Tariquidar-related triazoles as potent, selective and stable inhibitors of ABCG2 (BCRP)

Article abstract of DOI:10.1016/j.ejmech.2020.112133

Tariquidar derivatives have been described as potent and selective ABCG2 inhibitors. However, their susceptibility to hydrolysis limits their applicability. The current study comprises the synthesis and characterization of novel tariquidar-related inhibit

Full text of DOI:10.1016/j.ejmech.2020.112133

Journal Pre-proof
Tariquidar-related triazoles as potent, selective and stable inhibitors of ABCG2
(BCRP)
Frauke Antoni, Manuel Bause, Matthias Scholler, Stefanie Bauer, Simone A. Stark,
Scott M. Jackson, Ioannis Manolaridis, Kaspar P. Locher, Burkhard König, Armin
Buschauer, Günther Bernhardt
PII:
S0223-5234(20)30100-8
DOI:
https://doi.org/10.1016/j.ejmech.2020.112133
EJMECH 112133
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 31 October 2019
Revised Date: 4 February 2020
Accepted Date: 6 February 2020
Please cite this article as: F. Antoni, M. Bause, M. Scholler, S. Bauer, S.A. Stark, S.M. Jackson, I.
Manolaridis, K.P. Locher, B. König, A. Buschauer, Gü. Bernhardt, Tariquidar-related triazoles as potent,
selective and stable inhibitors of ABCG2 (BCRP), European Journal of Medicinal Chemistry (2020), doi:
https://doi.org/10.1016/j.ejmech.2020.112133.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
© 2020 Published by Elsevier Masson SAS.

Tariquidar-Related Triazoles as Potent, Selective
and Stable Inhibitors of ABCG2 (BCRP)
Frauke Antonia,*, Manuel Bauseb, Matthias Schollera, Stefanie Bauera, Simone A. Starkb,
Scott M. Jacksonc, Ioannis Manolaridisc, Kaspar P. Locherc, Burkhard Königb,*, Armin
Buschauera,#, Günther Bernhardta
a Institute of Pharmacy, University of Regensburg, D-93040 Regensburg, Germany
b Institute of Organic Chemistry, University of Regensburg, D-93040 Regensburg, Germany
c Institute of Molecular Biology and Biophysics, ETH Zürich, CH-8093 Zürich, Switzerland
Abstract
Tariquidar derivatives have been described as potent and selective ABCG2 inhibitors.
However, their susceptibility to hydrolysis limits their applicability. The current study
comprises the synthesis and characterization of novel tariquidar-related inhibitors, obtained by
bioisosteric replacement of the labile moieties in our previous tariquidar analog UR-ME22-1
(9). CuAAC (“click” reaction) gave convenient access to a triazole core as a substitute for the
labile amide group and the labile ester moiety was replaced by different acyl groups in a
Sugasawa reaction. A stability assay proved the enhancement of the stability in blood plasma.
Compounds UR-MB108 (57) and UR-MB136 (59) inhibited ABCG2 in a Hoechst 33342
transport assay with an IC50 value of about 80 nM and belong to the most potent ABCG2
inhibitors described so far. Compound 57 was highly selective, whereas its PEGylated analog
59 showed some activity at ABCB1. Both 57 and 59 produced an ABCG2 ATPase-depressing
effect which is in agreement with our precedent cryo-EM study identifying 59 as an ATPase
inhibitor that exerts its effect via locking the inward-facing conformation. Thermostabilization
of ABCG2 by 57 and 59 can be taken as a hint to comparable binding to ABCG2. As
reference substances, compounds 57 and 59 allow additional mechanistic studies on ABCG2
inhibition. Due to their stability in blood plasma, they are also applicable in vivo. The highly
specific inhibitor 57 is suited for PET labeling, helping to further elucidate the
(patho)physiological role of ABCG2, e.g. at the BBB.
Keywords
ABCG2 transporter, BCRP, inhibitors, multidrug resistance, Hoechst33342, ATPase
1

1 Introduction
In 1990, a novel drug resistance-related membrane protein was discovered in a doxorubicin
resistant breast cancer cell line [1] and in 1998 it was identified as the most recent member of
the human ABC transporter superfamily and named breast cancer resistance protein (BCRP)
[2]. The Human Genome Nomenclature Committee termed the transporter ABCG2 [3].
ABCG2 as well as other ABC transporter subtypes such as ABCB1 (P-gp) und ABCC1
(MRP1) are not only associated with the chemoresistance of malignant tumors [4-7], but they
are also expressed at the blood-brain barrier (BBB), preventing the entry of a broad variety of
xenobiotics, including numerous drugs, into the central nervous system [8-10]. Therefore, co-
administration of an ABC inhibitor and a cytostatic drug represents an attractive strategy to
overcome the BBB and multidrug resistance (MDR) of various malignancies. This has been
demonstrated by several proof-of-concept studies, although ABC inhibitors are not in clinical
use yet [11-16]. ABCG2 has come into focus since apart from being one of the three major
subtypes conferring MDR in cancer cells (in addition to ABCB1 and ABCC1), it also appears
to be the most abundant subtype at the human BBB [17, 18].
Figure 1 shows a selection of ABCG2 inhibitors described so far. Among the first substances
found to affect ABCG2 activity are inhibitors of the ABCB1 transporter, primarily the
benzanilides elacridar (1) and tariquidar (2) [19-22]. Fumitremorgin C (FTC) (3), a fungal
toxin from A. fumigatus Fresen. containing a tetrahydro-β-carboline backbone, was the first
known specific inhibitor of the ABCG2 subtype [23]. Its neurotoxic effects, however,
preclude its use in vivo. Its nontoxic analog Ko143 (4) was developed in 2002 and is still
among the most potent known ABCG2 inhibitors [24]. It was recently reported, though, that it
also interferes with both, ABCB1 and ABCC1 transport, but only at higher concentrations
(low micromolar range) [25]. The suitability of Ko143 for in vivo studies is compromised by
the poor metabolic stability of the ester moiety [26, 27]. Lately, further derivatives of FTC
have been produced, e.g. compound 5 [28]. Besides the fungal toxin FTC, other natural
compounds and derivatives thereof were reported to antagonize ABCG2, the main structural
class being flavonoids such as flavones, chalcones and curcuminoids [29-34]. Chromone is a
partial structure present in many active flavonoids and was derivatized to create novel, potent
and nontoxic ABCG2 inhibitors such as compounds 6 and 7 [35]. The stability of these
compounds still remains to be determined, though. In an ongoing search for potent, selective
and stable inhibitors for ABCG2, further structure types have emerged, for instance the
relatively small acrylonitrile YHO-13177 (8) [36, 37].
2

Figure 1. Structures of known ABCG2 inhibitors: elacridar (1), tariquidar (2), fumitremorgin C (3), Ko143 (4), the β-
carboline derivative 5, the chromones 6 and 7 and the acrylonitrile YHO-13177 (8).
Until recently, the mechanism of ABCG2 inhibition could only be speculated about, since
there were no high-resolution structural data for ABCG2. The breakthrough came in 2017
with the high-resolution cryo-EM structure of ABCG2 [38]. One year later, we determined the
structure of ABCG2 in complex with two different inhibitors, including UR-MB136 (59), a
member of the series we present here [39]. The synthetic route to 59, the appertaining series
of novel ABCG2 inhibitors and their biological characterization have not been published yet,
and are subject of the present study. Heretofore, we modified the dual ABCB1/ABCG2
inhibitor tariquidar (2) to obtain the selective ABCG2 inhibitor UR-ME22-1 (9) (Figure 2)
[40]. The main, selectivity-yielding modification was the shift of the hetarycarboxamido
moiety from the ortho- to the meta-position of the central benzamide core. However, this shift
rendered the central benzamide bond prone to hydrolysis in murine plasma, presumably due
to inadequate steric hindrance, facilitating the attack of the amide bond by plasma esterases.
In an attempt to gain stable tariquidar analogs, the labile amide moiety was replaced by an
indole ring, yielding more stable inhibitors with higher efficacy than 9 [41]. However, it was
shown that these compounds were still hydrolysed to a certain extent, namely at the ester
3

bond attached to the central, trisubstituted phenyl ring. Here we present the synthesis and
pharmacological evaluation of novel tariquidar derivatives. As shown in Figure 2, we
introduced a triazole core as a bioisoster to the labile amide group in 9, which can be more
conveniently synthesized than an indole moiety. Furthermore, we synthesized compounds in
which the labile ester moiety was replaced by ketone groups. We aimed at ABCG2 inhibitors
that are potent, effective, selective, and suitable for complexing with ABCG2 in cryo-EM
experiments and, moreover, also stable in blood plasma, a prerequisite for in vivo studies.
Figure 2. Structure of UR-ME22-1 (9) and general structures of the title compounds developed from 9.
In order to rank our novel series we also purchased or prepared compounds 2-8 as reference
substances and investigated them in our assays.
2 Results and Discussion
2.1 Synthesis
“Key” step in the convergent synthesis of the title compounds was the production of the
triazole core in a copper-catalyzed azide-alkyne cycloaddition (CuAAC), a stepwise variant of
the Huisgen cycloaddition process, and the most often used “click” reaction [42]. Here we
used the polytriazole TBTA as a copper(I)-stabilizing ligand [43]. The azides for the CuAAC
were prepared in three steps (Scheme 1). In order to vary the linker length between the
tetrahydroisoquinoline moiety and the phenyl ring, either 1-(bromomethyl)-4-nitrobenzene
(10) or 1-(2-bromoethyl)-4-nitrobenzene (11) was combined with the respective
tetrahydroisoquinoline derivative (12-14) in an N-alkylation reaction affording compounds
15-19. In our previous studies on tariquidar analogs it proved to be advantageous in terms of
efficacy to introduce a polyethylene glycol (PEG) chain at the tetrahydroisoquinoline moiety,
which was explained by a solubilizing effect [41, 44, 45]. For this reason, we also synthesized
the PEGylated tetrahydroisoquinoline derivative 14 according to our previous report [44] and
alkylated the nitrogen. In the next two steps, the nitro group in compounds 15-19 was reduced
4

to an amine group, resulting in compounds 20-24, and then converted into an azide moiety,
affording compounds 25-29.
Scheme 1. Synthesis of the azides for the “click” reactions. Reagents and conditions: (a) K2CO3, MeCN, reflux, overnight;
(b) Pd/C, H2 (40 bar), EtOH, rt, overnight or SnCl2, EtOH, reflux, 2 h; (c) (I) NaNO2, 6 M HCl aq, - 5 °C, 1 h; (II) NaN3, -
5 °C, 1 h.
The synthesis of the target compounds containing an ester group is shown in Scheme 2.
Firstly, the alkyne (32) as the second reactant in the “click” reactions was prepared by
Sonogashira coupling of methyl 2-amino-4-bromobenzoate (30) and trimethylsilylacetylene to
31 and subsequent TMS-deprotection to compound 32. Then 32 was combined with the
azides 25-29, affording the triazoles 33-37 and the quinoline moiety was introduced by amide
coupling with quinoline-2-carboxylic acid, yielding the inhibitors 38-42.
5

Scheme 2. Synthesis of the inhibitors 38-42 bearing an ester moiety. Reagents and conditions: (a) ethynyltrimethylsilane,
Pd(PPh3)2Cl2, CuI, NEt3, THF, 60 °C, overnight; (b) TBAF, THF, 0 °C, 2 h;(c) CuSO4, sodium ascorbate, TBTA, DMF, rt,
24 h; (d) quinoline-2-carboxylic acid, HBTU, DIPEA, DCM, 0 °C → rt, 24 h.
In the second subseries, the labile ester moiety was replaced by different ketone groups
(Scheme 3). 3-Bromoaniline (43) was acylated in a Sugasawa reaction, a specific ortho-
acylation of anilines, to form the ketones 44-46. Compound 44 was first treated with
quinoline-2-carboxylic acid to form the amide 47 and then submitted to a Sonogashira
coupling with trimethylsilylacetylene (resulting in 48) and deprotected to 49. Precursors 45
and 46 were first coupled with trimethylsilylacetylene under Sonogashira conditions (forming
50 and 51), deprotected (to 52 and 53), and then treated with quinoline-2-carboxylic acid to
form the amide bond (yielding 54 and 55). The final step was the “click” reaction of the
alkynes 49, 54, and 55 with the azides 28 or 29 to form the inhibitors 56-59.
6

Scheme 3. Synthesis of the inhibitors 56-59 bearing a ketone moiety. Reagents and conditions: (a) (I) BCl3,DCM, 0 °C,
20 min; (II) AlCl3, respective nitrile, reflux, 24 h; (III) 2 M HCl aq, 0 °C, 10 min → 80 °C, 1 h; (b) quinoline-2-carboxylic
acid, HBTU, DIPEA, DCM, 0 °C → rt, 24 h; (c) ethynyltrimethylsilane, Pd(PPh3)2Cl2, CuI, NEt3, THF, 60 °C, overnight; (d)
TBAF, THF, 0 °C, 2 h; (e) CuSO4, sodium ascorbate, TBTA, DMF or THF, rt, 24 h.
The reference compounds 3 and 4 were commercially available; 2 [21], 5 [28], 6 [35], 7 [35],
and 8 [36, 37] were synthesized according to literature.
2.2 Inhibition of the ABCG2, ABCB1 and ABCC1 Transporter Function
The newly synthesized compounds (38-42 and 56-59), as well as our previous tariquidar
analog 9 and the reference compounds 2-8, were analyzed for inhibition of the ABCG2
transport activity in the Hoechst 33342 microplate assay. The DNA stain Hoechst 33342,
which emits blue fluorescence when bound to DNA, is a substrate of the ABCG2 transporter
and therefore extruded from ABCG2-expressing cells such as the MCF-7/Topo cells utilized
in this assay. When the ABCG2 transport function is inhibited, Hoechst 33342 accumulates in
the cells and can be detected fluorometrically, which allows for determination of the
inhibitory potency and efficacy of the target compounds. Examples of concentration-response
curves are shown in Figure 3. In order to assess the selectivity of the inhibitors, they were also
analyzed in the calcein-AM microplate assays with respect to inhibition of the ABCB1 and
7

ABCC1 transporter activity, using KB-V1 and MDCK.2-MRP1 cells, respectively. The assay
principle is analogous to the Hoechst 33342 assay. The dual ABCB1 and ABCC1 substrate
calcein-AM is added to the cells and exported by the respective transporter. Upon inhibition
of the transport, calcein-AM accumulates in the cells, becomes cleaved to calcein by
intracellular esterases and complexes Ca2+-ions, resulting in a strong green fluorescence. The
fluorescence can be detected and correlated with the inhibitory potencies of the test
compounds. The results of the transport assays are summarized in Table 1.
Figure 3. Concentration-dependent inhibition of ABCG2-mediated Hoechst efflux in MCF-7/Topo cells by (A) the reference
compounds FTC (3), Ko143 (4), 7 and 8 as well as (B) tariquidar (2) and its novel analogs 42, 57 and 59. The inhibition is
expressed relative to the maximal effect in the presence of 10 µM FTC set to 100%.
FTC (3), its well-known analog Ko143 (4) and the more recent analog 5, which all comprise a
tetrahydro-
102%, 116% and 111%, respectively. However 4, with an IC50 value of 92 nM, was over ten
times more potent than the other two tetrahydro- -carboline derivatives in our assay. Whereas
β-carboline moiety, were similarly effective at ABCG2 with an Imax value of
β
4 is reported to be more potent than its precursor 3 [24], the IC50 value of 5 determined here
(1.1 µM) was about five times higher than reported in literature (lit. 233 nM in Hoechst 33342
assay [28]; 237 nM in pheophorbide A assay [28]). This discrepancy probably reflects
differences in the assay set-up, for instance different concentrations of Hoechst 33342 were
applied (it was higher in our assay). In accordance with previous reports [25], 4 was not
entirely selective; in contrast to 3 and 5 it showed activity at the ABCB1 and ABCC1
transporter. The reference compounds 6-8 were almost equieffective with 4, but less potent.
The chromone 7 inhibited ABCG2 with an IC50 value of 707 nM (lit. 110 nM [35]), being
almost three times more potent than the chromone 6, which showed an IC50 value of 1.8 µM
(lit. 170 nM [35]). Here again, the IC50 values determined by us were higher than those
reported in literature. Furthermore, chromone 7, as opposed to chromone 6, showed very low
inhibition of ABCB1 and ABCC1. The acrylonitrile 8 inhibited ABCG2 with a far higher IC50
value (3.3 µM) than 4. This compound is interesting insofar as it appears to be a triple
8

ABCG2, ABCB1 and ABCC1 inhibitor, which might have a different application than
specific inhibitors.
In accordance with previous publications [22], tariquidar (2) was a dual ABCG2 and ABCC1
inhibitor, but showed much higher inhibitory potency and also higher efficacy at the ABCB1
than at the ABCG2 transporter. Structural modifications at the benzamide core of 2 led to a
drastic increase in selectivity for ABCG2 over ABCB1 (compound 9), confirming our
previous results [40]. Compound 9 showed only very low inhibition of ABCB1, but inhibited
ABCG2 with an IC50 value of 101 nM. The maximal response of 99% was higher than that
determined in the flow cytometric mitoxantrone efflux assay we used before. The replacement
of the labile amide moiety in 9 by a triazole ring resulted in the novel inhibitors 38-42. This
replacement was well-tolerated, the efficacy of the new compounds being 85-95% and the
potencies in the three-digit nanomolar range. Apparently the length of the linker between the
tetrahydroisoquinoline moiety and the phenyl ring does not clearly correlate with the
inhibitory activity. Compounds 38 and 41, as well as 39 and 42, only differ in the length of
the linker (one vs. two methylene groups). Whereas compounds 39 and 42 were equipotent
(IC50 values of about 150 nM), compound 38 (IC50 of 350 nM) was almost twice as potent as
41. The introduction of a PEG chain increased the potency by a factor of two (38 vs. 39) or
four (41 vs. 42). The replacement of the two methoxy groups at the tetrahydroisoquinoline
moiety by hydrogen atoms (40) increased the potency slightly. Taken together, variations in
the linker and the moieties attached to the tetrahydroisoquinoline ring seem to be well-
tolerated. In terms of selectivity, compounds 38-42 showed great differences. While all of
them were inactive at ABCC1, they showed very dissimilar effects on ABCB1. Here again,
there was no correlation between the length of the linker and ABCB1 inhibition. Both 38 and
39, with an identical linker (one methylene group), inhibited ABCB1 with a maximal
response between 20% and 30% (up to 100 µM). 40, 41 and 42 (two methylene groups)
diverged very much depending on the substituents at the tetrahydroisoquinoline moiety. 41
was inactive, 40 was 11% effective (up to 100 µM) and 42, containing the PEG chain, was
80% effective and exhibited an IC50 value in the low micromolar range, making it a dual
ABCG2 and ABCB1 inhibitor. It seems as if the PEG chain can confer activity at ABCB1.
Since, however, this was not the case for 39 as compared to 38, this effect appears to depend
on the specific structure.
9

Table 1. Inhibitory effect of the reference compounds 2-8, our previous tariquidar analog 9 and the new inhibitors 38-42 and
56-59 on the transport activity of ABCG2, ABCB1 and ABCC1.
ABCG2a
Imax [%]d,e
ABCB1b
Imax [%]d,f
ABCC1c
Imax [%]d,g
Compound
2 (tariquidar)
3 (FTC)
IC50 [nM]d
6223 ± 707
1483 ± 194
92 ± 9
IC50 [nM]d
356 ± 3
IC50 [nM]d
inactiveh,i
inactive
6650 ± 2856
inactive
inactivei
3599 ± 302
≥ 39750
inactive
inactive
inactive
inactive
inactive
inactive
inactive
inactive
≥ 19040
inactive
72 ± 3
100 ± 3
–
–
102 ± 3
116 ± 2
111 ± 3
108 ± 5i
109 ± 12
106 ± 9
99 ± 1
inactive
≥ 25520j
inactive
inactivei
6756 ± 946
≥ 65480
≥ 12080
≥ 10730
≥ 25030
≥ 38270
inactive
1623 ± 405
inactive
inactive
≥ 21830
≥ 44460
–
4 (Ko143)
≥ 11
–
29 ± 8
1115 ± 32
1800 ± 300i
707 ± 98
3345 ± 342
101 ± 17
350 ± 53
153 ± 12
486 ± 5
–
5
–
–
6
10 ± 1
≥ 53
≥ 14
≥ 25
≥ 21
≥ 11
–
15 ± 1
7
8 (YHO-13177)
9 (UR-ME22-1)
38 (UR-MB86)
39 (UR-MB84)
40 (UR-MB81)
41 (UR-MB19)
42 (UR-MB95)
56 (UR-St1)
57 (UR-MB108)
58 (UR-St2)
59 (UR-MB136)
≥ 81
–
87 ± 2
–
85 ± 3
–
95 ± 5
–
617 ± 179
144 ± 35
140 ± 40
79 ± 5
90 ± 3
–
86 ± 3
80 ± 7
–
–
97 ± 2
–
91 ± 2
–
–
295 ± 28
81 ± 16
90 ± 0.4
90 ± 3
≥ 24
≥ 42
≥ 14
–
a Hoechst 33342 microplate assay using ABCG2-expressing MCF-7/Topo cells.
b Calcein-AM microplate assay using ABCB1-expressing KB-V1 cells.
c Calcein-AM microplate assay using ABCC1-expressing MDCK.2-MRP1 cells.
d Mean values ± SEM from two to seven independent experiments, each performed in triplicate.
e Maximal inhibitory effect (Imax) relative to the response to FTC at a concentration of 10 µM (100%).
f Maximal inhibitory effect (Imax) relative to the response to tariquidar at a concentration of 10 µM (100%).
g Maximal inhibitory effect (Imax) relative to the response to reversan at a concentration of 30 µM (100%).
h Inactive: response ≤ 10% up to a concentration of 100 µM.
i Data taken from our previous studies [46].
j No plateau was reached up to a concentration of 100 µM. In these cases, minimum IC50 values and Imax values were stated.
The second modification of 9 was the replacement of the labile ester moiety at the central,
trisubstituted phenyl core by ketones, yielding the inhibitors 56-59. This alteration increased
the efficacy at ABCG2 to 90-97%. The IC50 values were quite similar in the two to three digit
nanomolar range. The length of the acyl group attached to the benzamide core was varied.
The propionyl group (in 57) proved to be superior to the smaller acetyl group (in 56) and also
to the bulkier isobutyryl moiety (in 58) in terms of inhibitory potency and selectivity, 58
10

showing moderate ABCB1 and ABCC1 activity. In this subseries, the introduction of a PEG
chain (59) left the inhibitory potency at ABCG2 unchanged, but, here again, caused an
inhibitory effect on ABCB1, with the maximal response being 42% (up to a concentration of
100 µM). It is conceivable that the PEG chain confers affinity to the binding site in ABCB1.
Except for 58, none of the ketone-containing inhibitors showed inhibition of ABCC1.
2.3 Effect on the ABCG2 ATPase Activity
On the basis of the results obtained from the functional transport assays, the two most potent
novel inhibitors (57 and 59), their precursor 9 and the reference compounds 2-8 were
examined with regard to their effect on ABCG2 ATPase activity. The ATPase assay,
requiring relatively high levels of transporter protein, was performed with ABCG2-expressing
Sf9 membranes. ABC transporters gain the energy for the transport of their substrates from
ATP hydrolysis to ADP and inorganic phosphate. The latter was determined in a colorimetric
reaction after incubation of the membrane preparations with different concentrations of test
compounds, resulting in concentration-response curves (examples depicted in Figure 4A+B).
In a variant of this assay, the ABCG2 transporters in the membranes were additionally
stimulated with the ABCG2 substrate sulfasalazine to investigate whether the test compounds
reverse the activating effect caused by sulfasalazine, or not (Figure 4C+D). The results of the
two assay variants are summarized in Table 2.
11

Figure 4. (A+B) Concentration-dependent stimulation or suppression of the ATPase activity in ABCG2-expressing Sf9
membranes by (A) the reference compounds FTC (3), Ko143 (4), 6 and 8 as well as (B) tariquidar (2), its analog UR-ME22-1
(9) and its novel analogs 57 and 59. The effect is expressed relative to the basal ATPase activity (0%) and the maximal
stimulatory effect in the presence of 30 µM sulfasalazine set to 100%. (C+D) Concentration-dependent inhibition of the
sulfasalazine (3 µM) -stimulated ATPase activity in ABCG2-expressing Sf9 membranes by (C) the reference compounds
FTC (3), Ko143 (4) and 7 as well as (D) tariquidar (2), its analog UR-ME22-1 (9) and its novel analogs 57 and 59. The
inhibition is expressed relative to the maximal inhibitory effect in the presence of 10 µM FTC set to 100%.
In the absence of inhibitors, ABCG2 showed relatively high basal ATPase activity, which was
set to zero in our assay. The test compounds, all being transport inhibitors, behaved in three
different manners with respect to ABCG2 ATPase activity. Firstly, most of them decreased it
to a level below the basal activity (below zero) or secondly had no effect on it. In these two
cases it can be assumed that the test compounds are not transported and the inhibition of the
transport of other substrates may be, at least in part, attributed to the inhibition of the ATPase
activity of the transporter. A reduction of the ATPase activity was evoked by FTC (3) and its
analogs Ko143 (4) and 5, and the tariquidar analogs 9, 57 and 59, their maximal effects being
between - 20% and - 37% and their inhibitory potencies in the two-digit (4, 57 and 59) to
three-digit (3, 5 and 9) nanomolar range. The compounds were also able to inhibit
sulfasalazine-stimulated ATPase activity with an efficacy around 100% (3, 9, 57 and 59) or
more (4, 5) and potencies again in the two-digit (57 and 59) to three digit (4, 5 and 9)
nanomolar range. Only 3 shows a higher IC50 value of 1.8 µM. The chromone 7 exhibited no
effect on the basal ATPase activity, but was able to reverse activation by sulfasalazine.
12

Thirdly, some compounds increased the ATPase activity and can therefore be considered
transporter substrates, although the coupling of ATP-hydrolysis to transport has been
challenged [47]. It is conceivable that transport inhibitors that are strong ATPase activators at
the same time compete with other substrates, e.g. Hoechst 33342, for transport by diffusing
rapidly back into the cell (so-called “fast-diffusers”) [48]. To explore if they also compete for
the same binding site, further interaction type analyses would be required. It seems probable,
though, because we recently identified one central multidrug-binding site [39] (i.e. for
substrates and inhibitors). The acrylonitrile 8 activated the ATPase activity as efficiently as
the standard activator sulfasalazine (namely with 99%) and even exhibited a lower EC50 value
(327 nM vs. 916 nM for sulfasalazine). Compound 8 might be such a “fast-diffuser”, an
assumption also supported by its low molecular mass.
Table 2. Effect of the reference compounds 2-8, our previous tariquidar analog 9 and the new inhibitors 57 and 59 on the
ATPase activity of ABCG2 with and without pre-stimulation with sulfasalazine.
Effect on ATPase activitya
Inhibition on stimulated ATPase activityb
IC50 [nM]c Imax [%]c,f
–
Compound
Sulfasalazine
2 (tariquidar)
3 (FTC)
EC50 [nM]c,d
Emax [%]c,e
916 ± 2
211 ± 31
412 ± 240
70 ± 17
134 ± 3
274 ± 86
–
96 ± 4
–
25 ± 3
≥ 1180
1808 ± 203
271 ± 44
226 ± 30
194 ± 128
745 ± 74
–
≥ 79
- 20 ± 2
- 31 ± 1
- 37 ± 2
45 ± 3
104 ± 2
124 ± 5
141 ± 8
19 ± 4
113 ± 4
–
4 (Ko143)
5
6
0 ± 1
7
8 (YHO-13177)
9 (UR-ME22-1)
57 (UR-MB108)
59 (UR-MB136)
327 ± 34
105 ± 14
70 ± 28
52 ± 38
99 ± 7
- 27 ± 7
- 21 ± 4
- 23 ± 0.1
294 ± 17
96 ± 2
105 ± 3
99 ± 3
112 ± 1
93 ± 8
a ATPase microplate assay using ABCG2-expressing Sf9 membranes.
b
ATPase microplate assay using ABCG2-expressing Sf9 membranes stimulated with sulfasalazine at a concentration of
3 µM.
c Mean values ± SEM from two to four independent experiments, each performed in duplicate.
d
Half-maximal effective concentrations were termed EC50 (not IC50) values, even if the ATPase activity was reduced, in
order to avoid confusion with the data from the ATPase assay with sulfasalazine stimulation.
e Maximal effect (Emax) on the ATPase activity relative to the basal ATPase activity (0%) and to the response to sulfasalazine
at a concentration of 30 µM (100%).
f Maximal inhibitory effect (Imax) relative to the response to FTC at a concentration of 10 µM (100%).
13

Transport inhibitors that exert only a partially activating effect on the ATPase activity were
tariquidar (2), showing a maximal effect of 25% and standing in contrast with its ATPase-
suppressing analogs (9, 57 and 59), and the chromone 6 (45%), as opposed to the chromone 7
with no effect. Nonetheless, both 2 and 7 were able to partly antagonize ATPase stimulation
by sulfasalazine.
The ATPase results from our novel tariquidar derivatives are consistent with the structural
data on ABCG2 inhibition we published recently [39]. We showed that compound 59 and a
derivative of 4 both bound to the multidrug-binding site identified in these experiments
(cavity 1; located in the transmembrane domains) and locked the inward-facing conformation,
thereby inhibiting the ATPase activity (of the cytoplasmic, nucleotide-binding domains) due
to conformational coupling.
Figure 5 shows the comportment of the reference compounds FTC (3) and its analogs Ko143
(4) and 5, in relation to our previous tariquidar analog 9 and our most potent novel analogs 57
and 59, all of which suppressed the ABCG2 ATPase activity. Depicted are half-maximal
(effective or inhibitory) concentrations, determined in the three different functional assays
performed, namely the Hoechst 33342 transport assay and the ATPase assay with and without
pre-stimulation with sulfasalazine. Of course, it is expected that half-maximal concentrations
differ when determined in different assays, depending on the experimental conditions.
However, the differences in the experimental setup should become obvious by a constant
factor between the IC50/EC50 values from the different assays, provided that all compounds
share the same mechanism of action. To identify outliers and to show commonality, we
plotted the functional data obtained from different assays in a radar chart. Ko143 (4) and the
tariquidar analogs 9, 57 and 59 showed similar behavior – their potencies in the three assays
coincided fairly well. FTC (3) and its derivative 5 on the other hand, did not show comparable
half-maximal concentrations. Their potencies determined in the Hoechst transport assay were
much higher than those in the ATPase assay and compound 3 also showed a higher half-
maximal concentration in the ATPase assay with sulfasalazine-stimulation than without. Our
previous structural study [39] showed that one 59 molecule or two molecules of a derivative
of 4 completely occupied the volume of cavity 1. We concluded that these inhibitors act
competitively at the binding site with substrates, which was the second inhibition mechanism
identified, apart from ATPase inhibition. So one could speculate that not only 59 but also the
other inhibitors compete for the binding site with substrates, and similar potencies in the three
assays might suggest that the inhibitors 4, 9, 57 and 59 bind to ABCG2 with higher affinities
than the substrates Hoechst 33342 and sulfasalazine. Vice versa, higher potencies in assays in
14

the presence of substrates might be taken as a hint to displaceability of 3 and 5 by the
respective substrates at the concentrations employed.
Figure 5. Comparison of half-maximal effective/ inhibitory concentrations produced by the compounds 3, 4, 5, 7, 9, 57 and
59 in the Hoechst 33342 transport assay (black), the ATPase assay (orange) and the ATPase assay with pre-stimulation with
sulfasalazine (blue).
2.4 Thermostabilization of ABCG2
We analyzed the thermostabilization of ABCG2 by our two most potent novel tariquidar
analogs (57 and 59) and the reference compound Ko143 (4) in a size-exclusion
chromatography-based thermostability assay (SEC-TS). For this purpose, purified ABCG2
was incubated with or without inhibitor at increasing temperatures and subjected to SEC. The
main peak heights were plotted against the temperature to obtain the “melting points” (Tm
values) of ABCG2. The curves for ABCG2 in the presence and absence of compound 57 are
depicted in Figure 6; the curves for ABCG2 in the presence and absence of compounds 4 and
59 were included in our previous report [39] (on the cryo-EM structure of ABCG2 in complex
with inhibitors).
Figure 6. SEC-TS of ABCG2 before or after the addition of 57 at a concentration of 10 µM. The peak heights are expressed
relative to the peak height of ABCG2 without inhibitor at 4 °C (HP-SEC analysis, UV detection at 280 nm).
15

The melting points determined in this assay are listed in Table 3. Compounds 4, 57 and 59
showed approximately 7 °C thermostabilization of purified ABCG2. Since the three
compounds stabilized ABCG2 to the same extent, it can be deduced that these compounds
bind to ABCG2 with similar affinities.
Table 3. Tm values of ABCG2 alone and in complex with 4, 57 or 59 determined in a SEC-TS.
sample
Tm [°C]a
ABCG2
53.3 ± 0.3
60.0 ± 0.3
60.5 ± 0.3
59.0 ± 0.5
4 (Ko143)
57 (UR-MB108)
59 (UR-MB136)
a Best fit values ± SD.
2.5 Cytotoxicity and Reversal of Drug Resistance
The effect of the two most potent novel transport and ATPase inhibitors (57 and 59) and the
most potent reference compound Ko143 (4) on the proliferation as well as the ability to
overcome drug resistance of MCF-7/Topo cells were investigated in a kinetic
chemosensitivity assay, using the inhibitors alone and in combination with the cytostatic
topotecan, an ABCG2 substrate. The results are shown in Figure 7. Incubation of the cells
with 4 had no effect on cell proliferation up to a concentration of 3 µM. A slight cytotoxic
effect was observed at a concentration of 10 µM. When incubated with topotecan alone at a
concentration of 100 nM, the cells were not affected due to their resistance against the
cytostatic. By contrast, the combination of 100 nM topotecan with a per se nontoxic
concentration of 4 led to a complete reversal of resistance. Inhibitor 57 was nontoxic up to a
concentration of 1 µM. At higher concentrations, however, it showed cytotoxicity, which
might be an interesting property in view of addressing tumor stem cells [49]. By analogy with
4, 57 reversed topotecan resistance when administered at nontoxic concentrations in
combination with the cytostatic (100 nM). The inhibitor 59, which only differs from 57 in the
PEG chain attached to the tetrahydroisoquinoline moiety, was also able to reverse drug
resistance in MCF-7/Topo cells and was superior to 57 in so far as it did not exhibit marked
cytotoxicity at concentrations up to 10 µM.
16

Figure 7. (A-C) Antiproliferative activities of Ko143 (4) (A) and the novel tariquidar analogs 57 (B) and 59 (C) against
MCF-7/Topo cells upon long-term incubation. (D-F) Reversal of ABCG2-mediated drug resistance against topotecan on
proliferating MCF-7/Topo cells: effect of 100 nM topotecan (Topo) alone and in combination with different concentrations of
4 (D), 57 (E) or 59 (F). Antiproliferative effects correspond to the left y-axes. The growth curves of untreated control cells
(open circles) correspond to the right y-axes. Data are mean values ± SEM of two to four independent experiments, each
performed in octuplicate.
In conclusion, the reversal of drug resistance in MCF-7/Topo cells by the compounds 4, 57
and 59 is in good agreement with the data from the transport assays (Figure 3, Table 1) and
confirms the three substances as inhibitors of ABCG2-mediated drug efflux.
2.6 Stability in Blood Plasma
Prerequisite for in vivo studies is the stability of the test compounds in blood plasma. We first
replaced the labile benzamide bond in 9 by a triazole core (affording compounds 38-42) and
then introduced ketones at the central trisubstituted phenyl ring as alternatives to the labile
ester moiety, resulting in the inhibitors 56-59. Exemplarily, the stability of 59 upon incubation
in murine blood plasma at 37 °C was demonstrated by HPLC-analysis. As depicted in
Figure 8, the peak area of 59 remains virtually unchanged over a period of 24 hours and no
additional peaks were detected at 220 nm. This experiment demonstrates that the bioisosteric
replacement of the labile amide and ester moieties in 9 provided for stable inhibitors.
17

Figure 8. Chromatograms illustrating the stability of compound 59 upon incubation in murine blood plasma at 37 °C over a
period of 24 h. RP-HPLC analysis, UV detection at 220 nm.
3 Conclusion
The study presented is a continuation of our previous investigations on tariquidar analogs as
ABCG2 inhibitors. Among the title compounds, UR-MB108 (57) and UR-MB136 (59)
excelled in terms of inhibitory potency in the Hoechst 33342 transport assay. With IC50 values
of around 80 nM, they are among the most potent ABCG2 inhibitors known so far. Reversal
of drug resistance in MCF-7/Topo breast cancer cells confirmed their ability to inhibit drug
efflux. The PEGylated compound 59 also showed low activity at ABCB1, whereas 57 was
highly ABCG2-selective. Both compounds suppressed the basal ABCG2 ATPase activity and
were able to antagonize the ATPase-activating effect of sulfasalazine. Furthermore, the two
compounds thermostabilized ABCG2 to the same extent, suggesting comparable binding
affinities. Compound 59 was already analyzed in our precedent structural study [39] of
ABCG2 in complex with inhibitors (but its synthesis and biological characterization has not
been described until this report). We showed that one 59 molecule occupied the cavity for
multidrug binding (in the transmembrane domain), locked the inward-facing conformation
and inhibited the cytosolic ATPase subunit via conformational coupling. Compounds 57 and
59 differ only in the PEG chain attached at the tetrahydroisoquinoline moiety in 59 and it is
probable that they bind similarly to the central multidrug binding site, an assumption which is
supported by their same behavior in the functional assays and the thermostability assay.
Respecting future research, we expect compounds 57 and 59 to be of potential value as
reference substances in mechanistic studies on ABCG2 regarding transport and inhibition.
Compound 57 is beneficial, because it is more conveniently synthesized than 59. Furthermore,
in the light of their stability in blood plasma, compounds 57 and 59 can be employed as
pharmacological tools for in vivo investigations on the (patho-)physiological role of the
18

ABCG2 transporter and as a means to overcome MDR and the BBB. The high specificity of
compound 57 towards ABCG2 makes it convenient, for instance, as PET tracer for imaging
11
the ABCG2 transporter at the BBB, the two methoxy groups in 57 being suited for C-PET
labeling. In addition, it could be used for in vivo studies with pharmacological interventions
specifically on the ABCG2 isoform, for example to improve the efficacy of photodynamic
therapy of non-melanoma skin cancer.
4 Experimental Section
4.1 Chemistry: Experimental Protocols
4.1.1 General Experimental Conditions
Chemicals and solvents were purchased from commercial suppliers (Sigma Aldrich, Munich,
Germany; Merck, Darmstadt, Germany; VWR, Darmstadt, Germany; Thermo Fisher
Scientific, Waltham, MA, USA; TCI, Eschborn, Germany) and used without further
purification unless stated otherwise. The catalyst Pd(PPh3)2Cl2 [50] for the Sonogashira
couplings and the ligand tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine) [43] (TBTA) for
the CuAAC were synthesized according to published procedures. Per analysis grade solvents
were used for reaction mixtures and technical grade solvents for chromatography (automated
flash column chromatography and TLC). Reactions requiring anhydrous conditions were
carried out in dried reaction vessels under an atmosphere of nitrogen and anhydrous solvents
were used. Millipore water was used throughout for the preparation of buffers and HPLC
eluents. Acetonitrile for analytical and preparative HPLC (gradient grade) was obtained from
Merck (Darmstadt, Germany).
Thin layer chromatography was performed on TLC Silica gel 60 F254 aluminium plates
(Merck, Darmstadt, Germany). Visualization was accomplished by UV irradiation at
wavelengths of 254 nm and 366 nm or by staining with potassium permanganate or Ehrlich’s
reagent.
Automated flash column chromatography was performed on an Isolera Spektra One device
(Biotage, Uppsala, Sweden). Silica gel 60 (0.040-0.063 mm) for flash column
chromatography (Merck, Darmstadt, Germany) was used.
Melting points were determined in open capillaries on an OptiMelt MPA 100 apparatus
(Stanford Research Systems, Sunnyvale, CA, USA) and are uncorrected.
1
NMR spectra were recorded on an Avance 300 instrument (7.05 T, H: 300.1 MHz, 13C:
1
13
75.5 MHz), an Avance 400 instrument (9.40 T, H: 400 MHz, C: 101 MHz) or an Avance
19

1
600 instrument with cryogenic probe (14.1 T, H: 600 MHz, 13C: 151 MHz) (Bruker,
Karlsruhe, Germany) with TMS as external standard.
High-resolution mass spectrometry (HRMS) analysis was performed on an Agilent 6540
UHD Accurate-Mass Q-TOF LC/MS system (Agilent Technologies, Santa Clara, CA, USA),
using an ESI source.
IR spectra were recorded with a Golden Gate Single Reflection Diamond ATR System
(Specac, Orpington, UK) in an Excalibur FTS 3000 FT-IR-Spectrometer (Bio-Rad, München,
Germany).
Preparative HPLC was performed on a system from Knauer (Berlin, Germany), consisting
of two K-1800 pumps and a K-2001 detector. A Kinetex® XB-C18 (5 µm, 100 Å, 250 mm x
21.2 mm; Phenomenex, Aschaffenburg, Germany) served as RP-column at a flow-rate of
15 mL/min. Mixtures of acetonitrile and 0.1% aq TFA were used as mobile phase. The
detection wavelength was set to 220 nm throughout. The solvent mixtures were removed by
lyophilization, using an Alpha 2-4 LD lyophilisation apparatus (Christ, Osterode am Harz,
Germany) equipped with an RZ 6 rotary vane vacuum pump (Vacuubrand, Wertheim,
Germany).
Analytical HPLC was performed on a system from Agilent Technologies (Santa Clara, CA,
USA) (Series 1100) composed of a G1312A binary pump equipped with a G1379A degasser,
a G1329A ALS autosampler, a G1316A COLCOM thermostated column compartment and a
G1314A VWD detector. A Kinetex® C8 (2.6 µm, 100 Å, 100 mm × 4.6 mm; Phenomenex,
Aschaffenburg, Germany) served as RP-column at a flow rate of 1 mL/min. Oven temperature
was set to 30 °C throughout. Mixtures of acetonitrile (A) and 0.05% aq TFA (B) were used as
mobile phase. The detection wavelength was set to 220 nm throughout. Solutions for injection
(40 µM) were prepared by diluting a stock solution in DMSO with a mixture of A and B
corresponding to the mixture at the start of the gradient. The injection volume was 50 µL. The
following linear gradient was applied: 0-12 min: A/B 30:70-95:5, 12-15 min: A/B 95:5.
4.1.2 General Procedure for N-Alkylation
The 1,2,3,4-tetrahydroisoquinoline derivative (1 eq), 1-(bromomethyl)-4-nitrobenzene (1 eq)
or 1-(2-bromoethyl)-4-nitrobenzene (1 eq) and K2CO3 (3 eq) were refluxed in MeCN
overnight. The reaction mixture was cooled to rt, filtered and the filtrate concentrated under
reduced pressure. The residual solid was taken up in DCM and washed with H2O twice. The
organic layer was dried over MgSO4, the solvent removed under reduced pressure and the
residue purified by flash column chromatography.
20

4.1.3 General Procedure for the Reduction of Nitro Compounds
The nitrobenzene derivative (1 eq) was dissolved in EtOH, a 10% Pd/C catalyst (10 wt%)
was added and the suspension was stirred rapidly in a hydrogen atmosphere (40 bar) at rt
overnight. The catalyst was filtered off, the solvent was evaporated and the residue was
subjected to flash column chromatography.
4.1.4 General Procedure for Azide Formation
The amine (1 eq) was dissolved in 6 M HCl aq and cooled to - 5 °C. An aqueous solution of
NaNO2 (1.0 eq) was added and the mixture was stirred for 1 h. The solution was neutralized
with an aqueous solution of NaOAc while being cooled continuously. An aqueous solution of
NaN3 (1.2 eq) was added dropwise. After another h of cooling, the mixture was slowly
warmed to rt, and the product was extracted with EtOAc. The combined organic layers were
washed with H2O and brine, dried over MgSO4 and concentrated. The product was used for
subsequent conversion without further purification.
4.1.5 General Procedure for the Ortho Acylation of 3-Bromoaniline
To a cooled (0 °C) solution of BCl3 (1 M solution in heptane, 1.1 eq) in DCM 3-bromoaniline
(1 eq) was added under a stream of nitrogen gas. The mixture was stirred for 20 min and
AlCl3 (1.1 eq) and propionitrile (1 eq) were added subsequently. Stirring continued for 24 h
under reflux. After cooling to 0 °C, 2 M HCl aq was added. The mixture was stirred rapidly
for 10 min and heated again to 80 °C for 1 h. The product was extracted with DCM and the
organic phase washed with 1 M NaOH aq, dried and concentrated. The residue was subjected
to flash column chromatography.
4.1.6 General Procedure for the Sonogashira Reaction
The bromobenzene derivative (1 eq), CuI (0.1 eq) and Pd(PPh3)2Cl2 (0.05 eq) were placed in a
sealed vial under nitrogen. A mixture of degassed THF and NEt3 (2:1) and
ethynyltrimethylsilane (1.5 eq) was added. The solution was stirred at 60 °C overnight. H2O
was added and the product was extracted with DCM. The organic phase was separated,
concentrated and purified by flash column chromatography.
4.1.7 General Procedure for TMS Deprotection
A solution of the respective trimethylsilyl-protected alkyne (1 eq) in THF was cooled in an ice
bath. Under stirring, TBAF (1.5 M in THF, 1.5 eq) was added and stirring was continued until
TLC analysis revealed complete consumption of the starting material (approximately 2 h).
The solvent was evaporated, saturated NH4Cl solution was added and the compound extracted
with DCM. The organic layer was separated, dried over Mg2SO4 and the solvent was
21

evaporated. In case of sufficient purity the product was used for subsequent reaction without
further purification. Otherwise, flash column chromatography was used to remove traces of
the starting material.
4.1.8 General Procedure for Amide Formation
DIPEA (10 eq), HBTU (3 eq) and quinoline-2-carboxylic acid (3 eq) were dissolved in
anhydrous DCM under a nitrogen atmosphere and cooled to 0 °C. The amine derivative was
added in small portions and the mixture was allowed to reach rt and stirred for 24 h. The
solution was washed with H2O and brine, twice each, dried and concentrated. The residue was
subjected to flash column chromatography.
4.1.9 General Procedure for the CuAAC
The azide (1.0-1.5 eq), the alkyne (1.0-1.3 eq), CuSO4 · 6H2O (0.1 eq), sodium ascorbate
(0.5 eq) and TBTA (0.1 eq) were dissolved in DMF or, in case of UR-St1 and UR-St2, in
THF and stirred for 24 h under a nitrogen atmosphere. DCM was added and the organic layer
was washed with H2O, dried over MgSO4 and concentrated. The crude mixture was purified
by flash column chromatography.
4.2 Biological Assays: Experimental Protocols
4.2.1 General Experimental Conditions
Materials. Commodity chemicals and solvents were purchased from commercial suppliers
(Sigma Aldrich, Munich, Germany; Merck, Darmstadt, Germany; VWR, Darmstadt,
Germany; Thermo Fisher Scientific, Waltham, MA, USA; Invitrogen, Karlsruhe, Germany).
Topotecan and vinblastine were obtained from Sigma Aldrich. Hoechst 33342 and calcein-
AM were procured from Biotium (Fremont, CA, USA). FTC was from Merck. Tariquidar was
synthesized in our laboratory according to literature[21] with slight modifications[51].
Reversan
was
obtained
from
Tocris
(Wiesbaden-Nordenstadt,
Germany).
Cholesterol/RAMEB, a complex of 5.4% cholesterol and randomly methylated β-
cyclodextrin, was acquired from CycloLab (Cyclodextrin Research and Development
Laboratory, Budapest, Hungary). Millipore water was used throughout for the preparation of
buffers, aqueous reagent solutions and HPLC eluents. The pH of buffers and aqueous reagent
solutions was adjusted with NaOH aq or HCl aq unless stated otherwise. Acetonitrile for
HPLC (gradient grade) was obtained from Merck. Mammalian cell lines were purchased from
the ATCC (American Type Culture Collection; Manassas, VA, USA), Sf9 cells were obtained
from CLS (Eppelheim, Germany). Tissue culture flasks were procured from Sarstedt
(Nümbrecht, Germany). Dulbecco’s Modified Eagle’s Medium - high glucose (DMEM/High;
22

with 4500 mg/L glucose, sodium pyruvate and sodium bicarbonate, without L-glutamine,
liquid, sterile-filtered, suitable for cell culture) and L-glutamine solution (200 mM, sterile-
filtered, BioXtra, suitable for cell culture) were from Sigma Aldrich. Insect-Xpress Medium
was obtained from Lonza (Verviers, Belgium). Fetal calf serum (FCS) and trypsin/EDTA and
were from Biochrom (Berlin, Germany). For all assays in microplate format, 96-well plates
(PS, clear, F-bottom, with lid, sterile) from Greiner Bio-One (Frickenhausen, Germany) were
used. The BaculoGold transfection kit was bought from BD Biosciences (San Jose, CA,
USA). The Bio-Rad protein assay kit was purchased from Bio-Rad Laboratories (Munich,
Germany). Syringe filters (Phenex-RC, 4 mm, 0.2 µm) used in the chemical stability assay
were from Phenomenex (Aschaffenburg, Germany).
Stock solutions. Topotecan and vinblastine were dissolved in 70% EtOH to give 100 µM
stock solutions. Hoechst 33342 stock solution (0.8 mM) was prepared in water and calcein-
AM stock solution (100 µM) in DMSO. The test compounds and the reference compounds
fumitremorgin C, tariquidar, reversan and sulfasalazine were dissolved in DMSO at 100 times
the final concentrations in the transport assays and the ATPase assay and at 1000 times the
final concentrations in the chemosensitivity assay. A 1.5 mM stock solution of sulfasalazine
in DMSO was prepared for stimulating ABCG2 in the inhibition mode of the ATPase Assay.
Furthermore, 3 mM stock solutions of the test compounds in DMSO were prepared for the
stability assay in blood plasma. If not stated otherwise, water served as solvent for other assay
reagents.
Instruments. Fluorescence and absorbance measurements in microplates were carried out
with a GENios Pro microplate reader (equipped with a Xenon arc lamp; Tecan, Grödig,
Austria). Analytical HPLC of compound 59 after incubation in murine plasma was performed
on a system from Agilent described above (in the chemistry part of the experimental section).
The HPLC conditions were as described, yet with two alterations. The following linear
gradient was applied: 0-12 min: A/B 20:80-95:5, 12-15 min: A/B 95:5. The injection volume
was 100 µL.
Software. All biological data were analyzed with GraphPad Prism 5 (GraphPad Software,
San Diego, CA, USA).
4.2.2 Cell Culture
All mammalian cells were cultured in DMEM/High supplemented with 2% (v/v) of a 200 mM
L-glutamine solution and 10% (v/v) FCS at 37 °C in a water-saturated atmosphere containing
5% CO2.
23

MCF-7/Topo cells, an ABCG2-overexpressing variant of the MCF-7 cell line (ATCC® HTB-
22), were obtained by passaging the MCF-7 cells with increasing amounts of topotecan in the
culture medium to achieve a maximum concentration of 550 nM within a period of about
40 d; after 3 passages at the maximum concentration the treated cells expressed sufficient
quantities of ABCG2 [40, 52]. They were cultured with 550 nM topotecan to maintain
overexpression of the ABCG2 transporter.
KB-V1 cells, an ABCB1-overexpressing variant of the KB cell line (ATCC® CCL-17), were
obtained by passaging the KB cells with increasing amounts of vinblastine in the culture
medium to achieve a maximum concentration of 330 nM within a period of about 90 d; after 3
passages at the maximum concentration the treated cells expressed sufficient quantities of
ABCB1 transporter [51, 53]. They were cultured with 330 nM vinblastine to maintain
overexpression of the ABCB1 transporter.
MDCK.2-MRP1 cells were a kind gift from Prof. Dr. P. Borst from the Netherland Cancer
Institute (Amsterdam, NL). They were obtained by transfecting MDCK.2 cells (ATCC® CRL-
2936) with human ABCC1 [54, 55]. Due to the strong adherence of this cell line,
trypsinization was performed using 2X trypsin/EDTA (0.1%./0.04%) for 30 min.
Sf9 cells (CLS 604328) were cultured in Insect Xpress medium supplemented with 5% (v/v)
FCS in 250 or 500 mL disposable Erlenmeyer flasks at 28 °C and under rotation at 150 rpm.
All cells were routinely monitored for mycoplasma contamination by PCR using the
Venor®GeM mycoplasma detection kit (Minerva Biolabs, Berlin, Germany) and were
negative.
4.2.3 Inhibition of ABCG2: Hoechst 33342 Transport Assay [44]
MCF-7/Topo cells were seeded into 96-well plates at a density of 20,000 cells per well
(100 µL/well) and allowed to attach to the surface of the microplates at 37 °C in a water-
saturated atmosphere containing 5% CO2 overnight. The next day, the culture medium was
removed and the cells were incubated with loading suspension (DMEM, supplemented as
described above, 8 µM Hoechst 33342 and the test compounds at increasing concentrations;
100 µL/well) for 2 h (37 °C, water-saturated atmosphere containing 5% CO2). FTC at a final
concentration of 10 µM served as reference compound (positive control); the vehicle DMSO
(1%) served as negative control. Each concentration was measured in triplicate, positive and
negative control 12-fold each. The supernatants were drained and the cells were fixed with
4% (w/w) PFA in PBS (100 µL/well) under light protection for 20 min. Afterwards, the MCF-
7/Topo cells were washed with PBS twice (250 µL/well each time) to remove residual dye
and overlaid with PBS (100 µL/well). The relative fluorescence intensities (λexc=340 nm,
24

λem=485 nm) were determined using a GENios Pro microplate reader. On each plate, the
optimal gain was calculated by the determination of the fluorescence intensities in the
presence of the reference substance FTC.
The data were normalized relative to the fluorescence intensity in the absence of an ABCG2
inhibitor (negative control) and the response elicited by the FTC (positive) control, which was
defined as 100% inhibition of the Hoechst 33342 efflux. IC50 values were calculated using
four parameter sigmoidal fits. Errors were expressed as standard error of the mean (SEM).
4.2.4 Inhibition of ABCB1: Calcein-AM Transport Assay [44]
KB-V1 cells were seeded into 96-well plates at a density of 20,000 cells per well
(100 µL/well) and allowed to attach to the surface of the microplates at 37 °C in a water-
saturated atmosphere containing 5% CO2 overnight. The next day, the culture medium was
removed and the cells were washed with loading buffer (120 mM NaCl, 5 mM KCl, 2 mM
MgCl2, 1.5 mM CaCl2, 25 mM HEPES, 10 mM glucose, pH 7.4) in order to remove
unspecific serum esterases. Afterwards, the cells were incubated with loading suspension
(loading buffer, 5 mg/mL BSA, 1.25 µL/mL 20% (w/w) Pluronic® F-127 in DMSO, 0.5 µM
calcein-AM and the test compounds at increasing concentrations; 100 µL/well) for 10 min
(37 °C, water-saturated atmosphere containing 5% CO2). Tariquidar at a final concentration of
10 µM served as reference compound (positive control); the vehicle DMSO (1%) served as
negative control. Each concentration was measured in triplicate, positive and negative control
12-fold each. The supernatants were drained and the cells were fixed with 4% (w/w) PFA in
PBS (100 µL/well) under light protection for 20 min. Afterwards, the KB-V1 cells were
washed with PBS twice (250 µL/well each time) to remove residual dye and overlaid with
PBS (100 µL/well). The relative fluorescence intensities (λexc=485 nm, λem=535 nm) were
determined using a GENios Pro microplate reader. On each plate, the optimal gain was
calculated by the determination of the fluorescence intensities in the presence of the reference
substance tariquidar.
The data were normalized relative to the fluorescence intensity in the absence of an ABCB1
inhibitor (negative control) and the response elicited by the tariquidar (positive) control,
which was defined as 100% inhibition of the calcein-AM efflux. IC50 values were calculated
using four parameter sigmoidal fits. Errors were expressed as standard error of the mean
(SEM).
4.2.5 Inhibition of ABCC1: Calcein-AM Transport Assay [41]
MDCK.2-MRP1 cells were seeded into 96-well plates at a density of 20,000 cells per well
(100 µL/well) and allowed to attach to the surface of the microplates in a water-saturated
25

atmosphere containing 5% CO2 at 37 °C overnight. The assay was performed and the data
evaluated as described for the calcein-AM Transport Assay for analyzing ABCB1 inhibitors
with two exceptions: reversan at a final concentration of 30 µM served as positive control and
the incubation time was 1 h.
4.2.6 ABCG2 ATPase Assay
Generation of recombinant baculoviruses encoding the ABCG2 protein (in the pVL1392
vector) was performed in Sf9 insect cells using the BD BaculoGold transfection kit according
to the manufacturer’s protocol. High-titer virus stock solutions were generated by 2-3
sequential amplifications. The supernatant fluid from the last amplification was stored at 4 °C
under light protection.
Membrane preparation. The protocol is based on the work of Sarkadi et al. [56]. Sf9 cells
with a density of 3 x 106 cells/mL (100 mL) were infected 1:500 with a high-titer ABCG2
baculovirus stock, incubated for 48 h and harvested by centrifugation at 4 °C and 500 g for
10 min. The cell pellet was re-suspended in Tris mannitol buffer (50 mM Tris, 300 mM
mannitol, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), pH 7; 100 mL) and centrifuged
again. Then the pellet was lysed and homogenized in TMEP buffer (50 mM Tris, 50 mM
mannitol, 1 mM EDTA, 10 µg/mL leupeptin, 10 µg/mL benzamidine, 0.5 mM PMSF, 2 mM
DTT, pH 7; 60 mL) by a Potter Elvehjem tissue homogenizer. Undisrupted cells and cellular
debris were pelleted by centrifugation at 4 °C and 500 g for 10 min and the supernatant,
containing the membranes, was removed carefully. For cholesterol loading[57, 58],
2.5 mg/mL Cholesterol/RAMEB complex (cholesterol content 5.4%) was added (150 mg
total) and the membranes were incubated at 4 °C for 20 min. After centrifugation at 4 °C and
100,000 g for 1 h the pellet (containing the membranes) was re-suspended in TMEP buffer
(30 mL), giving a protein concentration of 2.0-3.0 mg/mL, homogenized with a Potter
Elvehjem tissue homogenizer. All procedures during the membrane preparation were
performed at 4 °C and aliquots (500-1000 µL) were stored at - 80 °C until use.
Protein quantification was performed by the method of Bradford using the Bio-Rad protein
assay kit according to the manual.
Assay procedure. The assay was performed by analogy with the ABCB1 ATPase assay
procedure described by Sarkadi et al. [56]. The ATPase activity of the ABCG2 transporter
was estimated by measuring inorganic phosphate liberation. It was determined as
orthovanadate-sensitive ATPase activity in the presence of inhibitors in ABCG2 Sf9
membrane preparations. The assay was carried out in the activation mode, i.e. without
26

stimulation of ABCG2, and in the inhibition mode, where the ABCG2 transporters in the
membranes were stimulated with sulfasalazine.
Membranes containing 2.0-2.5 mg total protein were thawed on ice and pelleted by
centrifugation at 4 °C and 16,200 g for 10 min. Then they were suspended in assay buffer
(50 mM MOPS-Tris (100 mM MOPS, pH adjusted to 7.0 with 1.7 M Tris), 50 mM KCl,
5 mM NaN3, 2 mM EGTA, 2 mM DTT, 1 mM ouabain; 4.2 mL), mixed with 10% (w/w)
CHAPS (42 µL; decreases the high basal ABCG2 ATPase activity) and homogenized using a
syringe and a needle (27G). When the assay was performed in the inhibition mode, 1.5 mM
sulfasalazine (10.5 µL; final conc. 3 µM) was added to the suspension to activate ABCG2.
The suspension was split into 2 portions (2.0 mL each) and one portion was supplemented
with orthovanadate (100 mM Na3VO4, pH 10; 50 µL; final conc. 2 mM), the other with the
same amount of purified water. The two suspensions (w. and wo. orthovanadate) were
transferred into a 96-well plate on ice (40 µL/well; 20-25 µg protein/well) and pre-incubated
at 37 °C for 3 min.
The ATPase reaction was started by adding the reference and test compounds in assay buffer
containing 20 mM ATP (10 µL/well, giving a final volume of 50 µL/well; final conc. 4 mM)
with a multichannel pipette and the plate was incubated at 37 °C in a microplate shaker for
1 h. For this purpose, an ATP solution (200 mM MgxATP, 400 mM MgCl2, pH adjusted to
7.0 with 1.7 M Tris) was diluted 1:10 (v/v) with assay buffer and the test and reference
compounds were added 5-fold concentrated. Sulfasalazine at a final concentration of 30 µM
served as reference activator (positive control) in the activation mode; FTC at a final
concentration of 10 µM served as reference inhibitor (positive control) in the inhibition mode;
the vehicle DMSO (1% final content) served as negative control. Each concentration (w. and
wo. orthovanadate) was measured in duplicate, positive and negative control in quadruplicate
each.
The reaction was stopped by addition of 10% (w/w) SDS (30 µL/well); background control
(phosphate stemming from the assay buffer) (w. and wo. orthovanadate) was stopped before
starting the reaction with MgATP and was measured in duplicate.
Phosphate standards (0, 0.05, 0.1, 0.25, 1.5 and 2.0 mM NaH2PO4 in assay buffer; 50 µL/well;
each concentration measured in duplicate) were included on each plate for calibration.
The amount of phosphate was determined by adding a colorimetric reagent (1 part of
reagent A (35 mM ammonium molybdate, 15 mM zinc acetate) mixed with 4 parts of
reagent B (5% (w/w) ascorbic acid, pH 5.0; freshly prepared); 200 µL/well) and by incubating
27

at 37 °C for further 20 min. The absorbance (820 nm) as a parameter proportional to the
phosphate amount was measured, using a GENios Pro microplate reader.
Data obtained under the treatment with orthovanadate were subtracted from the data without
orthovanadate to obtain phosphate liberation resulting from ABCG2 activity. The ensuing
data were normalized relative to the absorbance in the absence of an ABCG2
activator/inhibitor (negative control) and the response elicited by the sulfasalazine (positive)
control, which was defined as 100% ATPase activity in the activation mode, or the response
elicited by the FTC (positive) control, which was defined as 0% ATPase activity in the
inhibition mode of the assay, where the membranes were activated with sulfasalazine. IC50
values were calculated using four parameter sigmoidal fits. Errors were expressed as standard
error of the mean (SEM).
4.2.7 Size-Exclusion Chromatography-Based Thermostability Assay (SEC-
TS)
Expression and purification of human ABCG2 was performed as described in our previous
report [39].
Assay procedure. The assay was carried out as described in our previous study [39]. In short:
Detergent-purified ABCG2 was incubated with or without the test compound at a
concentration of 10 µM for 10 min at room temperature. Samples were aliquoted (100 µL
each) and heated at one temperature, ranging from 30-75 °C, for 10 min in a Bio-Rad
Thermocycler. Then the samples were cooled on ice, centrifuged at 4 °C and 100,000 g for
20 min and subjected to HP-SEC, using a TSKgel G3000SWxl column (Tosoh Biosciences)
as stationary phase and a detection wavelength of 280 nm. Tm values were calculated using
four parameter sigmoidal fits. Errors were expressed as standard deviation (SD) of the best fit
values.
4.2.8 Chemosensitivity Assay [59]
MCF-7/Topo cells were seeded into 96-well plates at a density of 1500 cells per well
(100 µL/well) and allowed to attach to the surface of the microplates in a water-saturated
atmosphere containing 5% CO2 at 37 °C overnight. The next day, fresh medium containing
the test compounds at 2-fold final concentrations was added (100 µL/well; giving a final
volume of 200 µL/well). On each plate, vinblastine at a final concentration of 300 nM served
as reference cytostatic (positive control); the vehicle DMSO (0.1%) served as negative control
to monitor cell growth in the absence of a drug. Each concentration was measured 8-fold,
positive and negative control 16-fold each. Growth of the cells was stopped after different
28