Catechols and 3-hydroxypyridones as inhibitors of the DNA repair complex ERCC1-XPF

Article abstract of DOI:10.1016/j.bmcl.2015.08.031

Catechol-based inhibitors of ERCC1-XPF endonuclease activity were identified from a high-throughput screen. Exploration of the structure-activity relationships within this series yielded compound 13, which displayed an ERCC1-XPF IC₅₀ of 0.6 μM, high selectivity against FEN-1 and DNase I and activity in nucleotide excision repair, cisplatin enhancement and γH2AX assays in A375 melanoma cells. Screening of fragments as potential alternatives to the catechol group revealed that 3-hydroxypyridones are able to inhibit ERCC1-XPF with high ligand efficiency, and elaboration of the hit gave compounds 36 and 37 which showed promising ERCC1-XPF IC₅₀ values of <10 μM.

Full text of DOI:10.1016/j.bmcl.2015.08.031

Bioorganic & Medicinal Chemistry Letters 25 (2015) 4097–4103
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Catechols and 3-hydroxypyridones as inhibitors of the DNA repair
complex ERCC1-XPF
a
a
a
a
Timothy M. Chapman a, , Kevin J. Gillen , Claire Wallace , Maximillian T. Lee , Preeti Bakrania ,
Puneet Khurana a, Peter J. Coombs a, Laura Stennett a, Simon Fox a, Emilie A. Bureau a, Janet Brownlees a,
David W. Melton b, Barbara Saxty a
⇑
a Centre for Therapeutics Discovery, MRC Technology, 1-3 Burtonhole Lane, Mill Hill, London NW7 1AD, UK
b MRC Institute of Genetics and Molecular Medicine, University of Edinburgh, MRC Human Genetics Unit, Western General Hospital, Crewe Road, Edinburgh EH4 2XU, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Catechol-based inhibitors of ERCC1-XPF endonuclease activity were identified from a high-throughput
Received 24 June 2015
Revised 7 August 2015
Accepted 11 August 2015
Available online 17 August 2015
screen. Exploration of the structure-activity relationships within this series yielded compound 13, which
displayed an ERCC1-XPF IC50 of 0.6
tide excision repair, cisplatin enhancement and
l
M, high selectivity against FEN-1 and DNase I and activity in nucleo-
H2AX assays in A375 melanoma cells. Screening of frag-
c
ments as potential alternatives to the catechol group revealed that 3-hydroxypyridones are able to inhibit
ERCC1-XPF with high ligand efficiency, and elaboration of the hit gave compounds 36 and 37 which
showed promising ERCC1-XPF IC50 values of <10 lM.
Keywords:
DNA repair
ERCC1-XPF
Ó 2015 Elsevier Ltd. All rights reserved.
Catechol
3-Hydroxypyridone
Cisplatin potentiation
Platinum-based chemotherapeutics such as cisplatin result in
several forms of DNA damage, however their effectiveness can be
limited by efficient DNA repair processes. The ERCC1-XPF complex,
formed from the heterodimerization of excision-repair cross com-
plementation group 1 (ERCC1) and xeroderma pigmentosum F
(XPF), has structure-specific endonuclease activity and plays an
essential role in the removal of platinum-DNA adducts by
nucleotide excision repair (NER).1 It is also involved in interstrand
crosslink repair (ICR) and there is evidence that it is a good target
for intervention in melanoma, ovarian, lung and other cancers.1
Inhibition of ERCC1-XPF activity therefore represents an approach
to potentially enhance the effectiveness of chemotherapeutic
agents such as cisplatin, however very few inhibitors have been
reported in the literature.2
out the unwanted features and gain additional potency while
retaining good selectivity for ERCC1-XPF. Testing of initial ana-
logues showed that the catechol group was required for inhibitory
activity since mono- or di-methylation led to complete loss of
activity. Catechols are known to be effective binders of metal ions,
in particular magnesium and manganese; the activity of ERCC1-
XPF is metal-dependent and the presence of Mn2+ is required for
substrate turnover in the biochemical assay. It was therefore our
hypothesis that they could inhibit ERCC1-XPF via binding to a
metal ion at the endonuclease active site.
Synthesis and testing of early analogues showed that the
removal of the two bromine atoms and methyl group (Table 1,
compound 2) and replacing the hydrazone with an amide (3) were
all well tolerated, and this led to an improvement in potency and
ligand efficiency while retaining selectivity against FEN-1 and
DNase I. Variation of the linker lengths on the right-hand-side
portion of the amide core suggested that benzyl was preferred (4).
We sought to identify novel inhibitors of ERCC1-XPF using a
high-throughput fluorescence-based in vitro endonuclease assay3
and one of the hits obtained was compound 1 (Fig. 1) with an
IC50 value of approximately 30 lM and modest ligand efficiency.
The catechol motif is present in known nuclease inhibitors4 and
although the dibromo-substitution and hydrazone motif were con-
sidered undesirable, 1 showed good selectivity for ERCC1-XPF in
counter-screen assays against the nucleases FEN-1 and DNase I.3
We therefore sought to explore whether it was possible to design
Br
H
ERCC1-XPF IC50 = 33 µM
N
OH
OH
N
H
N
Ligand efficiency = 0.25
O
FEN-1 IC50 > 100 µM
DNase I IC50 > 100 µM
1
Br
⇑
Corresponding author.
Figure 1. Initial hit compound 1.
http://dx.doi.org/10.1016/j.bmcl.2015.08.031
0960-894X/Ó 2015 Elsevier Ltd. All rights reserved.

4098
T. M. Chapman et al. / Bioorg. Med. Chem. Lett. 25 (2015) 4097–4103
Table 1
Activity and selectivity data from analogues of initial hit
Compound
Structure
ERCC1-XPF IC50
2.0
(lM)
FEN-1 IC50
>100
(lM)
DNase I IC50
>100
(lM)
H
N
OH
OH
N
H
N
2
O
O
H
N
OH
OH
N
H
3
4
5
15.2
0.9
>100
>100
>100
>100
>100
>100
OH
H
N
N
H
OH
O
O
H
N
OH
OH
N
H
1.6
Table 2
Activity and selectivity data from left-hand side variations
O
OH
R1
N
H
OH
M)
a
a
Compound
R1
ERCC1-XPF IC50
7.3
(l
FEN-1 IC50
(
lM)
DNase I IC50
>100
(lM)
6
>100
>100
>100
O
7
8
12.1
13.3
>100
>100
9
24.5
1.8
>100
>100
>100
nt
10
11
12
1.0
>100
>100
>100
>100
Cl
Cl
0.80
Cl
13
14
15
0.63
2.6
>100
nt
>100
nt
Cl
O
1.6
>100
nt
16
17
24.0
24.6
nt
nt
>100
>100

T. M. Chapman et al. / Bioorg. Med. Chem. Lett. 25 (2015) 4097–4103
4099
Table 2 (continued)
a
a
Compound
R1
ERCC1-XPF IC50
1.4
(
lM)
FEN-1 IC50
(lM)
DNase I IC50
>100
(lM)
18
>100
N
N
19
20
21
4.0
nt
nt
25.6
7.9
nt
nt
>100
>100
N
N
O
O
N
22
23
>100
5.8
nt
>100
>100
F
>100
N
a
nt = not tested.
Table 3
Activity and selectivity data from right-hand side variations
Cl
O
R1
N
Cl
R2
Compound
NR1R2
ERCC1-XPF IC50
(
lM)
FEN-1 IC50
>100
(l
M)
DNase I IC50
>100
(lM)
OH
N
24
1.0
0.8
OH
OH
N
25
26
27
>100
>100
>100
>100
>100
>100
OH
OH
Cl
N
H
1.6
1.1
OH
OH
N
H
F
OH
Table 4
Activity and selectivity data from variations to amide core
Compound
Structure
ERCC1-XPF IC50
2.9
(lM)
FEN-1 IC50
>100
(
lM)
DNase I IC50
>100
(lM)
Cl
O
O
S
OH
OH
28
N
H
Cl
Cl
H
N
OH
OH
29
30
0.74
4.8
>100
>100
>100
>100
Cl
O
Cl
O
OH
OH
N
N
H
H
(continued on next page)

4100
T. M. Chapman et al. / Bioorg. Med. Chem. Lett. 25 (2015) 4097–4103
Table 4 (continued)
Compound
Structure
ERCC1-XPF IC50
(lM)
FEN-1 IC50
>100
(
lM)
DNase I IC50
>100
(lM)
Cl
O
31
4.1
OH
OH
N
N
OH
OH
N
N
Cl
Cl
32
33
12.2
7.8
>100
>100
>100
>100
O
OH
O
N
OH
N
H
N
(8, 9). However, the introduction of additional substituents
allowed potency to be regained, such as methyl (10), 4-chloro
(11) or 2-chloro (12). The 2,4-dichloro variant (13) was the most
H
O
N
O
Cl
H
N
OH
Br
OH
potent compound at 0.6 lM. The 4-methoxy compound (14) was
O
34
35
ERCC1-XPF IC50 ~ 100 µM
Cl
less potent than the 4-chloro, and the larger cyclohexyl variant
(15) was relatively well tolerated. Smaller alkyl groups such as iso-
propyl (16) and cyclopentyl (17) were weak, although the cyclo-
ERCC1-XPF IC50 = 24 µM
Ligand efficiency = 0.70
hexylmethyl (18) displayed an IC50 value of just over
1
Figure 2. 5-Bromo-3-hydroxypyridone fragment 34 and merged compound 35.
micromolar. The introduction of more polar or larger groups (19–
23) did not lead to potency improvements relative to 13.
Modifications to the right-hand side portion were explored
(Table 3) via amide coupling with the 2,4-dichlorobenzyl group
fixed: cyclisation (24) or N-methylation of the amide (25) did not
significantly affect potency. Halogen substituents were added to
With the benzylic catechol in place, a range of groups on the
left-hand side portion of the molecule were explored (Table 2),
with synthetic access through amide coupling. Replacing the ani-
line NH in 4 with carbon (6) or oxygen (7) both led to reductions
in potency, and shortening the linker also led to losses in potency
Table 5
Activity and selectivity data from 3-hydroxypyridone fragment substitution
R1
N
R2
R3
O
OH
R4
a
a
Compound
R1
H
R2
H
R3
R4
H
ERCC1-XPF IC50
8.7
(lM)
FEN-1 IC50
>100
(l
M)
DNase I IC50
>100
(lM)
36
F
N
N
37
H
H
H
5.2
>100
>100
Cl
F
38
39
H
H
H
H
H
H
13.1
13.5
>100
>100
>100
>100
O
N
H
40
H
H
H
H
14.6
>100
>100
N
N
41
42
Me
H
H
H
>100
55.2
>100
nt
nt
nt
nt
nt
nt
F
H
H
F
43
H
H
F
a
nt = not tested.

T. M. Chapman et al. / Bioorg. Med. Chem. Lett. 25 (2015) 4097–4103
4101
the catechol ring to explore the effect of modifying the steric and
electronic properties (26, 27) including tuning the pKa of the cate-
chol OH groups, and although this was tolerated it did not improve
potency.
Alterations to the amide core were investigated (Table 4): the
sulfonamide 28 showed reduced potency in comparison with 13,
whereas the reverse amide 29 was equipotent to the parent amide.
The ureas (30, 31) and oxadiazole variants (32, 33) also showed
weaker potency compared to 13.
Next, the potential for modifying the catechol motif was
explored; as the catechol represents a liability, especially with
respect to oxidation, replacement was desirable. Previous data sug-
gested that it would be necessary for an alternative group to have
good metal-binding capability, so in order to identify catechol vari-
ants we screened a number of fragments with metal-binding capa-
bility in our biochemical assay, with the intention that these could
then be merged with the existing left-hand side portion. This
identified 5-bromo-3-hydroxypyridone 34 as a fragment with a
long half-life; both compounds were not toxic to Hep-G2 cells up
to a compound concentration of 10 M.
l
Although the catechols and hydroxypyridones showed selectiv-
ity for ERCC1-XPF over FEN-1 and DNase I, catechols in particular
can show PAINS effects7,8 and interfere with assays through non-
specific mechanisms; therefore it was important to demonstrate
binding to target using a biophysical method. The interaction of
compounds 13 and 36 with ERCC1-XPF was examined by surface
plasmon resonance (SPR). Compound 13 showed a KD of ꢀ30
lM,
weaker than the biochemical IC50, with relatively slow binding
kinetics (Fig. 3). By contrast, the hydroxypyridone 36 showed fast
binding kinetics and a KD of 3.5 lM, in good agreement with the
biochemical IC50. Neither compound showed binding to a control
protein included in the study.
Compound 13 was profiled in three cell-based assays using
A375 melanoma cells to measure (1) direct nucleotide excision
repair (NER) using a recombinant GFP reporter assay,3,9 (2) poten-
tiation of cisplatin induced cell toxicity10 and (3) immunofluores-
promising IC50 of 24
lM and high ligand efficiency (Fig. 2); this
cence detection of c
H2AX, a marker of DNA damage.11
fragment has recently been shown to inhibit the activity of an
influenza endonuclease with a similar level of potency and bind
through a metal-based interaction.5,6
Firstly, in the NER assay, repair of UV damaged pEGFP plasmid
was monitored following treatment with compound for 24 h at
However, when it was merged with the left-hand side portion
of compound 13 to give 35, this compound showed very weak
activity against ERCC1-XPF, suggesting that the fragment might
be binding in a different way to the catechol. Instead we therefore
sought to exploit the fragment by substituting at the available
positions, and due to the relative lack of structural knowledge
150
100
around the protein binding site
employed. Firstly, replacing the bromine with a 4-fluorophenyl
group (Table 5, 36) gave a gain in potency to less than 10
and the 3-pyrazole substituted variant 37 showed an IC50 of
approximately 5 M. A wider range of R3 variants were investi-
a systematic approach was
lM
l
gated and selected examples are shown in Table 5; although none
showed improved potency compared to 37, three other variants
were within 3-fold (38–40). Substitution at the alternative vectors
R1, R2 or R4 appeared to be poorly tolerated (41–43).
50
Compound 13
Compounds 13 and 36 were profiled in ADMET assays (Table 6).
Compound 13 showed relatively short mouse and human microso-
mal half-lives and acceptable permeability, whereas 36 showed a
0
Log [M] compound 13
Table 6
Measured in vitro ADMET profiles for compounds 13 and 36
Figure 4. Compound 13 NER in A375 melanoma cell line transfected with UV
damaged pEGFP and a luciferase reporter plasmid (pGL3) to control for transfection
efficiency. The fluorescence read out generated as the pEGFP plasmid is repaired is
normalised to both luciferase levels and readings for undamaged DNA to generate
the DNA Repair Index. Inhibition of NER is detected at higher concentrations of
Compound Log D Cytotox Hep-G2 MLM
HLM
PAMPA
v-50/lM
t1/2/mins t1/2/mins Papp/nmsÀ1
13
36
2.7
2.0
>10
>10
23
>400
28
>400
39
44
compound, IC50 = 15.6 lM.
Figure 3. SPR sensorgrams for compounds 13 (left) and 36 (right).

4102
T. M. Chapman et al. / Bioorg. Med. Chem. Lett. 25 (2015) 4097–4103
concentrations of up to 60
l
M. Compound 13 showed inhibition of
combination with compound 13 as indicated by a right shift
towards higher numbers of foci per cell.
NER with an IC50 of 15.6
l
M (Fig. 4).9
Secondly, compound 13 was profiled in a cisplatin combination
In summary, we have shown that compounds containing a cat-
echol or a 3-hydroxypyridone motif inhibited ERCC1-XPF with sub-
micromolar or low micromolar potency in the endonuclease bio-
chemical assay and displayed high selectivity over the other two
nucleases tested. Exemplars from both series showed binding to
ERCC1-XPF by SPR, and the leading example compound 13 showed
inhibition of DNA repair in A375 melanoma cells across three assay
formats. Further studies are required to understand the binding
mode and cellular mechanism of action of these compounds.
assay10 (Fig. 5a). The compound alone showed a reduction in cell
number at 4.96 and 10 lM (Fig. 5b), however at compound con-
centrations where no toxicity was observed with compound only,
there was an enhancement of cisplatin activity up to 1.5-fold
(PF50) as shown in the table in Figure 5.
Thirdly, compounds were assayed in a
is a marker of DNA repair11 (Fig. 6). A delay in repair of
was observed after 6 h post treatment of cells with cisplatin in
cH2AX foci assay which
c
H2AX foci
Figure 5. Cisplatin potentiation assay dose-response curves for compound 13. A375 cells were treated in 96-well plates and cell count measured using an InCell 2000 high
content imager (GE Healthcare). Curves show activity of cisplatin alone (0 M compound) and then in combination with the indicated concentrations of compound. A shift in
cisplatin potency was observed with increasing concentration of compound up to 2.03-fold with 10 M compound (a). The compound alone showed some cell death at the 2
M (b). Results are from 4 separate experiments and the error bars are SD.
l
l
highest concentrations, 4.96 and 10
l
DMSO
150
100
50
DMSO + cisplatin (5µM)
Cpd 13 (10µM) + cisplatin (5µM)
0
0
0
0
0
0
0
0
10 20 30 40 50 60 70 80 90
10 11 12 13 14 15
gamma H2AX foci
Cumulative Gaussian -- Percents
DMSO + cisplatin (5 M) Cpd13 (10 M) + cisplatin (5 M)
Best-fit values
Mean
SEM
DMSO
38.54
0.5
µ
µ
µ
45.43
0.61
50.25
0.44
Figure 6. Compound 13 activity in a
c
H2AX foci imaging assay.
cH2AX foci count in A375 cells pretreated with compound for 1 h, exposed to 5 lM cisplatin for 1 h and
continued exposure to compound at 10
l
M for 6 h. Levels of H2AX were quantified by immunofluorescence using a specific antibody (Millipore) and results analysed on an
c
InCell 2000 High Content Imager (GE Healthcare). Relative cumulative frequency plots were used in order to visualise the shift in the response to cisplatin and compound
treatment. Compound 13 causes a delay in DNA repair as illustrated by the rightward shift of the curve showing that cells contain larger numbers of foci. The combined
treatment of cisplatin and compound 13 is significantly more than cisplatin alone (p = 0.0012 by Sum of Squares F test of the EC50). Results are from 2 representative
experiments (triplicate tests in each) and the error bars are SEM.

T. M. Chapman et al. / Bioorg. Med. Chem. Lett. 25 (2015) 4097–4103
4103
6. Parhi, A. K.; Xiang, A.; Bauman, J. D.; Patel, D.; Vijayan, R. S.; Das, K.; Arnold, E.;
Lavoie, E. J. Bioorg. Med. Chem. 2013, 21, 6435.
Acknowledgement
7. Baell, J. B.; Holloway, G. A. J. Med. Chem. 2010, 53, 2719.
8. Baell, J.; Walters, M. A. Nature 2014, 513, 481.
The authors thank David Tickle, Sadhia Khan, Nathalie Bouloc
and Zaynab Isseljee for in vitro ADMET and Martin Wear for pro-
duction of ERCC1-XPF protein.
9. NER-GFP reporter assay: it should be noted that in this assay, the luciferase
signal provides a control for transfection efficiency and for any non-NER-
related effects of the compounds. For each compound concentration, the effect
of the compound on NER was determined by dividing the GFP/luciferase ratio
for the damaged GFP plasmid by the same ratio for the non-damaged GFP
plasmid. This value was then divided by the equivalent value for control
cultures without any compound and plotted as % NER activity (DNA Repair
Index) against compound concentration to obtain the IC50. Levels of repaired
GFP expression can vary and, following normalization, larger error bars may be
generated than in samples where GFP repair is reduced to very low and
consistent levels by higher concentrations of active compound (see exemplar
compound in Fig. 4).
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2015.08.
031.
10. Cisplatin combination assay: compound activity was measured over
concentration range of 0.6–10 for 5 days to assess potential shifts in
potency of cell death caused by a dose response of cisplatin (0.1–3 M).
H2AX foci assay: A375 cells pre-treated with compound for 1 h were exposed
to 5 M cisplatin for 1 h to induce DNA damage and therefore increase levels of
H2AX foci. The cisplatin was removed and replaced with 1, 3 and 10
compound. Cells were fixed with 4% paraformaldehyde 0, 17, 24 and 48 h after
cisplatin removal and processed for immunocytochemistry using a H2AX
a
References and notes
lM
l
1. McNeil, E. M.; Melton, D. W. Nucleic Acids Res. 2012, 40, 9990.
2. Jordheim, L. P.; Barakat, K. H.; Heinrich-Balard, L.; Matera, E. L.; Cros-Perrial, E.;
Bouledrak, K.; El Sabeh, R.; Perez-Pineiro, R.; Wishart, D. S.; Cohen, R.;
Tuszynski, J.; Dumontet, C. Mol. Pharmacol. 2013, 84, 12.
3. McNeil, E. M.; Ritchie, A. M.; Astell, K. R.; Shave, S.; Houston, D. R.; Bakrania, P.;
Jones, H. M.; Khurana, P.; Wallace, C.; Chapman, T.; Wear, M. A.; Walkinshaw,
M. D.; Saxty, B.; Melton, D. W. DNA Repair 2015, 31, 19.
4. Chen, E.; Swift, R. V.; Alderson, N.; Feher, V. A.; Feng, G. S.; Amaro, R. E. ACS Med.
Chem. Lett. 2014, 5, 61.
5. Bauman, J. D.; Patel, D.; Baker, S. F.; Vijayan, R. S.; Xiang, A.; Parhi, A. K.;
Martinez-Sobrido, L.; LaVoie, E. J.; Das, K.; Arnold, E. ACS Chem. Biol. 2013, 8,
2501.
11.
c
l
c
lM
c
antibody (Millipore). The number of foci per cell was quantified using a high
content imager (InCell 2000, GE Healthcare). Cell numbers containing 0–10,
10–20, 30–40 etc foci were binned and expressed as relative cumulative
frequency.