoline-2,4-dicarboxylic acid (67): To a mixture of 6-
fluoro-1H-indole-2,3-dione (5.0 g, 30.3 mmol, 66, CAS-RN 324-03-8)
in 33% aq. potassium hydroxide solu-
tion (75 mL) was added pyruvic acid
(4.67 g, 53.0 mmol) and this mixture
was heated at 408C for 18 h. After
cooling to room temperature 10%
N4-[1-(4-Cyanobenzyl)-5-methyl-3-(trifluoromethyl)-1H-pyrazol-4-
yl]-7-fluoroquinoline-2,4-dicarboxamide (19, BAY-876): To a solu-
tion of 4-{[4-amino-5-methyl-3-(trifluoromethyl)-1H-pyrazol-1-yl]me-
aq. sulfuric acid was added until pH
reached about 1. The precipitate was
isolated by filtration and dried in va-
cuo to give the desired compound
67 (6.02 g, 85%), which was used without further purification:
1H NMR (300 MHz, [D6]DMSO): d=7.78 (ddd, J=9.4, 8.5, 2.8 Hz,
1H), 7.99 (dd, J=10.0, 2.6 Hz, 1H), 8.42 (s, 1H), 8.89 ppm (dd, J=
9.5, 6.3 Hz, 1H); LC–MS (ESI+): tR =0.56 min, m/z 236.1 [M+H]+.
Dimethyl 7-fluoroquinoline-2,4-dicarboxylate (68): A mixture of
7-fluoroquinoline-2,4-dicarboxylic acid (6.0 g, 25.5 mmol, 67) and
thionyl chloride (28 mL, 383 mmol)
was heated at 808C for 2 days. After
cooling to 258C the resulting suspen-
sion was evaporated to dryness in va-
cuo. This crude product was sus-
pended in methanol (47 mL) and
held at reflux for 3 h. After cooling to
258C the solid was isolated by filtra-
tion to give compound 68 (3.06 g,
thyl}benzonitrile (144 mg, 0.51 mmol, 65) in DMSO (2.3 mL) was
added HATU (195 mg, 0.51 mmol), N,N-diisopropylethylamine
(112 mL, 0.64 mmol) and 2-carbamoyl-7-fluoroquinoline-4-carboxylic
acid (100 mg, 0.43 mmol, 70). The reaction mixture was stirred for
1 h at 258C. This mixture was directly purified by preparative HPLC
to obtain the desired compound 19 (98 mg, 46%): 1H NMR
(300 MHz, [D6]DMSO): d=2.27 (s, 3H), 5.61 (s, 2H), 7.38 (d, J=
8.3 Hz, 2H), 7.74–7.84 (m, 1H), 7.86–7.95 (m, 3H), 7.97 (brs, 1H),
8.24–8.33 (m, 2H), 8.40 (brs, 1H), 10.48 ppm (s, 1H); 13C NMR
(101 MHz, [D6]DMSO): d=9.3 (s, CH3), 53.1 (s, CH2), 111.0 (s, C),
113.3 (d, JCꢀF =20.3 Hz, CH), 114.8 (s, C), 116.4 (s, CH), 118.6 (s, C),
119.7 (d, JCꢀF =25.7 Hz, CH), 121.4 (q, JCꢀF =269.1 Hz, C), 122.4(s, C),
128.1 (d, JCꢀF =10.3 Hz, CH), 128.2 (s, 2CH), 132.9 (s, 2CH), 136.2 (q,
46%), which was used without fur-
ther purification: 1H NMR (300 MHz, [D6]DMSO): d=3.99 (s, 3H),
4.01 (s, 3H), 7.85 (ddd, J=9.2, 8.4, 2.6 Hz, 1H), 8.07 (dd, J=9.8,
2.6 Hz, 1H), 8.45 (s, 1H), 8.80 ppm (dd, J=9.5, 6.1 Hz, 1H); LC–MS
(ESI+): tR =1.07 min, m/z 264.0 [M+H]+.
J
J
CꢀF =35.6 Hz, C), 138.7 (s, C), 141.7 (s, C), 142.6 (s, C), 147.9 (d,
CꢀF =13.0 Hz, C), 151.4 (s, C), 163.0 (d, JCꢀF =250.4 Hz, C), 165.5 (s,
Methyl 2-carbamoyl-7-fluoroquinoline-4-carboxylate (69): To a so-
lution of dimethyl 7-fluoroquinoline-2,4-dicarboxylate (3.05 g,
11.6 mmol, 68) in methanol (42 mL)
C), 166.1 ppm (s, C); LC–MS (ESI+): tR =1.11 min, m/z 497.1 [M+
H]+.
was added a 7m solution of ammo-
nia in methanol (41 mL, 290 mmol)
and stirred for 3.5 h at 508C. After
cooling to 258C, the precipitate was
Biology
isolated by filtration and dried to
give the desired compound 69
(2.33 g, 81%), which was used with-
out further purification: 1H NMR
Materials and methods: Cytochalasin B and buffers were obtained
from Sigma–Aldrich. All other materials were of reagent grade and
were obtained from commercial sources.
Ultra-high-throughput screen (uHTS) with human GLUT1: It is
well known that a combination of small-molecule inhibitors of mi-
tochondrial electron transport chain and glucose catabolism syn-
ergistically suppress ATP production.[40] For uHTS, CHO-K1 cells
were stable transfected with human GLUT1 and a constitutively ex-
pressing luciferase as described previously.[41] Cells were seeded in
1536 microtiter plates with a density of 1000 cells per well and
starved for 24 h in glucose free DMEM in the presence of 1% FCS.
Prior to measurements cells were incubated for 30 min at 378C in
the presence of 10 mm rotenone to fully block oxidative phosphor-
ylation. Test compounds and caged luciferin were loaded simulta-
neously. Before application of 0.5 mm glucose and corresponding
activation of GLUT1, basal ATP was indirectly measured by lucifer-
ase activity in order to identify effects on cellular ATP levels inde-
pendent of glucose; 10 min kinetic luciferase recordings after ap-
(400 MHz, [D6]DMSO): d=4.03 (s,
3H), 7.83 (ddd, J=9.4, 8.4, 2.8 Hz,
1H), 7.94 (dd, J=9.9, 2.8 Hz, 1H), 7.97 (brs, 1H), 8.39 (brs, 1H),
8.52 (s, 1H), 8.83 ppm (dd, J=9.4, 6.1 Hz, 1H); LC–MS (ESI+): tR =
0.95 min, m/z 249.1 [M+H]+.
2-Carbamoyl-7-fluoroquinoline-4-carboxylic acid (70): To a solu-
tion of methyl 2-carbamoyl-7-fluoroquinoline-4-carboxylate (3.00 g,
12.1 mmol, 69) in methanol (56 mL)
and tetrahydrofuran (20 mL) was
added a solution of sodium hydrox-
ide (4.35 g, 109 mmol) in water
(111 mL). This mixture was stirred for
1 h at 258C and then concentrated
in vacuo. The residue was diluted
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
9
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
C. B. Thompson, Science 2009, 324, 1029–1033; g) R. A. Gatenby, R. J.
Gillies, Int. J. Biochem. Cell Biol. 2007, 39, 1358–1366.
plication of 500 mm glucose allowed the investigation of com-
pound induced inhibition of GLUT1.
[2] R. J. Gillies, I. Robey, R. A. Gatenby, J. Nucl. Med. 2008, 49, 24S–42S.
[3] T. Amann, C. Hellerbrand, Expert Opin. Ther. Targets 2009, 13, 1411–
1427.
GLUT isoform specificity testing: For specificity testing between
GLUT1, GLUT2, GLUT3 and GLUT4 we used DLD1 (for GLUT1),
DLD1GLUT1ꢀ/ꢀ (Horizon discovery, for GLUT3), CHO-hGLUT2 and
CHO-hGLUT4 (GLUT2 and 4) cells in combination with an oxidative
phosphorylation inhibitor (rotenone 1 mm). Cell lines were main-
tained in DMEM medium supplemented with 10% FCS and 1%
penicillin-streptomycin solution and 2% Glutamax under standard
conditions. The cells were treated with trypsin and seeded into
384 plates at a density of 4000 cells per well. The cells were then
cultured overnight in glucose free media containing 1% FCS to
reduce intracellular ATP levels. For GLUT1/2/3, after 16 h the cells
were incubated with appropriate glucose concentration or in case
of GLUT2 fructose concentration (0.1m for GLUT1, 0.3m for GLUT3
and 30 mm fructose for GLUT2, respectively) with or without com-
pounds and 1 mm rotenone for 15 min. The CellTiter-Gloꢂ Lumines-
cent Cell Viability Assay from Promega was then used to measure
ATP levels. Assay was normalized to the control cytochalasin B (IC50
GLUT1: 0.1 mm GLUT2: 2.8 mm, GLUT3: 0.12 mm, GLUT4: 0.28 mm),
assay variance: 9%, IC50 calculation R2 >0.9. For GLUT4, after 16 h
the glucose free medium was removed and cells were adapted to
KCl free tyrode buffer for 3 h. Compounds and rotenone were
added and after 20 min cells were incubated with glucose (0.1m
final concentration) for 15 min. The CellTiter-Gloꢂ Luminescent Cell
Viability Assay from Promega was then used to measure ATP
levels.
[4] X. Fu, G. Zhang, R. Liu, J. Wei, D. Zhang-Negrerie, X. Jian, Q. Gao, J.
Chem. Inf. Model. 2016, 56, 517–526.
[5] R. A. Medina, G. I. Owen, Biol. Res. 2002, 35, 9–26.
[6] T. Nishioka, Y. Oda, Y. Seino, T. Yamamoto, N. Inagaki, H. Yano, H. Imura,
R. Shigemoto, H. Kikuchi, Cancer Res. 1992, 52, 3972–3979.
[7] a) R. S. Brown, R. L. Wahl, Cancer 1993, 72, 2979–2985; b) A. Krzeslak, K.
Wojcik-Krowiranda, E. Forma, P. Jozwiak, H. Romanowicz, A. Bienkiewicz,
M. Brys, Pathol. Oncol. Res. 2012, 18, 721–728.
[8] a) H. E. Rashed, S. A. Ahmed, M. Abdelgawad, Life Sci. J. 2015, 12, 162–
169; b) Y.-M. Shen, G. Arbman, B. Olsson, X.-F. Sun, Int. J. Biol. Markers
2011, 26, 166–172.
[9] Y. Nagase, K. Takata, N. Moriyama, Y. Aso, T. Murakami, H. Hirano, J. Urol.
1995, 153, 798–801.
[10] H. Sasaki, M. Shitara, K. Yokota, Y. Hikosaka, S. Moriyama, M. Yano, Y.
Fujii, Mol. Med. Rep. 2012, 5, 599–602.
[11] Y. Cai, J.-j. Zhai, B.-b. Feng, X.-z. Duan, X.-j. He, J. Obstet. Gynaecol. Res.
2014, 40, 1925–1930.
[12] a) K. Reinicke, P. Sotomayor, P. Cisterna, C. Delgado, F. Nualart, A. Godoy,
J. Cell. Biochem. 2012, 113, 553–562; b) P. Effert, A. J. Beniers, Y. Tamimi,
S. Handt, G. Jakse, Anticancer Res. 2004, 24, 3057–3063.
[13] R. C. Osthus, H. Shim, S. Kim, Q. Li, R. Reddy, M. Mukherjee, Y. Xu, D.
Wonsey, L. A. Lee, C. V. Dang, J. Biol. Chem. 2000, 275, 21797–21800.
[14] J. J.-C. Sheu, B. Guan, F.-J. Tsai, E. Y.-T. Hsiao, C.-M. Chen, R. Seruca, T.-L.
Wang, I.-M. Shih, Am. J. Pathol. 2012, 180, 1179–1188.
[15] F. Schwartzenberg-Bar-Yoseph, M. Armoni, E. Karnieli, Cancer Res. 2004,
64, 2627–2633.
Glucose competition: For the glucose competition DLD1 cells
were treated with trypsin and seeded into 384 plates at a density
of 4000 cells per well. The cells were then cultured overnight in
glucose free media containing 1% FCS to reduce intracellular ATP
levels. After 16 h the cells were incubated with different glucose
concentration (0.1; 1 and 10 mm, respectively) together with com-
pound (30 mm to 1 nm) and 1 mm rotenone for 15 min. The CellTit-
er-Gloꢂ Luminescent Cell Viability Assay from Promega was then
used to measure ATP levels.
[16] a) H. Xu, B. Li, W. Yu, H. Wang, X. Zhao, Y. Yao, D. Huang, Nucl. Med.
Commun. 2013, 34, 953–958; b) A. Wincewicz, M. Sulkowska, M. Koda,
S. Sulkowski, Pathol. Oncol. Res. 2007, 13, 15–20.
[17] C. C. Barron, P. J. Bilan, T. Tsakiridis, E. Tsiani, Metab. Clin. Exp. 2016, 65,
124–139.
[18] a) T. A. D. Smith, Nucl. Med. Biol. 2001, 28, 1–4; b) R. S. Brown, J. Y.
Leung, S. J. Fisher, K. A. Frey, S. P. Ethier, R. L. Wahl, J. Nucl. Med. 1996,
37, 1042–1047.
[19] L. Gnudi, G. Viberti, L. Raij, V. Rodriguez, D. Burt, P. Cortes, B. Hartley, S.
Thomas, S. Maestrini, G. Gruden, Hypertension 2003, 42, 19–24.
[20] S. D. Hughes, C. Quaade, J. H. Johnson, S. Ferber, C. B. Newgard, J. Biol.
Chem. 1993, 268, 15205–15212.
[21] S. J. Vannucci, F. Maher, I. A. Simpson, Glia 1997, 21, 2–21.
[22] a) D. Leto, A. R. Saltiel, Nat. Rev. Mol. Cell Biol. 2012, 13, 383–396;
b) N. J. Bryant, R. Govers, D. E. James, Nat. Rev. Mol. Cell Biol. 2002, 3,
267–277.
Acknowledgements
The authors thank the following people for their valuable assis-
tance and input in the experimental procedures: Dr. Marcus Kop-
pitz, Dr. Jens Geisler, Dr. Joachim Kuhnke, Dirk Schneider, Dr. Mꢁl-
anie Hꢁroult, Dr. Charlotte Christine Kopitz, Dr. Carolyn Sperl, Dr.
Heike Petrul, Dr. Maria Quanz, Dr. Luisella Toschi, Dr. Sylvia Grue-
newald, Dr. Andrea Haegebarth, and Dr. Holger Hess-Stumpp.
The authors also thank Dr. Ludwig Zorn for his valuable support
and precise proofreading during the preparation of this manu-
script. BAY-876 is available as a chemical probe from the Struc-
tural Genomics Consortium (SGC, www.thesgc.org).
[23] a) M. Salas, P. Obando, L. Ojeda, P. Ojeda, A. Perez, M. Vargas-Uribe, C. I.
Rivas, J. C. Vera, A. M. Reyes, Am. J. Physiol. 2013, 305, C90–C99; b) K.-H.
Jung, J. H. Lee, C. H. T. Quach, J.-Y. Paik, H. Oh, J. W. Park, E. J. Lee, S.-H.
Moon, K.-H. Lee, J. Nucl. Med. 2013, 54, 2161–2167.
[24] H.-J. Martin, F. Kornmann, G. F. Fuhrmann, Chem.-Biol. Interact. 2003,
146, 225–235.
[25] R. E. Falk, Pat. No. CA1319107 C, 1995.
[26] Y. Liu, Y. Cao, W. Zhang, S. Bergmeier, Y. Qian, H. Akbar, R. Colvin, J.
Ding, L. Tong, S. Wu, J. Hines, X. Chen, Mol. Cancer Ther. 2012, 11,
1672–1682.
[27] C. Granchi, Y. Qian, H. Y. Lee, I. Paterni, C. Pasero, J. Iegre, K. E. Carlson,
T. Tuccinardi, X. Chen, J. A. Katzenellenbogen, P. J. Hergenrother, F. Min-
utolo, ChemMedChem 2015, 10, 1892–1900.
[28] D. Wang, P.-C. Chu, C.-N. Yang, R. Yan, Y.-C. Chuang, S. K. Kulp, C.-S.
Chen, J. Med. Chem. 2012, 55, 3827–3836.
[29] D. A. Chan, P. D. Sutphin, P. Nguyen, S. Turcotte, E. W. Lai, A. Banh, G. E.
Reynolds, J.-T. Chi, J. Wu, D. E. Solow-Cordero, M. Bonnet, J. U. Flanagan,
D. M. Bouley, E. E. Graves, W. A. Denny, M. P. Hay, A. J. Giaccia, Sci. Transl.
Med. 2011, 3, 94ra70.
[30] H. Siebeneicher, M. Bauser, B. Buchmann, I. Heisler, T. Mueller, R. Neu-
haus, H. Rehwinkel, J. Telser, L. Zorn, Bioorg. Med. Chem. Lett. 2016, 26,
1732–1737.
Keywords: medicinal chemistry · quinoline carboxamides ·
GLUT1 inhibitors · structure–activity relationships · Warburg
effect
[1] a) O. Warburg in Ueber den Stoffwechsel der Tumoren, Constable,
London, 1930; b) W. H. Koppenol, P. L. Bounds, C. V. Dang, Nat. Rev.
Cancer 2011, 11, 325–327; c) O. Warburg, Science 1956, 123, 309–314;
d) O. Warburg, Science 1956, 124, 269–270; e) J. W. Kim, C. V. Dang,
Cancer Res. 2006, 66, 8927–8930; f) M. G. Vander Heiden, L. C. Cantley,
&
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
10
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
[31] K. Kapoora, J. S. Finer-Moorea, B. P. Pedersena, L. Cabonia, A. Waight,
R. C. Hillig, P. Bringmann, I. Heisler, T. Mueller, H. Siebeneicher, R. M.
Stroud, Proc. Natl. Acad. Sci. USA 2016, 113, 4711–4716.
[32] Promega Corporation, 2800 Woods Hollow Road, Madison, WI 53711–
5399 (USA).
[33] P. E. Lindahl, K. E. Oberg, Exp. Cell Res. 1961, 23, 228–237.
[34] P. Ciapetti, B. Giethlen in The Practice of Medicinal Chemistry, 4th ed.
(Eds.: C. G. Wermuth, D. Aldous, P. Raboisson, D. Rognan), Elsevier,
London, 2015, pp. 181–241.
[35] M. Martinell Pedemonte, I. Navarro MuÇoz, M. Soler Lꢄpez, D. Morme-
neo Juliꢅn, M. Rosol, A. Llebaria Soldevila, J. Bofarull Aymamꢆ (Crystax
Pharmaceuticals S.L.), Int. PCT Pub. No. WO2009007399 A1, 2009.
[36] J. G. Buchanan, A. Stobie, R. H. Wightman, J. Chem. Soc. Perkin Trans. 1
1981, 2374–2378.
[38] a) R. R. Renshaw, H. L. Friedman, J. Am. Chem. Soc. 1939, 61, 3320–
3322; b) P. V. N. Reddy, K. C. Jensen, A. D. Mesecar, P. E. Fanwick, M.
Cushman, J. Med. Chem. 2012, 55, 367–377.
[39] L. A. Carpino, J. Am. Chem. Soc. 1993, 115, 4397–4398.
[40] a) O. A. Ulanovskaya, J. Janjic, M. Suzuki, S. S. Sabharwal, P. T. Schumack-
er, S. J. Kron, S. A. Kozmin, Nat. Chem. Biol. 2008, 4, 418–424; b) O. A.
Ulanovskaya, J. Cui, S. J. Kron, S. A. Kozmin, Chem. Biol. 2011, 18, 222–
230.
[41] F. F. Craig, A. C. Simmonds, D. Watmore, F. McCapra, M. R. H. White, Bio-
chem. J. 1991, 276, 637–641.
Received: May 31, 2016
Revised: July 20, 2016
[37] J. N. Sangshetti, A. S. Zambare, I. Gonjari, D. B. Shinde, Mini-Rev. Org.
Chem. 2014, 11, 225–250.
Published online on && &&, 0000
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
11
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

FULL PAPERS
H. Siebeneicher, A. Cleve, H. Rehwinkel,
R. Neuhaus, I. Heisler, T. Mꢀller, M. Bauser,
B. Buchmann*
A GLUT of selectivity! Based on an HTS
hit, extensive SAR studies at all regions
of the structure finally led to the very
potent GLUT1 inhibitor BAY-876, having
more than 100-fold selectivity against
other GLUT members. The good in vitro
and in vivo pharmacokinetics properties
of BAY-876 make it an excellent chemi-
cal probe to further evaluate the scope
and limitations of GLUT1 inhibition as
a new therapeutic option.
&& – &&
Identification and Optimization of the
First Highly Selective GLUT1 Inhibitor
BAY-876
&
ChemMedChem 2016, 11, 1 – 12
www.chemmedchem.org
12
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!