Computer-aided drug design of small molecule inhibitors of the ERCC1-XPF protein–protein interaction

Article abstract of DOI:10.1111/cbdd.13660

The heterodimer of DNA excision repair protein ERCC-1 and DNA repair endonuclease XPF (ERCC1-XPF) is a 5′–3′ structure-specific endonuclease essential for the nucleotide excision repair (NER) pathway, and it is also involved in other DNA repair pathways. In cancer cells, ERCC1-XPF plays a central role in repairing DNA damage induced by chemotherapeutics including platinum-based and cross-linking agents; thus, its inhibition is a promising strategy to enhance the effect of these therapies. In this study, we rationally modified the structure of F06, a small molecule inhibitor of the ERCC1-XPF interaction (Molecular Pharmacology, 84, 2013 and 12), to improve its binding to the target. We followed a multi-step computational approach to investigate potential modification sites of F06, rationally design and rank a library of analogues, and identify candidates for chemical synthesis and in vitro testing. Our top compound, B5, showed an improved half-maximum inhibitory concentration (IC₅₀) value of 0.49 μM for the inhibition of ERCC1-XPF endonuclease activit, and lays the foundation for further testing and optimization. Also, the computational approach reported here can be used to develop DNA repair inhibitors targeting the ERCC1-XPF complex.

Full text of DOI:10.1111/cbdd.13660

DR JACK TUSZYNSKI (Orcid ID : 0000-0001-9976-0429)
Article type : Research Article
Computer-Aided Drug Design of Small Molecule Inhibitors of the ERCC1-XPF
Protein-Protein Interaction
Running title: Computer-aided drug design of ERCC1-XPF inhibitors
Francesco Gentile1,˩#, Ahmed H. Elmenoufy2,3#, Gloria Ciniero4,5, David Jay6, Feridoun Karimi-
Busheri6, Khaled H. Barakat7, Michael Weinfeld6,8, Frederick G. West2,8 and Jack A. Tuszynski1,4,6
1Department of Physics, University of Alberta, Edmonton, AB T6G 2E1, Canada
2Department of Chemistry, University of Alberta, Edmonton, AB T6G 2G2, Canada
*
3Department of Pharmaceutical Chemistry, College of Pharmacy, Misr University for Science and
Technology, P.O. Box: 77, 6th of October City 12568, Egypt
4Department of Mechanical and Aerospace Engineering, Politecnico di Torino, Torino 10129, Italy
5Università di Torino, Torino 10124, Italy
6Department of Oncology, Cross Cancer Institute, University of Alberta, Edmonton, AB T6G 1Z2,
Canada
7Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, Edmonton, AB T6G 2H1,
Canada
This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process, which may lead to
differences between this version and the Version of Record. Please cite this article as doi:
10.1111/CBDD.13660
This article is protected by copyright. All rights reserved

8Cancer Research Institute of Northern Alberta, University of Alberta, Edmonton, AB T6G 2E1,
Canada
˩Current address: Vancouver Prostate Centre, University of British Columbia, Vancouver, BC V6H
3Z6, Canada
#These authors contributed equally to this work
*Correspondence: jackt@ualberta.ca, tel. +1-780-964-4517
Acknowledgments
This work was supported by an Alberta Cancer Foundation Transformative Program (project 26603)
awarded to MW, FGW and JAT. FG was supported by an Alberta Innovates graduate student
scholarship and a Novartis Pharmaceuticals Canada Inc. graduate scholarship. AHM was supported
by a Marshall Syska scholarship. Computer simulations were carried out using the Pharmamatrix
cluster at the University of Alberta.
Conflict of interest
The authors declare no conflict of interest.
Abstract
The heterodimer of DNA excision repair protein ERCC-1 and DNA repair endonuclease XPF
(ERCC1-XPF) is a 5´-3´ structure-specific endonuclease essential for the nucleotide excision repair
(NER) pathway, and it is also involved in other DNA repair pathways. In cancer cells, ERCC1-XPF
plays a central role in repairing DNA damage induced by chemotherapeutics including platinum-
based and crosslinking agents, thus its inhibition is a promising strategy to enhance the effect of these
therapies. In this study, we rationally modified the structure of F06, a small molecule inhibitor of the
ERCC1-XPF interaction (Jordheim et al., 2013), to improve its binding to the target. We followed a
multi-step computational approach to investigate potential modification sites of F06, rationally design
and rank a library of analogues, and identify candidates for chemical synthesis and in vitro testing.
This article is protected by copyright. All rights reserved

Our top compound, B5, showed an improved half-maximum inhibitory concentration (IC50) value of
0.49 µM for the inhibition of ERCC1-XPF endonuclease activity, and lays the foundation for further
testing and optimization. Also, the computational approach reported here can be used to develop DNA
repair inhibitors targeting the ERCC1-XPF complex.
Keywords
Computer-aided drug design, molecular dynamics, virtual screening, DNA repair, ERCC1-XPF, small
molecules, protein-protein interaction, chemotherapy
Introduction
Genomic instability is associated with the aging process (López-Otín, Blasco, Partridge, Serrano, &
Kroemer, 2013) and it is an enabling characteristic to acquire the hallmarks of cancer (Hanahan &
Weinberg, 2011). Therefore, it is not surprising that cells require a well-functioning DNA repair
apparatus to be protected from endogenous and exogenous damaging agents and to preserve their
healthy status. Likewise, defective DNA repair pathways are associated with rare genetic diseases,
such as xeroderma pigmentosum, Cockayne’s syndrome and trichothiodystrophy, whose affected
individuals show propensity for cancer, neurological deficiencies and physical abnormalities
(Bukowska & Karwowski, 2018; Cleaver, Lam, & Revet, 2009). Although DNA repair is essential to
maintain genomic stability, these pathways can interfere with cancer therapies by repairing the
damage inflicted on tumor cells, allowing cancer to progress. Combination cancer therapy is a
relatively new strategy in which a DNA-damaging agent (e.g. cisplatin) and a DNA repair inhibitor
are jointly administered to the patient in order to potentiate the effect of the former (Helleday,
Petermann, Lundin, Hodgson, & Sharma, 2008). The therapeutic potential of DNA repair inhibitors
was recently demonstrated by the successful development of inhibitors for poly(ADP-ribose)
polymerase (PARP) enzymes, with several molecules that either were approved for clinical use or
entered clinical trials (Ohmoto & Yachida, 2017). Despite the increasing interest in DNA repair
proteins as potential drug targets, there are very few known inhibitors besides the aforementioned
PARP inhibitors. In particular, there are a number of other, complementary DNA repair mechanisms
that could be exploited for anti-cancer therapy (Brown, O’Carrigan, Jackson, & Yap, 2017).
This article is protected by copyright. All rights reserved

The heterodimeric protein ERCC1-XPF, made up of the DNA excision repair protein ERCC1 and the
DNA repair endonuclease XPF, is a 5´-3´ structure-specific endonuclease which cleaves double-
strand/single-strand DNA junctions (McNeil & Melton, 2012). The catalytic activity of ERCC1-XPF
is essential for the nucleotide excision repair (NER) pathway in order to repair bulky, helix-distorting
damages such as UV-induced cyclobutane pyrimidine dimers, and is also involved in the repair of
double-strand breaks (DSBs) and inter-strand crosslinks (ICLs) (Faridounnia, Folkers, & Boelens,
2018) (see Figure 1). NER and ICL repair are the main mechanisms responsible for removing damage
induced by platinum-based and other crosslinking chemotherapy. Inhibiting these processes sensitize
cancer cells to chemotherapy (Arora, Kothandapani, Tillison, Kalman-Maltese, & Patrick, 2010;
McNeil & Melton, 2012; Mendoza et al., 2013), hence ERCC1-XPF is an attractive target for
developing DNA repair inhibitors (McNeil & Melton, 2012).
Due to its central role in mediating therapeutic damage inflicted to cancer cells, ERCC1-XPF has
been the subject of a number of studies aimed at developing DNA repair inhibitors (Gentile, A.
Tuszynski, & H. Barakat, 2016). A first line of research was devoted to the discovery of inhibitors of
the NER-specific interaction between the DNA repair protein complementing XP-A cells (XPA) and
ERCC1 (Barakat et al., 2012; Gentile, Tuszynski, & Barakat, 2016). ERCC1-XPF is recruited to the
damaged zone through this interaction in NER, and XPA-ERCC1 inhibitors showed promising
preclinical potential to enhance the effect of cisplatin on tumors, although this approach would be
effective on NER only without affecting other ERCC1-XPF-mediated pathways that do not involve
XPA (Barakat et al., 2012). The catalytic site of the endonuclease has been explored as a target for
inhibition as well. The site is on the XPF nuclease domain and contains a number of charged residues
and metal ions, which make it an attractive target for small molecule binding. Recently, several
inhibitors have been reported, some of which have been shown to enhance chemotherapy in cell and
xenograft tumor models (Arora et al., 2016; Chapman, Gillen, et al., 2015; Chapman, Wallace, et al.,
2015; Gentile, Barakat, & Tuszynski, 2018; McNeil et al., 2015). However, the similarity between
catalytic sites of several human divalent cation-based DNA-cleaving enzymes constitutes an obstacle
to designing XPF-specific inhibitors, and an experimental structure of the human XPF nuclease
domain has not been solved yet (McNeil & Melton, 2012). Lastly, the ERCC1-XPF protein-protein
interaction represents another interesting target for pharmacological inhibition since dimerization is
This article is protected by copyright. All rights reserved

required to generate a functional endonuclease. Despite the high-affinity, hydrophobic-driven
interaction between the double helix-hairpin-helix (HhH2) domains of ERCC1 and XPF, the
interaction is specific to ERCC1-XPF and the deletion of a single residue (F293) at the ERCC1 side is
enough to abrogate dimerization and nuclease activity. These characteristics make the ERCC1-XPF
dimerization interface a promising, yet challenging, drug target for DNA repair inhibition (McNeil &
Melton, 2012).
Some of us previously reported F06, the first ERCC1-XPF dimerization inhibitor, which binds at the
XPF pocket interacting with ERCC1’s F293, disrupts the protein-protein interaction, selectively
blocks ERCC1-XPF-mediated DNA repair, and sensitizes cancer cell lines to UV and DNA-damaging
drugs such as cisplatin and mitomycin C (Jordheim et al., 2013). More recently, we successfully
combined computer-aided drug design (CADD) with a structure-activity relationship approach to
extend the piperazine ring of F06 with different substituents, in order to improve the biological
activity of this class of compounds. In this way we assessed F06 as a suitable scaffold for developing
DNA repair inhibitor drug candidates (Elmenoufy et al., 2019). Building on our previous results, we
report a new CADD strategy for further improvement of F06, which uses molecular dynamics (MD)
to characterize in detail the structural changes introduced by F06 binding to XPF compared to XPF
free form and ERCC1-XPF heterodimer. We also evaluated the hydrogen bond network of F06 with
XPF in order to design F06 analogues with enhanced protein interaction patterns. In contrast to our
earlier study where we modified a solvent-exposed group of F06 (Elmenoufy et al., 2019), here we
focused on different modifications such as truncation of the lateral branch of F06 and substitution of
the methoxy group with hydrogen bond acceptor groups to improve buried ligand-receptor
interactions. Through our CADD-driven medicinal chemistry strategy, we developed B5, an analogue
showing 3-fold increased inhibition of the ERCC1-XPF incision activity in vitro. Our results confirm
F06 as a suitable scaffold for designing DNA repair inhibitors, and propose a rational approach to
improve the affinity of the compounds to the target.
Methods
Molecular dynamics of XPF HhH2 domain alone, F06-bound and ERCC1-bound
This article is protected by copyright. All rights reserved

We computationally predicted F06 as binding to a pocket of the HhH2 domain of XPF, replacing the
side chain of F293 from ERCC1 in the dimer (Jordheim et al., 2013). Using classical MD simulations,
we compared the unbound, F06-bound and ERCC1-bound form of the HhH2 domain of XPF
(residues 822 to 905). For the F06-XPF complex, we used the structure previously obtained from
docking simulations (Elmenoufy et al., 2019). The unbound and ERCC1-bound forms were extracted
from the NMR ensemble of ERCC1-XPF dimerized HhH2 domains (PDB 1Z00) (Tripsianes et al.,
2005), where the bound ERCC1 structure was excluded and included in the simulation, respectively.
The protonation states of the systems were assigned using the H++ server by setting pH=7, a salinity
of 0.15 M, a dielectric constant of 10 for the solute and a solvent dielectric constant of 80 as input
parameters (Anandakrishnan, Aguilar, & Onufriev, 2012). In AmberTools14 tleap (Case et al., 2015),
we assigned Amber ff14SB force field parameters to the proteins (Maier et al., 2015), and the 12-6-4
parameters for mono- and divalent ions in TIP3P water to the ions as previously recommended (Li,
Roberts, Chakravorty, & Merz, 2013; Li, Song, & Merz, 2015). We assigned the parameters of the
Generalized Amber Force Field (GAFF) (Wang, Wolf, Caldwell, Kollman, & Case, 2004) and AM1-
BCC charges (Jakalian, Jack, & Bayly, 2002) to F06 using Antechamber (Wang, Wang, Kollmann, &
Case, 2005). The three systems were solvated with octahedral boxes of TIP3P explicit water
molecules (Jorgensen, Chandrasekhar, Madura, Impey, & Klein, 1983) with 15 Å of buffer. The
required numbers of Na+ and Cl- ions were added to neutralize the overall charge of the system and
establish a 0.15 M physiological ionic concentration. We then ran MD simulations of the systems
using pmemd.cuda (Salomon-Ferrer, Götz, Poole, Le Grand, & Walker, 2013) with the following
procedure: relaxation of ions and water molecules through 1000 steps of steepest descent and 1000
steps of conjugate gradient minimization, keeping protein and ligand atoms harmonically restrained
with a force constant of 500 kcal/mol/Å2. 2000 steps of steepest descent, followed by 3000 conjugate
gradient steps without any restraint. Gradual heating from 0 to 300 K over 100 ps using the Langevin
thermostat, keeping the backbone and ligand heavy atoms harmonically restrained (2 kcal/mol/Å2),
and using a time step of 0.5 fs and NVT conditions. Gradual release of the restraints in four NPT
phases of 50 ps where the force constant was reduced by 0.5 kcal/mol/Å2 at each phase, with a time
step of 2 fs. 250 ns of NPT production simulation with a time step of 2 fs. The cut-off for long-range
interactions was set to 9 Å.
This article is protected by copyright. All rights reserved

After the simulations, we calculated the mass-weighted root-mean square deviation (RMSD) of the
heavy atoms of protein backbone and ligand using cpptraj from AmberTools14. We also carried out a
secondary structure analysis over the MD trajectories using the dictionary of protein secondary
structure (DSSP) method from cpptraj (Kabsch & Sander, 1983).
Analysis of the binding mode of F06
In order to investigate the dynamics of the F06-XPF interactions, we analyzed the hydrogen bonds
occurring between F06 and the receptor and F06 and solvent, respectively, during the equilibrated part
of the MD trajectory, using cpptraj. We defined as hydrogen bonds non-bonded interactions occurring
between donor-hydrogen and acceptor atoms, which deviated no more than 60° from linearity and
with a maximum distance of 3.5 Å between the two heavy atoms.
Using the MOE (Molecular Operating Environment) Pharmacophore Query Editor and the EHT
scheme, we generated a set of essential pharmacophore features of F06, based on the docking pose of
the compound (Lin, 2004). Electrostatic maps of the binding site were generated as reported
previously (Elmenoufy et al., 2019).
Computer-aided design of F06 analogues
As a starting point for our computer-aided generation of analogues, we used the previously obtained
docking pose of F06. We created an extensive, non-redundant collection of multi-conformational
fragments by merging the MOE Linker library (Chemical Computing Group Inc., 2015), the
ChEMBL fragment database (Chemical Computing Group Inc., 2015), the Cambridge Structure
Linker database (Groom, Bruno, Lightfoot, & Ward, 2016), and a collection of fragments derived
from the ZINC15 3D database (Sterling & Irwin, 2015). This last library was generated using the
MOE2015 SD Pipelining Command tools (Chemical Computing Group Inc., 2015) to obtain different
protonation states and tautomers, remove non-lead-like, reactive and large-ring structures, and
generate molecular fragments. Up to five conformations for each ZINC15-derived fragment were
obtained using Conformation Import and imposing a strain limit of 4 kcal/mol. We then used the
collection of fragments for MOE2015 Scaffold Replacement (Chemical Computing Group Inc., 2015)
of F06 substructures. We also used the MOE MedChem transformation tool (175 modification rules)
This article is protected by copyright. All rights reserved

(Chemical Computing Group Inc., 2015) on the same F06 groups, with 6 iterations and a protein-
ligand clash energy limit of 0.5 kcal/mol/ligand atom. For both Scaffold Replacement and MedChem
transformations, we ran minimization of the resulting poses with protein side chains free to move and
generalized Born Volume Integral/Weighted Surface Area (GBVI/WSA) function for scoring (Labute,
2008).
Structure-based virtual screening
In order to rank the analogues generated in the previous step, we adopted the same structure-based
virtual screening (VS) strategy that we successfully used before (Elmenoufy et al., 2019). Molecular
docking simulations were run with MOE Dock (Chemical Computing Group Inc., 2015). The
pharmacophore model was used for the placement step, in which 30 high-scoring poses were returned,
based on the London dG scores (Labute, 2008). Poses were minimized together with the side chains
of the binding site (Induced Fit), and one pose was returned as the top-scored by the GBVI/WSA
scoring function. We calculated logP values of the molecules using the SlogP function in MOE
(Wildman & Crippen, 1999), as well as docking ligand efficiencies (final score/number of ligand
heavy atoms).
We ran MD simulations of the top 200 analogue-XPF complexes resulting from docking. The systems
were prepared in the same way as the F06-XPF system, and simulation setups were the same except
for the length of the production simulations (2 ns). For MD snapshots extracted every 10 ps from the
production simulations, we calculated the enthalpic portion of the binding energy using the Molecular
Mechanics/Generalized Born Surface Area (MM/GBSA) method (Onufriev, Bashford, & Case, 2004)
implemented in the MMPBSA.py script (Miller et al., 2012). In MM/GBSA, the free energy change
due to ligand binding is calculated as
(
)
(Eq. 1)
∆퐺푏푖푛푑, 푠표푙푣 = ∆퐺푀푀, 푣푎푐 +∆퐺푠표푙푣, 푐표푚푝푙푒푥 ― ∆퐺푠표푙푣, 푙푖푔푎푛푑 +∆퐺푠표푙푣, 푝푟표푡푒푖푛 ―푇∆푆
where
푀푀, 푣푎푐 includes averaged non-bonded molecular mechanics terms (electrostatic and van der
∆퐺
Waals) occurring between protein and ligand. Solvation terms are modeled as
(Eq. 2)
∆퐺푠표푙푣 = ∆퐺푠표푙푣,푝표푙푎푟 +∆퐺푠표푙푣,푛푝표푙푎푟
This article is protected by copyright. All rights reserved

where the polar contribution of the solvent is calculated by the Generalized Born equation (Genheden
& Ryde, 2015). Here, we used the GBOCBII model and the modified radii from Onufriev et al.
(Onufriev et al., 2004), and an ionic concentration of 0.15 M. The hydrophobic contribution to the
solvation free energy is calculated as
(Eq. 3)
with (surface tension) as 0.005 kcal/mol/Å2, and the solvent-accessible surface area (
∆퐺푠표푙푣, 푛푝표푙푎푟 = 훾 ∙ 푆퐴푆퐴
)
푆퐴푆퐴
γ
calculated using the linear combinations of pairwise overlaps (LCPO) model method (Weiser,
Shenkin, & Still, 1999). Averaged values were estimated for the ligand-receptor (LR) complex,
receptor (R) and ligand (L) alone, and final values were calculated as a difference between these three
values (LR-R-L).
The entropy change is represented in the free energy equation by the
term. was calculated for
푆
푇∆푆
each trajectory by using the Amber normal mode analysis (NMA) method for the LR, R and L
systems and then calculating the difference between the three systems (Kassem, Ahmed, El-Sheikh, &
Barakat, 2015). Pairwise decomposition of binding energies was also carried out for the top
compounds and the residues of the XPF pocket.
Preparation of F06-based analogues (Gen B compounds)
General synthesis of final Gen B compounds was achieved through a nucleophilic substitution
reaction of the Gen B1 (2-hydroxy 6,9 chloroacridine) and acyl chloride or chloride substituted
derivatives to afford compounds Gen B3 and Gen B4, respectively. Synthesis of Gen B1 was
achieved through dealkylation of 2-methoxy 6,9 chloroacrdine using BBr3 in dichloromethane under
nitrogen for 8 hours at room temperature. Syntheses of F06, Gen B2 and Gen B5 were performed
according to our reported procedure (Elmenoufy et al., 2019) with minor modifications, which was
through a one pot three sequential addition reaction; Mannich reaction of p-acetamidophenol with
formaldehyde and appropriate secondary amine in 2-propanol was carried out under reflux for 12 h.
The solvent and excess of the unreacted formaldehyde from the resulting mixture were removed under
vacuum and, without isolating the compound, the resulting viscous residue was treated with 6 M HCl
This article is protected by copyright. All rights reserved

to deacetylate the acetamido group to furnish the primary amine. Afterwards, an equimolar amount of
2-methoxy 6,9 chloroacridine or 2-hydroxy 6,9 chloroacridine was added, affording after heating
compounds F06 or Gen B2 and Gen B5 respectively in moderate to good yields after isolation. The
synthesis is general, facile and reproducible. The high-performance liquid chromatography (HPLC)
analyses were performed using an Agilent 1100 LC/MSD instrument. Elution was done with a
gradient of 10−95% solvent B in solvent A (solvent A was 0.1% TFA in water, and solvent B was 0.1%
acetic acid in MeCN) through an Agilent column eclipse XDB- C18 (4.6 × 250 mm, 5 μm) column at
1.0 mL/min. Area % purity was measured at 210 and 254 nm. The purity of the most active
compound Gen B5 was assessed by HPLC (>98% purity). All synthesized compounds were
1
13
characterized by H NMR, C NMR, HRMS, IR and the purity of most active compound B5 was
determined by HPLC.
ERCC1-XPF protein preparation
Human ERCC1-XPF wild-type proteins were obtained as previously described (Bowles et al., 2012;
Elmenoufy et al., 2019).
Microplate fluorescence incision assay
ERCC1-XPF incision was measured in a microplate incision assay as previously described (Bowles et
al., 2012; Elmenoufy et al., 2019).
Results and discussion
Molecular dynamics simulations
The MD simulation of the unbound form of XPF revealed a dramatic change from the initial structure
extracted from the complex with ERCC1, with ~8 Å of difference in RMSD (shown in red in Figure
2a) from the ERCC1-bound form (shown in green in Figure 2a). Remarkably, despite the small size of
F06 compared to the ERCC1 HhH2 domain, the binding of the ligand to the XPF pocket (which hosts
ERCC1’s F293 in the heterodimer) led to a stable XPF conformation which diverged from the
ERCC1-bound equilibrated structure by only ~2 Å in terms of RMSD (shown in blue in Figure 2a).
This article is protected by copyright. All rights reserved

MD results suggest that the binding of F06 to the F293 hotspot-interacting pocket on XPF (de Laat,
Sijbers, Odijk, Jaspers, & Hoeijmakers, 1998; McNeil & Melton, 2012; Sijbers et al., 1996;
Tripsianes et al., 2005) can substitute for ERCC1 as a binding partner of XPF and induce an
conformation close to the ERCC1-bound one. The RMSD of the heavy atoms of F06 during the
production MD simulation fluctuated around an average value of 1 Å from the starting conformation,
indicating a stable pose of the ligand within the binding site of XPF (Figure 2b).
To assess further the structural differences between the three XPF models, we extracted and
superposed the lowest potential energy structures of XPF from the MD trajectories. The ERCC1-
bound (shown in green in Figure 3a) and F06-bound (shown in blue in Figure 3a) form showed a
similar tertiary structure, suggesting again that F06 may mimic the natural ERCC1 partner and cause
subsequently a similar conformation of XPF to be present. The unbound XPF (shown in red in Figure
3a) showed a less ordered conformation compared to the other two structures. The main structural
change observable in the unbound form was the loss of the HhH2 typical domain structure and the
disruption of the H1 alpha helix (residue 834-845) (Figure 3b), as revealed also by the DSSP analysis
performed over the last 100 ns of MD simulation (Figure 3c). Importantly, H1 is part of the F06
binding site. Overall, the ERCC1-bound and the F06-bound structures showed more globular shapes
compared to the unbound form, with less exposure of the hydrophobic regions devoted to interact
with ERCC1, in agreement with previous experimental observations (McNeil & Melton, 2012;
Tripsianes et al., 2005).
Interactions of F06 with the XPF binding site
In our earlier work (Elmenoufy et al., 2019) we synthesized and tested analogues generated by
modifying an hydrophobic, solvent-exposed moiety of F06. In the present work, we focused on polar
groups of F06. We carried out hydrogen bond analysis of the last 100 ns of the XPF-F06 trajectory, in
order to assess polar interactions established by the ligand with the receptor and solvent molecules
(Table 1). The charge-assisted hydrogen bond between the N2-H atoms of F06 (Figure 4) and the
negatively charged side chain of E831 of XPF was the most prevalent interaction, present in 37% of
the last 100 ns of the MD trajectory. N3 of F06 also established hydrogen bond interactions with the
backbone atoms of V859 and M856 for 26% of the equilibrated trajectory. N4 was the main hydrogen
This article is protected by copyright. All rights reserved

bond acceptor atom of F06, establishing a hydrogen bond with N834 of XPF for 9% of the 100 ns
(Table 1). These three interactions showed also good hydrogen bond average geometries in terms of
distances (2.88 – 3.10 Å) and angles (~160o) (Schaeffer, 2015). The other polar moieties of F06 were
involved in hydrogen bonds with solvent molecules during the simulation, rather than specific
interactions with XPF (Table 1). Interestingly, O2 of F06 was the only polar atom not involved in any
hydrogen bond with XPF, despite its buried position and both acceptor and donor-favourable
interaction spots present around it (Figure S1 in Supplementary Material).
Virtual screening of F06 analogues
We generated a virtual library of analogues by modifying two groups of F06 (Figure 5), namely 1) the
methoxy group of site 1, in order to establish an additional favourable hydrogen bond; and 2) the
lateral branch of the acridine scaffold of F06 to reduce the solvent-exposed portion of the compound
(site 2). We also included the piperazine ring extension of compound 4, the most active F06 analogue
identified in the previous medicinal chemistry campaign (Elmenoufy et al., 2019).
The resulting library includes 5650 structures. Because different features were introduced in the
analogues, we defined an essential pharmacophore model based on the conserved acridine core, with
three aromatic features accounting for the aromatic rings anchoring the compounds to the
hydrophobic floor of the binding site (Figure S2 in Supplementary Material), and we used it for
pharmacophore-assisted docking in MOE, followed by MD-MM/GBSA-based rescoring. Among the
top ranked virtual hits, we selected 5 compounds for synthesis based on visual inspection of their
poses, MM/GBSA scores, and synthetic chemistry feasibility. These compounds (Figure 6) were
synthesized successfully and their activities were validated with a cell-free assay of ERCC1-XPF
incision activity (discussed below). Gen B1 was a truncated version of F06 with the hydroxy acridine
rather than methoxy group branched out of the acridine. Gen B2 was a more complete analogue of
F06 but with the hydroxy group replacing the methoxy group. Gen B3 and B4 were functionalized to
have phenyl groups tethered with ester or ether linkage, respectively, at site 1. Gen B5 was composed
of N-methyl piperazine extension at site 1 and the hydroxy acridine at site 2. Binding energies and
other computationally derived values for F06 and the series of analogues are reported in Table 2.
This article is protected by copyright. All rights reserved

Inhibition of ERCC1-XPF DNA incision activity by F06 analogues
In order to assess the ability of the F06 analogues to inhibit ERCC1-XPF, we used an in vitro real-
time fluorescence-based incision assay that we previously developed (Elmenoufy et al., 2019). In this
assay, the activity of ERCC1-XPF results from an increase of time-dependent fluorescent signal
caused by the release of a fluorescent 8-base DNA fragment upon cleavage (Figure 7). Compared
with the absence of compounds, a lower initial velocity of the reaction (V0, signal curve slope) is
expected when an ERCC1-XPF inhibitor is present. Two analogues showed increased activity when
compared to F06, namely compound B2 and B5. In particular, B5 was the best inhibitor of this
generation of analogues, and the second-highest scored structure from VS; the calculated logP value
for B5 (2.31) was also the lowest observed between all the F06 analogues we have investigated so far
(Elmenoufy et al., 2019). Different concentrations of B5 ranging from 0 to 10 µM were plotted
against the velocities in presence of the inhibitor at such concentrations, and half-maximum inhibitory
concentration (IC50) was estimated as 0.49 ± 0.04 µM, while IC50 of F06 was 1.86 ± 0.25 µM
(Elmenoufy et al., 2019). The improved inhibitory activities of B2 and B5 suggest the replacement of
the methoxy group with a hydrogen bond donor group together with the conservation or extension of
the piperazine ring as a viable strategy for further improving the F06 structure. Interestingly, the
modest values of binding energies calculated computationally for B1 were compensated by a very
satisfactory ligand efficiency (-0.31 kcal/mol/heavy atom) and an inhibitory activity was observed in
the assay, suggesting that this analogue could be considered for further investigations due to its small
size and the absence of the lateral branch exposed to solvent in F06.
Since B5 was the most active analogue of the series, we carried out a detailed analysis of its predicted
binding pose based on the 2 ns-long MD simulation and MM-GBSA calculations, and compared it
with the analysis for F06. The lowest potential energy conformation of the XPF-B5 complex extracted
from the 2 ns MD simulation (Figure 8a) showed a hydrophobic stacking of the acridine moiety of B5
between the side chains of Y833 and K860 of F06 (Figure 8a and b), resulting in more favourable
contributions to the overall binding affinity by these residues compared to XPF-F06 (Figure 8c).
N834 was involved in a hydrogen bond with B5, similarly to F06 (Figure 8a and b). The hydroxyl
group substituting the methoxy in B5 was involved in a new hydrogen bond established with the side
This article is protected by copyright. All rights reserved

chain of V859, not observed for F06 or our previous compound 4 (Elmenoufy et al., 2019). Consistent
with our previous observations, the acridine moiety was buried inside the XPF pocket while the
piperazine ring extension derived from compound 4 resulted in enhanced interactions with the solvent
and E829 (Figure 8b), although the latter was less strong than the one observed for F06 (Figure 8c).
Conclusions
Interest is growing in the role of the ERCC1-XPF endonuclease as a drug target in cancer treatment,
due to its critical role in repairing DNA damage inflicted by platinum-based and cross-linking
chemotherapy. Abnormal regulation of levels of XPF, ERCC1 or related proteins has been linked
with modulation of the response to DNA-damaging treatments, either showing resistance in the case
of over-regulation or sensitization in the case of down-regulation. Thus, ERCC1-XPF inhibition is
viable strategy to circumvent DAN damage-related drug resistance and improve the outcome of
cancer therapy.
In this work we focused on designing small molecules able to disrupt the ERCC1-XPF protein-protein
interaction, since heterodimerization of ERCC1 and XPF is required to constitute a functional
complex (McNeil & Melton, 2012). We identified two compounds, B2 and B5, with improved
inhibition of the ERCC1-XPF incision activity compared to the parent compound F06. Compound B1,
a truncated version of B2 and B5, may also be a promising starting structure for further modifications
due to its modest inhibitory activity despite the replacement of the bulky lateral branch of F06 with a
chlorine.
To conclude, the workflow we introduced here provides a blueprint for future modification of other
sites of F06 in order to develop preclinical drug candidates for DNA repair inhibition. Likewise, the
set of analogues described in this study, especially B1, B2 and B5, and their structure-activity
relationships provide important information for the lead optimization process of F06. We plan to
design a next generation of compounds aiming to further optimize the lead compound B5, which will
be followed by more advanced in vitro and in vivo experimental testing.
Data availability statement
This article is protected by copyright. All rights reserved

The data that support the findings of this study are available from the corresponding author upon
reasonable request.
References
Anandakrishnan, R., Aguilar, B., & Onufriev, A. V. (2012). H++ 3.0: automating pK prediction and
the preparation of biomolecular structures for atomistic molecular modeling and simulations.
Nucleic Acids Research, 40(Web Server issue), W537-41. https://doi.org/10.1093/nar/gks375
Arora, S., Heyza, J., Zhang, H., Kalman-Maltese, V., Tillison, K., Floyd, A. M., … Patrick, S. M.
(2016). Identification of small molecule inhibitors of ERCC1-XPF that inhibit DNA repair and
potentiate cisplatin efficacy in cancer cells. Oncotarget, 7(46), 75104–75117.
https://doi.org/10.18632/oncotarget.12072
Arora, S., Kothandapani, A., Tillison, K., Kalman-Maltese, V., & Patrick, S. M. (2010).
Downregulation of XPF-ERCC1 enhances cisplatin efficacy in cancer cells. DNA Repair, 9(7),
745–753. https://doi.org/10.1016/j.dnarep.2010.03.010
Barakat, K. H., Jordheim, L. P., Perez-Pineiro, R., Wishart, D., Dumontet, C., & Tuszynski, J. A.
(2012). Virtual screening and biological evaluation of inhibitors targeting the XPA-ERCC1
interaction. PloS One, 7(12), e51329. https://doi.org/10.1371/journal.pone.0051329
Bowles, M., Lally, J., Fadden, A. J., Mouilleron, S., Hammonds, T., & Mcdonald, N. Q. (2012).
Fluorescence-based incision assay for human XPF-ERCC1 activity identifies important elements
of DNA junction recognition. Nucleic Acids Research, 40(13), e101.
https://doi.org/10.1093/nar/gks284
Brown, J. S., O’Carrigan, B., Jackson, S. P., & Yap, T. A. (2017). Targeting DNA Repair in Cancer:
Beyond PARP Inhibitors. Cancer Discovery, 7(1), 20–37. https://doi.org/10.1158/2159-
8290.CD-16-0860
Bukowska, B., & Karwowski, B. T. (2018). Actual state of knowledge in the field of diseases related
with defective nucleotide excision repair. Life Sciences, 195, 6–18.
This article is protected by copyright. All rights reserved

https://doi.org/10.1016/J.LFS.2017.12.035
Case, D. ., Berryman, J. T., Betz, R. M., Cerutti, D. S., Cheatham III, T. E., Darden, T. A., …
Kollman, P. A. (2015). Amber 15. https://doi.org/citeulike-article-id:2734527
Chapman, T. M., Gillen, K. J., Wallace, C., Lee, M. T., Bakrania, P., Khurana, P., … Saxty, B. (2015).
Catechols and 3-hydroxypyridones as inhibitors of the DNA repair complex ERCC1-XPF.
Bioorganic & Medicinal Chemistry Letters, 25(19), 4097–4103.
https://doi.org/10.1016/j.bmcl.2015.08.031
Chapman, T. M., Wallace, C., Gillen, K. J., Bakrania, P., Khurana, P., Coombs, P. J., … Saxty, B.
(2015). N-Hydroxyimides and hydroxypyrimidinones as inhibitors of the DNA repair complex
ERCC1-XPF. Bioorganic & Medicinal Chemistry Letters, 25(19), 4104–4108.
https://doi.org/10.1016/j.bmcl.2015.08.024
Chemical Computing Group Inc. (2015). Molecular Operating Environment (MOE), 2015. Retrieved
February 2, 2014, from http://www.chemcomp.com
Cleaver, J. E., Lam, E. T., & Revet, I. (2009). Disorders of nucleotide excision repair: the genetic and
molecular basis of heterogeneity. Nature Reviews Genetics, 10(11), 756–768.
https://doi.org/10.1038/nrg2663
de Laat, W. L., Sijbers, A. M., Odijk, H., Jaspers, N. G., & Hoeijmakers, J. H. (1998). Mapping of
interaction domains between human repair proteins ERCC1 and XPF. Nucleic Acids Research,
26(18), 4146–4152. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9722633
Elmenoufy, A. H., Gentile, F., Jay, D., Karimi-Busheri, F., Yang, X., Soueidan, O. M., … West, F. G.
(2019). Targeting DNA Repair in Tumor Cells via Inhibition of ERCC1-XPF. Journal of
Medicinal Chemistry, 62(17), acs.jmedchem.9b00326.
https://doi.org/10.1021/acs.jmedchem.9b00326
Faridounnia, M., Folkers, G. E., & Boelens, R. (2018, December 5). Function and interactions of
ERCC1-XPF in DNA damage response. Molecules. Multidisciplinary Digital Publishing Institute.
This article is protected by copyright. All rights reserved

https://doi.org/10.3390/molecules23123205
Genheden, S., & Ryde, U. (2015). The MM/PBSA and MM/GBSA methods to estimate ligand-
binding affinities. Expert Opinion on Drug Discovery, 10(5), 449–461.
https://doi.org/10.1517/17460441.2015.1032936
Gentile, F., A. Tuszynski, J., & H. Barakat, K. (2016). Modelling DNA Repair Pathways: Recent
Advances and Future Directions. Current Pharmaceutical Design, 22(23), 3527–3546.
https://doi.org/10.2174/1381612822666160420141435
Gentile, F., Barakat, K., & Tuszynski, J. (2018). Computational Characterization of Small Molecules
Binding to the Human XPF Active Site and Virtual Screening to Identify Potential New DNA
Repair Inhibitors Targeting the ERCC1-XPF Endonuclease. International Journal of Molecular
Sciences, 19(5), 1328. https://doi.org/10.3390/ijms19051328
Gentile, F., Tuszynski, J. A., & Barakat, K. H. (2016). New design of nucleotide excision repair (NER)
inhibitors for combination cancer therapy. Journal of Molecular Graphics and Modelling, 65(1),
71–82. https://doi.org/10.1016/j.jmgm.2016.02.010
Groom, C. R., Bruno, I. J., Lightfoot, M. P., & Ward, S. C. (2016). The Cambridge structural database.
Acta Crystallographica Section B: Structural Science, Crystal Engineering and Materials, 72(2),
171–179. https://doi.org/10.1107/S2052520616003954
Hanahan, D., & Weinberg, R. A. (2011, March 4). Hallmarks of cancer: The next generation. Cell.
Elsevier. https://doi.org/10.1016/j.cell.2011.02.013
Helleday, T., Petermann, E., Lundin, C., Hodgson, B., & Sharma, R. A. (2008). DNA repair pathways
as targets for cancer therapy. Nature Reviews. Cancer, 8(3), 193–204.
https://doi.org/10.1038/nrc2342
Jakalian, A., Jack, D. B., & Bayly, C. I. (2002). Fast, efficient generation of high-quality atomic
charges. AM1-BCC model: II. Parameterization and validation. Journal of Computational
Chemistry, 23(16), 1623–1641. https://doi.org/10.1002/jcc.10128
This article is protected by copyright. All rights reserved

Jordheim, L. P., Barakat, K. H., Heinrich-Balard, L., Matera, E.-L., Cros-Perrial, E., Bouledrak,
K., … Dumontet, C. (2013). Small molecule inhibitors of ERCC1-XPF protein-protein
interaction synergize alkylating agents in cancer cells. Molecular Pharmacology, 84(1), 12–24.
https://doi.org/10.1124/mol.112.082347
Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., & Klein, M. L. (1983).
Comparison of simple potential functions for simulating liquid water. The Journal of Chemical
Physics, 79(2), 926. https://doi.org/10.1063/1.445869
Kabsch, W., & Sander, C. (1983). Dictionary of protein secondary structure: Pattern recognition of
hydrogen‐bonded and geometrical features. Biopolymers, 22(12), 2577–2637.
https://doi.org/10.1002/bip.360221211
Kassem, S., Ahmed, M., El-Sheikh, S., & Barakat, K. H. (2015). Entropy in bimolecular simulations:
A comprehensive review of atomic fluctuations-based methods. Journal of Molecular Graphics
& Modelling, 62, 105–117. https://doi.org/10.1016/j.jmgm.2015.09.010
Labute, P. (2008). The generalized Born/volume integral implicit solvent model: estimation of the
free energy of hydration using London dispersion instead of atomic surface area. Journal of
Computational Chemistry, 29(10), 1693–1698. https://doi.org/10.1002/jcc.20933
Li, P., Roberts, B. P., Chakravorty, D. K., & Merz, K. M. (2013). Rational Design of Particle Mesh
Ewald Compatible Lennard-Jones Parameters for +2 Metal Cations in Explicit Solvent. Journal
of Chemical Theory and Computation, 9(6), 2733–2748. https://doi.org/10.1021/ct400146w
Li, P., Song, L. F., & Merz, K. M. (2015). Systematic parameterization of monovalent ions employing
the nonbonded model. Journal of Chemical Theory and Computation, 11(4), 1645–1657.
https://doi.org/10.1021/ct500918t
Lin, A. (2004). Overview of Pharmacophore Applications in MOE. Retrieved July 20, 2015, from
https://www.chemcomp.com/journal/ph4.htm
López-Otín, C., Blasco, M. A., Partridge, L., Serrano, M., & Kroemer, G. (2013). The Hallmarks of
This article is protected by copyright. All rights reserved

Aging. Cell, 153(6), 1194. https://doi.org/10.1016/J.CELL.2013.05.039
Maier, J. A., Martinez, C., Kasavajhala, K., Wickstrom, L., Hauser, K. E., & Simmerling, C. (2015).
ff14SB: Improving the Accuracy of Protein Side Chain and Backbone Parameters from ff99SB.
Journal of Chemical Theory and Computation, 11(8), 3696–3713.
https://doi.org/10.1021/acs.jctc.5b00255
McNeil, E. M., Astell, K. R., Ritchie, A.-M., Shave, S., Houston, D. R., Bakrania, P., … Melton, D.
W. (2015). Inhibition of the ERCC1-XPF structure-specific endonuclease to overcome cancer
chemoresistance. DNA Repair, 31, 19–28. https://doi.org/10.1016/j.dnarep.2015.04.002
McNeil, E. M., & Melton, D. W. (2012). DNA repair endonuclease ERCC1-XPF as a novel
therapeutic target to overcome chemoresistance in cancer therapy. Nucleic Acids Research,
40(20), 9990–10004. https://doi.org/10.1093/nar/gks818
Mendoza, J., Martínez, J., Hernández, C., Pérez-Montiel, D., Castro, C., Fabián-Morales, E., …
Herrera, L. A. (2013). Association between ERCC1 and XPA expression and polymorphisms
and the response to cisplatin in testicular germ cell tumours. British Journal of Cancer, 109(1),
68–75. https://doi.org/10.1038/bjc.2013.303
Miller, B. R., McGee, T. D., Swails, J. M., Homeyer, N., Gohlke, H., & Roitberg, A. E. (2012).
MMPBSA.pyꢀ: An Efficient Program for End-State Free Energy Calculations. Journal of
Chemical Theory and Computation, 8(9), 3314–3321. https://doi.org/10.1021/ct300418h
Ohmoto, A., & Yachida, S. (2017). Current status of poly(ADP-ribose) polymerase inhibitors and
future directions. OncoTargets and Therapy, 10, 5195–5208.
https://doi.org/10.2147/OTT.S139336
Onufriev, A., Bashford, D., & Case, D. A. (2004). Exploring protein native states and large-scale
conformational changes with a modified generalized born model. Proteins, 55(2), 383–394.
https://doi.org/10.1002/prot.20033
Salomon-Ferrer, R., Götz, A. W., Poole, D., Le Grand, S., & Walker, R. C. (2013). Routine
This article is protected by copyright. All rights reserved

Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 2. Explicit Solvent
Particle Mesh Ewald. Journal of Chemical Theory and Computation, 9(9), 3878–3888.
https://doi.org/10.1021/ct400314y
Schaeffer, L. (2015). The Role of Functional Groups in Drug-Receptor Interactions. In The Practice
of Medicinal Chemistry: Fourth Edition (pp. 359–378). Academic Press.
https://doi.org/10.1016/B978-0-12-417205-0.00014-6
Sijbers, A. M., van der Spek, P. J., Odijk, H., van den Berg, J., van Duin, M., Westerveld, A., …
Hoeijmakers, J. H. (1996). Mutational analysis of the human nucleotide excision repair gene
ERCC1. Nucleic Acids Research, 24(17), 3370–3380. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/8811092
Sterling, T., & Irwin, J. J. (2015). ZINC 15 - Ligand Discovery for Everyone. Journal of Chemical
Information and Modeling, 55(11), 2324–2337. https://doi.org/10.1021/acs.jcim.5b00559
Tripsianes, K., Folkers, G., Ab, E., Das, D., Odijk, H., Jaspers, N. G. J., … Boelens, R. (2005). The
structure of the human ERCC1/XPF interaction domains reveals a complementary role for the
two proteins in nucleotide excision repair. Structure, 13(12), 1849–1858.
https://doi.org/10.1016/j.str.2005.08.014
Wang, J., Wang, W., Kollmann, P., & Case, D. (2005). Antechamber, An Accessory Software
Package For Molecular Mechanical Calculation. J. Comput. Chem., 25, 1157–1174.
Wang, J., Wolf, R. M., Caldwell, J. W., Kollman, P. A., & Case, D. A. (2004). Development and
testing of a general amber force field. Journal of Computational Chemistry, 25(9), 1157–1174.
https://doi.org/10.1002/jcc.20035
Weiser, J., Shenkin, P. S., & Still, W. C. (1999). Approximate atomic surfaces from linear
combinations of pairwise overlaps (LCPO). Journal of Computational Chemistry, 20(2), 217–
230. https://doi.org/10.1002/(SICI)1096-987X(19990130)20:2<217::AID-JCC4>3.0.CO;2-A
Wildman, S. A., & Crippen, G. M. (1999). Prediction of physicochemical parameters by atomic
This article is protected by copyright. All rights reserved

contributions. Journal of Chemical Information and Computer Sciences, 39(5), 868–873.
https://doi.org/10.1021/ci990307l
Tables
Table 1 Main ligand-receptor and ligand-solvent hydrogen bonds observed during the last 100 ns of
MD simulation of the XPF-F06 complex. Only the bonds present in at least 5% of the trajectory are
reported. Names of F06 polar atoms follow the scheme of Figure 4.
Donor
F06@N2
F06@N3
F06@N3
F06@N2
F06@O1
F06@N3
F06@N2
N834
Acceptor
E831
Ratioa
0.37
0.15
0.11
0.08
0.92
0.75
0.68
0.09
0.96
0.82
0.49
0.41
0.13
Average distance (Å)
Average angle (º)
160.24
2.88
2.98
3.19
2.98
2.78
3.02
2.95
3.10
3.02
2.94
3.28
3.07
3.24
V859
159.68
M856
142.63
S830
140.18
Solventb
Solventb
Solventb
F06@N4
F06@N4
F06@N1
F06@CL1
F06@O1
F06@N3
160.81
160.20
155.90
155.05
Solventc
Solventc
Solventc
Solventc
Solventc
154.51
160.17
138.81
148.57
145.08
aRatio = (frames where the interaction was present)/(total number of frames)
bWater, Na+ and Cl-
cWater
Table 2 Computational results for the series of F06 analogues selected for chemical synthesis. The list
is in ascending order based on MM/GBSA scores.
GBVI/WSAa
(kcal/mol)
Ligand efficiencyb
(kcal/mol/ha)
MM/GBSA
(kcal/mol)
Compound
LogP
This article is protected by copyright. All rights reserved

F06
-6.70
-8.56
-7.00
-5.17
-5.21
-5.90
-0.20
-0.21
-0.19
-0.13
-0.31
-0.18
-17.78
-14.19
-12.44
-8.15
4.10
5.31
2.31
5.93
4.40
3.79
B3
B5
B4
B1
B2
-7.11
-3.79
aFrom MOE Dock
bDocking binding energy/heavy atoms (ha)
Figure legends
Figure 1 The ERCC1-XPF complex is involved in three major DNA repair pathways: a) in NER,
adduct-caused distortions of the double helix are detected and the double stranded filament is opened
and stabilized by the repair machinery; ERCC1-XPF and XPG endonucleases are then recruited at 5′
and 3′, respectively, for the cleavage and removal of the single-strand lesion. The free 3′-OH terminal
is then used as a primer for the synthesis of the missing fragment. b) ICLs seriously hinder the
replication machinery, eventually leading to formation of double-strand breaks. ERCC1-XPF is
involved in the incision either side of an ICL from one strand allowing the passage of polymerases
and the filling of the gaps. c) In DSB repair pathway the damaged DNA can be fixed either by
homologous recombination or microhomology-mediated end joining. Independently of the repair
mechanism, ERCC1-XPF is needed to remove 3′ single strand flaps formed by non-homologous
fragments at the edges of the former break, to allow ligases to seal gaps and restore the integrity of the
double helix.
Figure 2 a) RMSDs of backbone atoms of unbound (red), F06-bound (blue) and ERCC1-bound XPF
(green) from MD simulations. The initial structure was the same in all three cases, i.e., the XPF
conformation bound to ERCC1. The systems reached a RMSD plateau after ~150 ns of simulation in
explicit solvent. Binding of F06, despite the reduced size of the ligand compared to the ERCC1 HhH2
domain, partially restored the RMSD trend to the ERCC1-bound form, indicating a stabilizing effect
upon binding to the pocket of XPF, and suggesting F06 as a dimerization inhibitor able to replace
This article is protected by copyright. All rights reserved

ERCC1 as a binding partner. b) RMSD of F06 heavy atoms. The ligand kept a stable pose within the
binding pocket of XPF, with an average displacement of ~1 Å from the starting structure.
Figure 3 Structural differences between bound and unbound forms of XPF. a) ERCC1-bound (green)
and F06-bound (blue) displayed a similar tertiary structure of the domain, with a well-formed HhH2
domain, despite the difference in size and interaction interface between ERCC1 and F06. The
unbound form (red) lost the HhH2 structure and adopted a less globular conformation. b) Residues
constituting H1 (in yellow) showed an alpha helical structure in the two bound forms, while a more
disordered structure was observed in the unbound form. H1 is part of the binding site of F06. C)
DSSP analysis of H1 residues in the last 50000 frames (100 ns) of MD simulations, confirming the
prevalence of alpha helical conformations for the XPF bound forms, differently from the unbound
form.
Figure 4 Protonated 3D structure of F06. Polar atoms are labelled with their names.
Figure 5 Modification sites of F06 explored using a CADD strategy. The choice of sites was driven
by the analysis of the binding pose of the original compound F06.
Figure 6 VS-derived compounds selected for chemical synthesis, exploring a variety of modifications
of site 1 and 2 of F06.
Figure 7 Representative plot of the change of fluorescence versus time. Increase of relative
fluorescence units (RFU) is due to the release of a fluorescent 8-base fragment and indicates ERCC1-
XPF-mediated DNA cleavage. Control refers to ERCC1-XPF activity in the presence of carrier but
not inhibitor.
Figure 8 Detailed analysis of the predicted binding mode of B5 to XPF derived from the 2 ns-long
MD simulation. a) lowest potential energy XPF (white and green) - B5 (yellow) conformation
extracted from the simulation. The acridine moiety was stacked between the side chains of Y833 and
K860. Hydrogen bonds between B5 and N834 and V859 were observed. b) Ligand interaction
diagram of the lowest potential energy conformation; the lateral branch of B5 containing the
piperazine ring extension was solvent-exposed (blue clouds) thanks to its polar and charged groups
This article is protected by copyright. All rights reserved

that favourably interacted with the surrounding water molecules. c) Per-residue decomposition of the
total MM/GBSA binding energy of B5 (blue) and F06 (orange).
This article is protected by copyright. All rights reserved

cbdd_13660_f1.tif
This article is protected by copyright. All rights reserved

cbdd_13660_f2.tif
This article is protected by copyright. All rights reserved

cbdd_13660_f3.tif
This article is protected by copyright. All rights reserved

cbdd_13660_f4.tif
This article is protected by copyright. All rights reserved

cbdd_13660_f5.tif
This article is protected by copyright. All rights reserved

cbdd_13660_f6.tif
This article is protected by copyright. All rights reserved