research papers
Insights into the binding of PARP inhibitors to the
of human tankyrase-2aPrincess Margaret Cancer Center, University Health Network, Toronto, Ontario, Canada, bHauptman–Woodward Medical Research Institute, IMCA-CAT, Advanced Photon Source, Argonne National Laboratory, Argonne, Illinois USA, cDepartments of Biochemistry, Molecular Genetics, and Medical Biophysics, University of Toronto, Toronto, Ontario, Canada, dSt Michael's Hospital, Division of Rheumatology, Departments of Medicine, Immunology and Medical Biophysics, University of Toronto, Toronto, Ontario, Canada, and eDepartment of Pharmacology and Toxicology, University of Toronto, Toronto, Ontario, Canada
*Correspondence e-mail: rottapel@gmail.com, nchirgadze@gmail.com
The poly(ADP-ribose) polymerase (PARP) family represents a new class of therapeutic targets with diverse potential disease indications. PARP1 and PARP2 inhibitors have been developed for breast and ovarian tumors manifesting double-stranded DNA-repair defects, whereas tankyrase 1 and 2 (TNKS1 and TNKS2, also known as PARP5a and PARP5b, respectively) inhibitors have been developed for tumors with elevated β-catenin activity. As the clinical relevance of PARP inhibitors continues to be actively explored, there is heightened interest in the design of selective inhibitors based on the detailed structural features of how small-molecule inhibitors bind to each of the PARP family members. Here, the high-resolution crystal structures of the human TNKS2 PARP domain in complex with 16 various PARP inhibitors are reported, including the compounds BSI-201, AZD-2281 and ABT-888, which are currently in Phase 2 or 3 clinical trials. These structures provide insight into the inhibitor-binding modes for the tankyrase PARP domain and valuable information to guide the rational design of future tankyrase-specific inhibitors.
Keywords: cancer; poly(ADP-ribose) polymerase; TNKS2; structure-based drug discovery; structural biology.
PDB references: TNKS2–EB-47, 4tk5; TNKS2–DR-2313, 4pnl; TNKS2–3,4-CPQ-5-C, 4tju; TNKS2–BSI-201, 4tki; TNKS2–TIQ-A, 4pnr; TNKS2–5-AIQ, 4pnq; TNKS2–4-HQN, 4pnn; TNKS2–3-AB, 4pml; TNKS2–AZD-2281, 4tkg; TNKS2–NU-1025, 4pnm; TNKS2–PJ-34, 4tjw; TNKS2–INH2BP, 4pns; TNKS2–DPQ, 4tk0; TNKS2–ABT-888, 4tjy; TNKS2–IWR-1, 4tkf; TNKS2–1,5-IQD, 4pnt
1. Introduction
The post-translational modification of proteins by poly(ADP-ribosyl)ation is catalyzed by a group of 22 related enzymes which are members of the poly(ADP-ribosylation) polymerase (PARP) family (Schreiber et al., 2006; Gagné et al., 2006). The most extensively studied PARP family member is PARP1, which modifies a number of DNA-binding proteins with ADP-ribose chains in response to DNA damage (D'Amours et al., 1999; de Murcia et al., 1997; Wang et al., 1997). Other PARP family members are involved in diverse cellular functions including control of chromatin structure, organization of the mitotic spindle and regulation of signal transduction pathways (Schreiber et al., 2006).
PARP enzymes catalyze the transfer of an ADP-ribose moiety to aspartate, glutamate, asparagine, arginine or lysine residues of acceptor proteins (reviewed in Hottiger et al., 2010). The repeating units of ADP-ribose linked by glycosidic bonds can result in polymers that are hundreds of units long, branched and carry a highly polyanionic charge. Poly(ADP-ribose) (PAR) modification is reversible through the action of poly(ADP-ribose) glycohydrolase (PARG; Bonicalzi et al., 2005), while the final ADP-ribose moiety attached to the protein is removed by ADP-ribosyl protein lyase (Oka et al., 1984). ADP-ribosylarginine hydrolase-3 (ARH3), an enzyme unrelated to PARG, has also been shown to be capable of PAR hydrolysis (Oka et al., 2006).
PARP family members share a homologous et al., 2010). Common to all active PARP catalytic domains is a conserved signature sequence defined by a `catalytic triad' of histidine, tyrosine and glutamic acid.
typically located at the C-terminus of the protein, while the N-terminal sequences contain diverse protein–nucleotide binding or protein-interaction domains. To date, only PARP1, PARP2, PARP3, PARP4, TNKS1 and TNKS2 have been confirmed to be catalytically active (RouleauFour distinct PAR-binding motifs have been identified: (i) the PAR-binding basic/hydrophobic motif present in DNA-damage checkpoint proteins (Pleschke et al., 2000) and in heterogeneous nuclear ribonucleoproteins (Gagné et al., 2003), (ii) the PAR-binding zinc-finger domain (PBZ domain) contained in the CHFR E3 ubiquitin ligase and the DNA-damage response proteins aprataxin and PNK-like factor (APLF; Ahel et al., 2008), (iii) the mono-ADP-ribose-binding macro domain found in histone H2A (Karras et al., 2000) and (iv) the WWE domain in RNF146 that recognizes PAR by interacting with iso-ADP-ribose (iso-ADPR) within the poly(ADP-ribose) chain (Wang et al., 2012). The recognition of ADP-ribose modifications by proteins containing PAR-binding domains can mediate the assembly of multiprotein complexes.
TNKS1 and TNKS2 display a high degree of sequence identity (85% of residues identical overall, with 94% identity in the PARP catalytic domains). TNKS1 and TNKS2 share a common domain organization with a large N-terminal ankyrin domain divided into five ankyrin-repeat clusters (ARCs) involved in substrate recognition, a sterile alpha motif (SAM) domain required for dimerization, followed by the C-terminal PARP domain (Hsiao & Smith, 2008), as shown in Fig. 1. TNKS1 contains a unique histidine-, proline- and serine-rich N-terminal region (HPS domain) of unknown function that is not present in TNKS2. TNKS1 was originally identified as a binding partner of the telomerase inhibitor TRF1 and promotes telomere elongation by suppressing the protein expression of TRF1 through an ADP-ribose-dependent ubiquitin pathway (Smith et al., 1998). Tankyrase enzymes are now appreciated to poly(ADP-ribosyl)ate (PARsylate) a number of target proteins (Hsiao & Smith, 2008) which contain a common RXXPXG ARC-binding consensus sequence (Sbodio & Chi, 2002; Guettler et al., 2011). TNKS1-deficient cells manifest a cell-cycle defect (Dynek & Smith, 2004), increased sister-telomere association (Canudas et al., 2007), spindle dysfunction (Chang et al., 2005) and altered Glut4/IRAP distribution in adipocytes (Yeh et al., 2007). TNKS2 has been identified as a binding partner of Grb14 (Lyons et al., 2001). TNKS2 has also been shown to bind to TRF1 (Hsiao et al., 2006) and IRAP (Sbodio & Chi, 2002), suggesting functional redundancy between TNKS1 and TNKS2. While both TNKS1 and TNKS2 knockout mice are viable with a decreased body-weight phenotype (Hsiao et al., 2006), TNKS1/TNKS2 compound homozygote knockout mice are embryonically lethal by day 9.5, supporting genetic redundancy between the two proteins (Chiang et al., 2008). Both TNKS and TNKS2 bind to and suppress Axin2, a negative regulator of β-catenin, suggesting that they may represent novel druggable targets for cancers dependent on active β-catenin (Huang et al., 2009). Loss of TNKS2-dependent negative regulation of the adapter protein 3BP2 underlies the pathogenic mechanism of cherubism, an autosomal dominant disorder affecting cranial bone development (Levaot et al., 2011). TNKS2 negatively regulates the steady-state levels of the Src-binding adapter protein 3BP2 in macrophages and osteoclasts. Ribosylation of 3BP2 by TNKS2 creates a binding for the E3-ubiquitin ligase RNF146, which ubiquitylates 3BP2, leading to its destruction by the proteasome (Levaot et al., 2011). Mutation of the 3BP2 TNKS2 binding site in cherubism patients results in a hypermorphic mutation of 3BP2, leading to its increased expression, activation of Src and hyperactive osteoclasts.
The crystal structures of the catalytic domains of TNKS1 and TNKS2 are highly similar to one another but reveal a number of differences when compared with the et al., 2008; Karlberg, Markova et al., 2010). The nine core β-strands and four α-helices of the TNKS and the histidine, tyrosine and glutamic acid (HYE) triad are conserved in PARP1. However, the N-terminal α-helical domain of PARP1 is entirely absent in TNKS. TNKS1 and TNKS2 also have a much smaller and disordered B-loop than PARP1. This region has been linked to substrate specificity in some mono-ADP-ribosylating enzymes (Karlberg, Markova et al., 2010). TNKS1 and TNKS2 also differ from other PARP family members in that they harbor a short zinc-binding motif containing residues Thr1079–His1093 within the Three cysteine residues (Cys1081, Cys1089 and Cys1092) and His1084 coordinate a zinc ion, which is located about 20 Å away from the catalytic site. The function of this is still unknown. A second structural feature which distinguishes the TNKS from PARP1 is that the donor NAD+-binding site (D-loop) of TNKS is in a closed configuration compared with PARP1.
of PARP1 (LehtiöThe development of inhibitors directed against members of the PARP family has focused mainly on PARP1 and PARP2. Several candidate clinical leads, including BSI-201, AZD-2281 and ABT-888, have progressed to Phase 2 and 3 clinical trials for patients with BRCA mutations in breast or ovarian cancer (Sandhu et al., 2011; Domagala et al., 2011; Fogelman et al., 2011; Liang & Tan, 2010; Weil & Chen, 2011). structures of the TNKS with small-molecule ligands (Karlberg, Markova et al., 2010; Wahlberg et al., 2012; Narwal et al., 2012) have provided detailed information about the modes of binding of general PARP inhibitors in comparison to TNKS2-selective inhibitors. Here, we report high-resolution structures of the TNKS2 PARP with 16 known PARP inhibitors and provide a consensus structural model for the selectivity of TNKS inhibition distinct from that of PARP1 and PARP2.
2. Materials and methods
2.1. Inhibitor compounds
3-Aminobenzamide (3-AB), 2-(dimethylamino)-N-(5,6-dihydro-6-oxophenanthridin-2-yl)acetamide (PJ-34), 5-aminoisoquinolinone (5-AIQ) and 1-piperazineacetamide-4-[1-(6-amino-9H-purin-9-yl)-1-deoxy-D-ribofuranuron]-N-(2,3-dihydro-1H-isoindol-4-yl)-1-one (EB-47) were purchased from Sigma–Aldrich, Calbiochem or Inotek Pharmaceuticals (Beverly, Massachusetts, USA); the rest of the 12 compounds were purchased from other commercial suppliers. Fresh stock solutions of these compounds were prepared in 1% DMSO or distilled water.
2.2. Cloning
A pET-28a vector containing the sequence coding for the PARP domain (NdeI/XhoI) of TNKS2 was used as a template to generate several PCR fragments of the PARP domain. These PCR fragments were then cloned into pET-28a_LIC (GenBank accession EF442785) and p15TV-L (GenBank accession EF456736) vectors employing a ligation-independent cloning technique (Clontech Laboratories In-Fusion PCR Cloning Kit). Of the eight generated constructs (domain boundaries corresponding to Glu938–Gly1166, Gly950–Gly1162, Ser959–Val1164 and Gly939–Arg1159), one clone (PARP domain boundary Ser959–Val1164 cloned into p15TV_L vector) demonstrated the best expression of soluble protein. This clone, p15TV-L (PARP Ser959–Val1164), was subsequently denoted Tank2.4-6 and was chosen for protein purification.
2.3. Protein expression
Tank2.4-6 DNA was transformed into Escherichia coli BL-21(DE3) RIPL cells (Stratagene, La Jolla, California, USA). Cells were grown on standard Terrific Broth (Sigma–Aldrich Canada Co., Oakville, Ontario, Canada) supplemented with 100 mg l−1 ampicillin and 34 mg ml−1 chloramphenicol in 1 l Tunair flasks at 37°C to an OD600 of 3.5; the temperature was then lowered to 16°C and IPTG was added to 0.2 mM. Expression was allowed to proceed overnight. The cells were then harvested by centrifugation, flash-frozen in liquid nitrogen and stored at −80°C.
2.4. Protein purification
Cells were thawed on ice and resuspended in binding buffer [100 mM HEPES pH 7.5, 500 mM NaCl, 5% glycerol, 0.2 mM tris(2-carboxyethyl)phosphine, 0.2 mM TCEP] supplemented with 0.5% CHAPS, 0.25 mM phenylmethylsulfonylfluoride and 0.5 mM benzamidine. After disruption by sonication and centrifugation at 60 000g for 40 min, the cell-free extracts were passed through a DE-52 column (5 cm diameter × 7.5 cm) which had been pre-equilibrated with the same buffer and were then loaded by gravity flow onto a 10 ml Ni–nitrilotriacetic acid (NTA) column (Qiagen, Germantown, Maryland, USA). The column was washed with five column volumes (CV) of wash buffer (100 mM HEPES pH 7.5, 500 mM NaCl, 5% glycerol, 15 mM imidazole, 0.2 mM TCEP) supplemented with 0.5% CHAPS, followed by five volumes of wash buffer. The His6-tagged protein was eluted with the same buffer containing 250 mM imidazole. This sample was concentrated using a Vivaspin unit (Sartorius NA, Edgewood, New York. USA) and loaded onto a 2.6 cm diameter × 60 cm Superdex 200 column (GE Healthcare) equilibrated with gel-filtration buffer (10 mM HEPES pH 7.5, 500 mM NaCl, 0.2 mM TCEP). Elution was carried out at a flow rate of 3 ml min−1 at 8°C and Tank2.4-6 was eluted as an apparent monomer. This sample was concentrated to ∼1 ml, diluted tenfold with ion-exchange buffer (20 mM MES buffer pH 6.5, 5% glycerol, 0.2 mM TCEP) and subjected to cation-exchange on a 1.6 cm diameter × 10 cm Source 30S column (GE Healthcare). The column was washed with 3 CV of 50 mM NaCl in the same buffer and developed with a 20 CV linear gradient of NaCl (50–500 mM). Tank2.4-6 eluted at ∼375 mM NaCl. It was immediately concentrated to 25 mg ml−1, divided into 1.25 mg aliquots, flash-frozen and stored at −80°C.
2.5. In vitro PARP assay
Purified PARP domain of TNKS2 and either BSA or recombinant full-length 3BP2 protein were incubated in PARP reaction buffer (50 mM Tris pH 8.0, 4 mM MgCl2, 0.2 mM dithiothreitol) containing 0.5 mM NAD+ as an exogenous source of ADP-ribose for 30 min at 25°C with or without PARP inhibitors. Reactions were stopped by adding sample buffer to the tubes. Samples were boiled and separated on a 4–20% SDS–PAGE gel. The gel was stained with Coomassie Blue, dried on a gel dryer and used for autoradiography analysis (Fig. 2).
2.6. Crystallization
The TNKS2 protein sample was prepared at a concentration of 15 mg ml−1 (0.06 mM) and incubated with 0.1 mM inhibitor for 1 h. 1.0 µl of the mixture was then transferred to a hanging drop and mixed with an equal volume of reservoir solution consisting of 0.2 M NaCl, 0.1 M HEPES buffer pH 7.5, 12–15% isopropanol. The rod-shaped crystals were fully grown after one week to standard dimensions of 100 × 30 × 30 µm. In co-crystallization experiments, the crystals were mounted and transferred into a droplet that contained identical components to the actual drop on the crystallization plate plus 0.1 mM of the respective inhibitor and 10% glycerol. Using a `co-crystallization plus soaking' technique, before introducing the cryoprotectant the crystals were soaked overnight in 10 mM inhibitor. An equal amount of inhibitor (10 mM) and 10% glycerol were added to the cryoprotectant. In `inhibitor replacement' experiments, the crystals were grown in the presence of 3-AB (the crystals were easy to reproduce and 3-AB has a relatively low affinity for TNKS2 when compared with the other inhibitors) and then replaced with the inhibitor of interest. In this approach, crystals were grown at room temperature with 0.1 mM 3-AB under the conditions described above. Prior to harvesting, crystals were soaked overnight with 5–10 mM of the respective replacement inhibitor. The cryoprotectant solution included 5–10 mM of the replacement inhibitor and 10% glycerol. Cryoprotected crystals were flash-cooled in liquid nitrogen for low-temperature X-ray screening and data collection.
2.7. X-ray data collection and processing
Synchrotron X-ray data sets for TNKS2 inhibitor complexes were collected at 100 K on beamlines 17-ID and 17-BM at the Advanced Photon Source, Argonne National Laboratory. In-house data sets were collected on a Rigaku FR-E SuperBright rotating-anode generator equipped with a Rigaku Saturn A200 CCD detector (Rigaku, The Woodlands, Texas, USA). The diffraction data were reduced and scaled with XDS (Kabsch, 2010).
2.8. and crystallographic refinement
The crystals of all complexes belonged to P212121, with unit-cell parameters around a = 74, b = 79, c = 153 Å and four molecules per The first complex was determined by with MOLREP (Vagin & Teplyakov, 2010) using TNKS1 (PDB entry 2rf5; Lehtiö et al., 2008) as a search model. The rest of the complex structures were determined by the difference Fourier method. Following the initial rigid-body interactive cycles of model building and were carried out using Coot (Emsley et al., 2010) and BUSTER-TNT (Bricogne et al., 2011). The coordinates and topologies of the ligands from this study were generated using the GlycoBioChem PRODRG2 server (Schüttelkopf & van Aalten, 2004). Ligands were introduced at the last stages of after most of the protein models of TNKS2 has been built. Water molecules as well as other solvent ligands were added based on the 2mFo − DFc map in Coot and were refined with BUSTER-TNT. Using phenix.refine (Afonine et al., 2012), a simulated-annealing map based on the final model without any inhibitors and waters was generated for each complex structure as a reference to avoid model bias. Owing to the crystal packing, the inhibitor electron density had different quality for each of the four TNKS2 molecules in the among which chain D had the worst density in most complexes, while chains A, B and C had equally high-quality electron density. In order to have a direct comparison, we choose chain C in our discussion below except for the situations where the chain is specifically mentioned. In the case of the BSI-201 complex structure, an additional experimental phasing map was generated using phenix.autosol (Adams et al., 2010), which proved that there were ten iodine sites per in the complex structure and that they corresponded to the ten BSI-201 positions in the final model. The are listed in Table 1. All figures except for Figs. 1 and 2 were produced using PyMOL (https://www.pymol.org).
|
3. Results
The overall et al., 2010; Fig. 3a). There are two prominent binding pockets: the NAD+ (donor) site and the acceptor site demarcated by the side chain of Tyr1050 in the D-loop in the closed conformation of the (Fig. 3b, left). Three conserved cysteine residues (i.e. Cys1081, Cys1089 and Cys1092) and one histidine (His1084) form a short zinc-binding motif which is unique to TNKS1 and TNKS2 (Fig. 3c). Two of the inhibitory structures reported here adopt this closed configuration (TNKS2–TIQ-A and TNKS2–BSI-201). The majority of the structures of the bound to inhibitor compounds, however, show an open conformation in which the side chain of Tyr1050 is displaced away from the NAD+ site, exposing a narrow and deeply buried pocket for binding the nicotinamide moiety (NI-subsite; Figs. 3b and 3c). The residues surrounding this subsite are highly conserved across the whole PARP family. The majority of PARP inhibitors have been designed to target this NI-subsite. A second structural feature of the NAD+-binding site is a binding pocket for the adenosine moiety of NAD+ (AD-subsite). This subsite encompasses a narrow cleft, which is surface-accessible. The residues surrounding the AD-subsite and the D-loop region are highly conserved in both TNKS and TNKS2, but are distinct compared with other PARP family members. An analysis of the PARP structures deposited in the Protein Data Bank provides little information about the protein–ligand interaction at the AD-subsite. Our study now provides evidence that the exploitation of ligand interactions at the AD-subsite could improve the design of TNKS2-specific inhibitors.
of the TNKS2 is similar to the structure of the of PARP2 determined in complex with the small-molecule inhibitor ABT-888 (Karlberg, Hammarström3.1. Group I: inhibitors that only target the NI-subsite (nicotinamide)
Most of the PARP inhibitors available in the public domain are based on first-generation inhibitors targeting the NI-subsite. In this study, we present eight protein–ligand complex structures that belong to this group. The main interactions between TNKS2 and these inhibitors are (i) hydrogen bonds to the backbone carbonyl and amide group of Gly1032 and the side chain of Ser1068 and (ii) π-stacking interactions between the aromatic ring(s) of the inhibitors with Tyr1060 and Tyr1071. These two specific interactions are observed in the complexes of all PARP family members published to date. The half-maximal inhibitory concentrations (IC50) measured for these inhibitors with TNKS2 are moderate, ranging from 0.45 µM to over 30 µM (Table 2), and the values are generally consistent with the binding interactions that these inhibitors undergo. The structural details are described below.
|
3.1.1. 3-Amino-benzamide (3-AB)
3-AB is the most studied first-generation PARP inhibitor. It has been co-crystallized with several PARP proteins, including PARP2 (PDB entry 3kcz; Karlberg, Hammarström et al., 2010), PARP10 (PDB entry 3hkv; Structural Genomics Consortium, unpublished work), PARP12 (PDB entry 2pqf; Structural Genomics Consortium, unpublished work) and PARP14 (PDB entry 3goy; Wahlberg et al., 2012). We determined the structure of the TNKS2 in complex with 3-AB to 1.9 Å resolution. Similar to the previously reported structures listed above, 3-AB sits on the bottom of the active site, mimicking the binding mode of nicotinamide. It forms three conserved hydrogen bonds to the backbone carbonyl and amide of Gly1032 and the side chain of Ser1068. The benzamide ring of 3-AB (A ring) is in an approximate position to stack with Tyr1071 and Tyr1060. Nevertheless, the 3′-substituted amide group of 3-AB forms a weak hydrogen-bond interaction with the Oη atom of Tyr1071 (3.1 Å), which pulls the plane of 3-AB closer to Tyr1071 and away from Tyr1060. It should be noted that the 3′ amide of 3-AB forms a hydrogen bond to a well defined isopropanol molecule (IPA) acquired from the crystallization solution. The alcohol links 3-AB to the catalytically important residue Glu1138 (Fig. 4a).
3.1.2. 2-Methyl-3,5,7,8-tetrahydro-4H-thiopyrano[4,3-d]pyrimidin-4-one (DR-2313)
DR-2313 is a potent, water-soluble competitive PARP inhibitor. It is also the first PARP inhibitor that does not contain a benzamide et al., 2005). The IC50 value for DR-2313 from our study is about ten times better than the value for 3-AB (Table 2). In the publicly available structure of the complex of PARP3 with DR-2313 at 2.1 Å resolution (PDB entry 3c4h; Lehtiö et al., 2009), the inhibitor adopts a binding mode similar to that seen in our 1.5 Å high-resolution structure (Fig. 4b). In addition to the three conserved hydrogen bonds associated with the B ring, and the π-stacking with Tyr1071, the bulkier S atom from the A ring also displays hydrophobic interactions with Glu1138, Lys1067, Phe1061 and Tyr1060.
which had previously been thought to be essential for good binding to PARP enzymes. In DR-2313, the amide group is fused into the B ring, a modification that improved the binding potency (Nakajima3.1.3. 8-Hydroxy-2-methyl-3-hydro-quinazolin-4-one (NU-1025)
NU-1025 forms the three hydrogen bonds with Gly1032 and Ser1068 along with the π-stacking interaction with Tyr1071 seen with our other structures described above. In the 2.2 Å resolution electron-density map, a water molecule can be identified which links the hydroxyl group from the A ring to the catalytically important Glu1138. This additional hydrogen bond mediated by the water molecule may contribute to the lower IC50 of NU-1025 compared with that of DR-2313 (Table 2, Fig. 4c). The second structural water molecule hydrogen-bonded to the hydroxyl group has a very weak interaction with the protein and for this reason contributes very little to the ligand potency increase (The same is true for 4-HQN, 5-AIQ and 1,5-IQD.) These interactions between NU-1025 and TNKS2 are similar to those previously reported for the PARP1–NU-1025 complex (PDB entry 4pax; Ruf et al., 1998).
3.1.4. 4-Hydroxyquinazoline (4-HQN)
4-HQN plays a role in modulating the kinase cascades and regulating transcription factors in a rodent septic shock model (Veres et al., 2004). Based on our 1.65 Å resolution structure, 4-HGN forms three hydrogen bonds and a π-stacking interaction with Tyr1071 of TNKS2. This configuration of interactions is similar to the interaction of other first-generation PARP inhibitors with TNKS2 (Fig. 4d) and may explain the similar IC50 of 4-HQN and 3-AB towards TNKS2 (Table 2).
3.1.5. 5-Aminoisoquinolinone (5-AIQ)
5-AIQ is an isoquinolinone derivative and has been reported to have moderating effects on the organ injury and dysfunction caused by haemorrhagic shock (McDonald et al., 2000). Our 1.9 Å resolution structure identifies the same three conserved hydrogen bonds and a π-stacking interaction as described for the NI-subsite inhibitors (Fig. 4e). In concert with this observation, the IC50 of 5-AIQ towards TNKS2 was similar to those of 3-AB and 4-HQN.
3.1.6. 1,5-Isoquinolinediol (1,5-IQD)
The 1.6 Å resolution structure shows that 1,5-IQD forms a water-mediated hydrogen bond from the hydroxyl group of the A ring to Glu1138 in addition to the three conserved hydrogen bonds and the π-stacking interaction, thus interacting with TNKS2 in a manner similar to the NU-1025 TNKS2 complex. The hydrogen bond formed between 1,5-IQD and Glu1138 is likely to contribute to the potent IC50 of this inhibitor of 1.5 µM, a level similar to that of NU-1025 (Fig. 4f).
3.1.7. Thieno-[2,3-c]-isoquinolin-5-one (TIQ-A)
The larger planar surface of its tricyclic ring allows TIQ-A to form an extended π-stacking with Tyr1060 and Tyr1071 of the TNKS2 The B ring forms three hydrogen bonds to the backbone of Gly1032 and one to the side chain of Ser1068 at the bottom of the NI-subsite of TNKS2. The S atom from the C ring also forms a strong van der Waals interaction with the main-chain carbonyl group of Gly1032 (Fig. 4g). The tricyclic lactam core of TIQ-A appears to be responsible for its tighter binding to TNKS2 when compared with the two-ring inhibitors discussed above. One surprising observation in the 1.7 Å resolution TNKS2–TIQ-A complex structure is that the side chain of Tyr1050, part of the D-loop, protrudes toward TIQ-A and forms a closed conformation. Additional hydrophobic interactions between the side chain of Tyr1050 and the five-membered thiophene C-ring cause this movement of the D-loop, which brings it closer to the NI-subsite compared with the six inhibitors described above. TIQ-A exhibits the best IC50 value towards TNKS2 (0.456 µM) compared with other inhibitors from this category (Table 2).
3.1.8. 5-Iodo-6-amino-1,2-benzopyrone (INH2BP)
INH2BP was designed as a noncovalent inhibitor of PARP1 (Bauer et al., 1995); however, it displays low potency towards other PARP family members. This inhibitor binds to the NI-subsite of TNKS2 in a mode different from that of all the other PARP inhibitors investigated in this study (Fig. 4h). Owing to the relatively large radius of the I atom, the inhibitor adopts a position with its iodine end pointing towards the AD-subsite. This orientation prevents the molecule from preserving the three critical hydrogen bonds to residues located at the bottom of the NI-subsite. Instead, only two potential hydrogen bonds to TNKS2 can be observed from INH2BP: one to the main-chain amide of Gly1032 and the other to the side chain of Ser1068 (Fig. 4h). The amino group on the opposite end of the bicyclic ring system forms three water-mediated hydrogen bonds to Tyr1071, Glu1138 and Gly1053 of TNKS2. It should be noted that the discontinuous 2mFo − DFc density (contoured at 1σ, black) around the iodine group could be caused by Fourier series truncation ripples, since iodine has a relatively large number of electrons and these ripples compete with normal positive density from the nearby C atom and cancel it out. This ripple effect can be observed as strong negative density surrounding the I atom (Fig. 4h; mFo − DFc map contoured at −3σ, colored red).
Inhibitory activities measured for the compounds in group I (Table 2) correlate well with the number of hydrogen bonds that the inhibitor molecules form to the NI-subsite of TNKS2. For example, TIQ-A is able to form four hydrogen bonds to TNKS2 and has an IC50 value of 456 nM, which is about three times lower than that for NU-1025 (IC50 = 1.4 µM), which forms three direct hydrogen bonds. In distinction, INH2BP forms two direct hydrogen bonds to the NI-subsite and has the least potent IC50 (>30 µM).
3.2. Group II: inhibitors that reach outside the NI-subsite but do not enter the AD-subsite
To develop more selective and potent inhibitors for individual PARP1 and PARP2, a series of compounds have been synthesized with substitutions designed to extend towards, but not reach, the AD-site and interact with the N-terminal helices within the
Here, we report the crystal structures of TNKS2 in complex with four such inhibitors.3.2.1. N-(6-Oxo-5,6-dihydro-phenanthridin-2-yl)-N,N-dimethylacetamide (PJ-34)
The tricyclic lactam core of PJ-34 creates three conserved hydrogen bonds and an extended π-sandwich stacking with Tyr1060 and Tyr1071 within the NI-subsite of the TNKS2 in a manner similar to TIQ-A. In distinction to the closed conformation observed in the TIQ-A structure, however, the tertiary amine extension protruding from the three-ring core of PJ-34 displaces the D-loop away from the AD-subsite and the side chain of Tyr1050 adopts an open conformation (Fig. 5a). The amine extension forms two weak water-mediated hydrogen bonds to the backbone of Tyr1050 and Gly1058. The twofold lower IC50 of PJ-34 (963 nM) compared with TIQ-A is likely to be a result of these distinct interactions with the TNKS2 catalytic domain
3.2.2. 2-[(R)-2-Methylpyrolidin-2-yl]-1H-benzimidazole-4-carboxamide (ABT-888, Veliparib)
A 3kjd; Karlberg, Hammarström et al., 2010). In our structure with TNKS2, ABT-888 adopts a similar orientation to that found in the PARP2–ABT-888 complex (Fig. 5b). At the bottom of the NI-subsite, the carboxamide group forms three hydrogen bonds to the backbone of Gly1032 and the side-chain hydroxyl of Ser1068. The N3 atom of the benzimidazole undergoes water-mediated hydrogen bonding to Glu1138. Fig. 5(b) presents the superposition of ABT-888 complexed to PARP2 and TNKS2; one noticeable difference is the 10° rotation of the pyrrolidine ring of ABT-888 towards Glu1138 in the TNKS2 structure. When complexed with PARP2, the N2 atom of the ABT-888 pyrrolidine forms a water-mediated interaction with a glutamate from the N-terminal helix-bundle domain of the enzyme. Since this helical structure is absent in TNKS2, the core plane of the ABT-888 scaffold moves towards Glu1138 within the TNKS2 catalytic domain.
of ABT-888 with PARP2 is available in the PDB (PDB entry3.2.3. 3-(4-Chlorophenyl)-quinoxaline-5-carboxamide (3,4-CPQ-5C)
As deduced from our 1.57 Å resolution structure, the binding mode of TNKS2 to 3,4-CPQ-5C at the NI-subsite is similar to what has been described for the NU-1025 and 1,5-IQD complexes. The three common hydrogen bonds and the π-stacking interaction lock the carboxamide moiety of the inhibitor tightly into the NI-subsite. Instead of a hydroxyl group, the N9 atom of the quinoxaline ring forms a water-mediated hydrogen bond to Glu1138. The structure of this inhibitor has also been analyzed in complex with PARP1 (PDB entry 1wok; Iwashita et al., 2005). It was suggested that the terminal phenyl group of this ligand could provide between PARP1 and PARP2. Since TNKS2 lacks the N-terminal helix-bundle domain, 3,4-CPQ-5C adopts a distinct binding mode with the TNKS2 such that the chlorophenyl group rests in a large pocket adjacent to the NI-subsite and undergoes some hydrophobic interactions with the side chain of Ile1075 (Fig. 5c).
3.2.4. 3,4-Dihydro-5-[4-(1-piperidinyl)buthoxyl)]-1(2H)-isoquinolinone (DPQ)
From our 1.8 Å resolution structure of the TNKS2–DPQ complex, we find that the DPQ molecule binds poorly to the protein, with only one of the four active sites (chain C) represented in the displaying electron density sufficient to build in a DPQ model. The isoquinolinone base of DPQ contributes to most of the interactions between the inhibitor and TNKS2, with three conserved hydrogen bonds with Gly1032 and Ser1068 as well as the π-stacking with Tyr1060 and Tyr1071 (Fig. 5d). As had been found in the 3,4-CPQ-5C complex structure noted above, the extension of the isoquinolinone core does not interact strongly with TNKS2 owing to the absence of the N-terminal helix-bundle domain in the TNKS2 This feature provides an explanation of why this group of inhibitors in general does not exhibit better selectivity and affinity for TNKS2 compared with PARP1 (as shown in Table 2).
3.3. Group III: inhibitors targeting the AD-subsite (adenosine moiety of NAD+)
3.3.1. 1-Piperazineacetamide-4-[1-(6-amino-9H-purin-9-yl)-1-deoxy-D-ribofuranuron]-N-(2,3-dihydro-1H-isoindol-4-yl)-1-one (EB-47)
EB-47 is an inhibitor that targets not only the NI-subsite but also the AD-subsite within the TNKS2 50 of 32 nM. In the EB-47 occupies the entire NAD-binding pocket, making it an excellent mimic of the NAD-binding mode. The isoindolinone core engages in the well known hydrogen bonds and π-stacking interactions with Tyr1060 and Tyr1071. At the other end, the adenosine moiety forms four hydrogen bonds to surrounding protein residues, with that between the 2′-hydroxyl of the ribose and the catalytically important His1031 seeming to be particularly strong (2.8 Å). In addition, a network of about ten water-mediated hydrogen bonds further locks the compound into the NAD+ donor site (Fig. 6a).
The piperazine and succinyl linkers connect the adenosine and isoindolinone cores, making EB-47 one of the most potent TNKS2 inhibitors, with an IC3.3.2. 4-[(3aR,4S,7R,7aS)-1,3,3a,4,7,7a-hexahydro-1,3-dioxo-4,7-methano-2H-isoindol-2-yl]-N-8-quinolinyl-benzamide (IWR-1)
This complex structure was obtained by applying the ligand-replacement technique (for details, see §2). In two of the four molecules in the the NI-subsite still contained a 3-AB molecule which had not been competed out of its binding pocket. As shown in Fig. 6(b), IWR-1 occupies only the AD-subsite not the NI-subsite, owing in part to hydrophobic interactions. A sandwich-like π-stacking interaction with Phe1035 and His1048 holds the adenine-mimicking quinoline ring in position. The substituted isoindole ring binds to the hydrophobic pocket surrounded by the side chains of Ile1075, Tyr1071 and Tyr1060 in the context of 3-AB bound to the NI-subsite. The O atoms of the benzamide group of IWR-1 and of one of the carbonyl substituents of the isoindole form hydrogen bonds to the backbone of Tyr1071 and Asp1045, respectively, contributing to the strong binding of this inhibitor. This structure is similar to that recently reported by Narwal et al. (2012) (PDB entry 3ua9), although their complex structure has different crystal packing in C2221.
3.3.3. 4-[3-(4-Cyclopropanecarbonylpiperazine-1-carbonyl)-4-fluorobenzyl]-2H-phthalazin-1-one (AZD-2281, Olaparib)
As shown in Fig. 6(c), the bicyclic ring of AZD-2281 forms the base of the inhibitor that locks into the NI-subsite by forming the three critical hydrogen bonds and the π-sandwich stacking interaction with TNKS2 residues. The central fluorobenzyl ring displaces the D-loop by forming two hydrogen bonds to backbone atoms of Ile1051 and Gly1058 within the TNKS2 The carbonyl linking the fluorobenzyl ring to the piperazine hydrogen-bonds to the backbone N atom of Tyr1060, while the ketone O atom between piperazine and the cyclopropyl group interacts with the backbone of Asp1045; it also forms a water-mediated hydrogen bond to the backbone amide of Gly1043. The cyclopropyl ring then fits well into the AD-subsite between the aromatic rings of Phe1035 and His1048. AZD-2281 and EB-47 interact with the NI-subsite and the AD-subsite in a similar manner. AZD-2281 may have a more favorable capacity to cross the blood–brain barrier than EB-47 given its lower molecular weight (434.5 versus 610.5 Da).
3.3.4. 4-Iodo-3-nitrobenzamide (BSI-201, Iniparib)
BSI-201 is a noncompetitive PARP1 inhibitor that interacts with the PARP1 zinc-binding site (Buki et al., 1991; Patel & Kaufmann, 2010). Mutagenesis studies have confirmed that this benzamide derivative targets and covalently modifies the Arg34 residue in the zinc-finger motif of PARP1 (Melisi et al., 2009; Ossovskaya & Sherman, 2009). We included BSI-201 in our analysis in anticipation of obtaining structural information on the TNKS2 PARP domain covalently modified by the BSI-201 adduct. To our surprise, two BSI-201 molecules (BSI-201a and BSI-201b) bound to the NAD+ pocket in two different configurations, with one molecule bound within the NI-subsite and the other in the AD-subsite. The unusual presence of these two BSI-201 molecules was further confirmed by the strong anomalous signal of the I atom of BSI-201 (Fig. 6d).
BSI-201a binds within the NI-subsite. Unlike other benzamide inhibitors, the amide group of BSI-201a does not initiate the standard hydrogen bonds that are the basis for the potency of TNKS2 inhibitors. Instead, the nitro group and the iodine linked to the 4-position of the aromatic ring face the center of the protein. The nitro group forms three hydrogen bonds, two with Ser1068 and one with Gly1032, mimicking the interaction pattern of the crucial amide group of other inhibitor molecules. The side chains of Lys1067 and Glu1138 adjust themselves to accommodate the interaction with the I atom. The amide group situated on the opposing side of the aromatic ring forms a hydrogen bond to the main-chain carbonyl of Gly1032. This amide group also forms three water-mediated hydrogen bonds to the main-chain atoms of Tyr1071, the main-chain atoms of Tyr1060 and the side-chain hydroxyl of Ser1033. The side chain of Tyr1050 swings towards BSI-201a to cover the NI-subsite, reminiscent of what was observed in the TIQ-A complex (Figs. 3c and 4g). Tyr1050 also contributes to the hydrophobic environment for BSI-201a binding. The major part of the D-loop moves about 2 Å towards the NAD+ donor site when compared with the structures of other complexes (Fig. 3c and 6a). This hydrogen-bonding network, together with hydrophobic interactions, holds the inhibitor tightly in the NI-subsite.
A second BSI-201 molecule (BSI-201b) is located in the AD-subsite. The majority of its binding energy comes from π-stacking with Phe1035. In addition, BSI-201b engages in one water-mediated hydrogen bond from its nitro group to the main chain of Asp1045. In addition to the BSI-201 molecules found in the substrate-binding site, two further BSI-201 molecules are bound to the allosteric site near residue Trp1006 in molecules A and D. Owing to the crystal packing, this allosteric site in molecules B and C is occupied by neighboring molecules. The aromatic ring of BSI-201 forms good π-stacking with the side chain of Trp1006.
4. Discussion
In recent years, the members of the PARP family have been the targets of intensive drug-development efforts. To date, more than 40 entries for PARP family members complexed with small-molecule ligands are available in the Protein Data Bank (PDB). Most of these structures were determined in complex with the first-generation PARP inhibitors.
In this report, we have determined 16 novel crystal structures of TNKS2 catalytic domain–inhibitor complexes and highlight several principles of TNKS2 PARP inhibition. The binding modes of the 16 inhibitors have been subdivided into three distinct groups. The first group includes inhibitors that only target the NI-subsite, the second group consists of inhibitors that interact with TNKS2 residues lying outside the NI-subsite but do not contact the AD-subsite, and the third group is represented by inhibitors that target only the AD-subsite. We found that inhibitors that bind to the AD-subsite, such as BSI-201, AZD-2281, IWR-1 and EB-47, dramatically improve the inhibitor potency and are antagonized by movement of the D-loop from the `closed' configuration. An inhibitor that is able to stabilize the D-loop in the closed conformation would be likely to contribute favorably to the energy of binding to TNKS2. We have also determined the high-resolution crystal structures of TIQ-A and BIS-201 complexes, which represent examples of inhibitors that bind to the closed loop conformation. Smaller ligands that interact with the NI-subsite tend to bind to different PARP family members with poor selectivity. Larger ligands that bind to both the NI-subsite and the AD-subsite are good candidates for scaffolds that may demonstrate improved selectivity as TNKS2-specific inhibitors. Although the NAD+-binding pockets of tankyrases and other PARP family members are highly conserved, this site may still be exploited to design tankyrase-specific inhibitors. For example, because TNKS and TNKS2 do not have the N-terminal helix-bundle domain located near the AD-subsite of the NAD+ pocket in PARP1, which interferes with binding of the dinucleotide, potent inhibitors such as AZD-2281, whose small cyclopropyl group extends into the AD-subsite, demonstrate greater activity against PARP1 than TNKS2. Another approach to target TNKS2 could be to design TNKS2-specific inhibitors based on the AD-subsite structure. Some inhibitors may demonstrate allosteric cooperativity between AD-subsite and NI-subsite binding. For example, IWR-1 may show greater binding and enhanced in the presence of an NI-subsite binder. Lastly, we have determined the of the complex of a PARP1 inhibitor that binds to TNKS2 in a completely unique mode. BSI-201 is a potent PARP1 inhibitor that covalently binds and inhibits PARP-1 (Ossovskaya & Sherman, 2009; Buki et al., 1991; Melisi et al., 2009). We observed that BSI-201 bound to TNKS2 in a stoichiometric ratio of 2:1 with the NI-subsites and AD-subsites each bound to one BSI-201 molecule. This mode of binding suggests a new inhibitory mode of noncovalent inhibition of BSI-201 directed towards the TNKS2 catalytic domain.
5. Conclusion
We believe that the high-resolution structural information that we have obtained and systematically analyzed in the context of inhibitor-binding activity experiments will serve as a strong foundation for future tankyrase-specific structure-based drug-discovery programs.
Supporting information
PDB references: TNKS2–EB-47, 4tk5; TNKS2–DR-2313, 4pnl; TNKS2–3,4-CPQ-5-C, 4tju; TNKS2–BSI-201, 4tki; TNKS2–TIQ-A, 4pnr; TNKS2–5-AIQ, 4pnq; TNKS2–4-HQN, 4pnn; TNKS2–3-AB, 4pml; TNKS2–AZD-2281, 4tkg; TNKS2–NU-1025, 4pnm; TNKS2–PJ-34, 4tjw; TNKS2–INH2BP, 4pns; TNKS2–DPQ, 4tk0; TNKS2–ABT-888, 4tjy; TNKS2–IWR-1, 4tkf; TNKS2–1,5-IQD, 4pnt
Acknowledgements
These studies were supported by the Ontario Research and Development Challenge Fund (99-SEP-0512) and the Canada Research Chair Program (EFP). This work was supported in part from grants from CIHR and the Selective Therapy Program funded jointly by the Terry Fox Research Institute and the Ontario Institute for Cancer Research (RR). The use of the IMCA-CAT beamline 17-ID at the Advanced Photon Source was supported by the companies of the Industrial Macromolecular Crystallography Association through a contract with the Hauptman-Woodward Medical Research Institute. Use of the Advanced Photon Source was supported by the US Department of Energy, Office of Science, Office of Basic Energy Sciences under Contract No. DE-AC02-06CH11357. We also appreciate the help of Aiping Dong in providing technical support and the Structural Genomics Consortium, University of Toronto for the use of their X-ray facilities. We would like to thank Dr A. Scotter for his help in preparing this manuscript. This research was funded in part by the Ontario Ministry of Health and Long Term Care (OMOHLTC). The views expressed do not necessarily reflect those of the OMOHLTC.
References
Adams, P. D. et al. (2010). Acta Cryst. D66, 213–221. Web of Science CrossRef CAS IUCr Journals Google Scholar
Afonine, P. V. et al. (2012). Acta Cryst. D68, 352–367. Web of Science CrossRef CAS IUCr Journals Google Scholar
Ahel, I., Ahel, D., Matsusaka, T., Clark, A. J., Pines, J., Boulton, S. J. & West, S. C. (2008). Nature (London), 451, 81–85. Web of Science CrossRef PubMed CAS Google Scholar
Bauer, P. I., Kirsten, E., Varadi, G., Young, L. J., Hakam, A., Comstock, J. A. & Kun, E. (1995). Biochimie, 77, 374–377. CrossRef CAS PubMed Web of Science Google Scholar
Bonicalzi, M.-E., Haince, J.-F., Droit, A. & Poirier, G. G. (2005). Cell. Mol. Life Sci. 62, 739–750. Web of Science CrossRef PubMed CAS Google Scholar
Bricogne, G., Blanc, E., Brandl, M., Flensburg, C., Keller, P., Paciorek, W., Roversi, P., Sharff, A., Smart, O. S., Vonrhein, C. & Womack, T. O. (2011). BUSTER v.2.10.0. Cambridge: Global Phasing Ltd. Google Scholar
Buki, K. G., Bauer, P. I., Mendeleyev, J., Hakam, A. & Kun, E. (1991). FEBS Lett. 290, 181–185. CrossRef PubMed CAS Web of Science Google Scholar
Canudas, S., Houghtaling, B. R., Kim, J. Y., Dynek, J. N., Chang, W. G. & Smith, S. (2007). EMBO J. 26, 4867–4878. Web of Science CrossRef PubMed CAS Google Scholar
Chang, P., Coughlin, M. & Mitchison, T. J. (2005). Nature Cell Biol. 7, 1133–1139. Web of Science CrossRef PubMed CAS Google Scholar
Chiang, Y. J., Hsiao, S. J., Yver, D., Cushman, S. W., Tessarollo, L., Smith, S. & Hodes, R. J. (2008). PLoS One, 3, e2639. Web of Science CrossRef PubMed Google Scholar
D'Amours, D., Desnoyers, S., D'Silva, I. & Poirier, G. (1999). Biochem. J. 342, 249–268. Web of Science PubMed CAS Google Scholar
Domagala, P., Lubinski, J. & Domagala, W. (2011). N. Engl. J. Med. 364, 1780–1781. Web of Science CAS PubMed Google Scholar
Dynek, J. N. & Smith, S. (2004). Science, 304, 97–100. Web of Science CrossRef PubMed CAS Google Scholar
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486–501. Web of Science CrossRef CAS IUCr Journals Google Scholar
Fogelman, D. R., Wolff, R. A., Kopetz, S., Javle, M., Bradley, C., Mok, I., Cabanillas, F. & Abbruzzese, J. L. (2011). Anticancer Res. 31, 1417–1420. Web of Science PubMed Google Scholar
Gagné, J.-P., Hendzel, M. J., Droit, A. & Poirier, G. G. (2006). Curr. Opin. Cell Biol. 18, 145–151. Web of Science PubMed Google Scholar
Gagné, J.-P., Hunter, J. M., Labrecque, B., Chabot, B. & Poirier, G. G. (2003). Biochem. J. 371, 331–340. Web of Science PubMed Google Scholar
Guettler, S., LaRose, J., Petsalaki, E., Gish, G., Scotter, A., Pawson, T., Rottapel, R. & Sicheri, F. (2011). Cell, 147, 1340–1354. Web of Science CrossRef CAS PubMed Google Scholar
Hottiger, M. O., Hassa, P. O., Lüscher, B., Schüler, H. & Koch-Nolte, F. (2010). Trends Biochem. Sci. 35, 208–219. Web of Science CrossRef CAS PubMed Google Scholar
Hsiao, S. J., Poitras, M. F., Cook, B. D., Liu, Y. & Smith, S. (2006). Mol. Cell. Biol. 26, 2044–2054. Web of Science CrossRef PubMed CAS Google Scholar
Hsiao, S. J. & Smith, S. (2008). Biochimie, 90, 83–92. Web of Science CrossRef PubMed CAS Google Scholar
Huang, S.-M. A. et al. (2009). Nature (London), 461, 614–620. Web of Science CrossRef PubMed CAS Google Scholar
Iwashita, A., Hattori, K., Yamamoto, H., Ishida, J., Kido, Y., Kamijo, K., Murano, K., Miyake, H., Kinoshita, T., Warizaya, M., Ohkubo, M., Matsuoka, N. & Mutoh, S. (2005). FEBS Lett. 579, 1389–1393. Web of Science CrossRef PubMed CAS Google Scholar
Kabsch, W. (2010). Acta Cryst. D66, 125–132. Web of Science CrossRef CAS IUCr Journals Google Scholar
Karlberg, T., Hammarström, M., Schütz, P., Svensson, L. & Schüler, H. (2010). Biochemistry, 49, 1056–1058. Web of Science CrossRef CAS PubMed Google Scholar
Karlberg, T., Markova, N., Johansson, I., Hammarström, M., Schütz, P., Weigelt, J. & Schüler, H. (2010). J. Med. Chem. 53, 5352–5355. Web of Science CrossRef CAS PubMed Google Scholar
Karras, G. I., Kustatscher, G., Buhecha, H. R., Allen, M. D., Pugieux, C., Sait, F., Bycroft, M. & Ladurner, A. G. (2000). EMBO J. 24, 1911–1920. Web of Science CrossRef Google Scholar
Lehtiö, L., Collins, R., van den Berg, S., Johansson, A., Dahlgren, L. G., Hammarström, M., Helleday, T., Holmberg-Schiavone, L., Karlberg, T. & Weigelt, J. (2008). J. Mol. Biol. 379, 136–145. Web of Science PubMed Google Scholar
Lehtiö, L., Jemth, A. S., Collins, R., Loseva, O., Johansson, A., Markova, N., Hammarström, M., Flores, A., Holmberg-Schiavone, L., Weigelt, J., Helleday, T., Schüler, H. & Karlberg, T. (2009). J. Med. Chem. 52, 3108–3111. Web of Science PubMed Google Scholar
Levaot, N., Voytyuk, O., Dimitriou, I., Sircoulomb, F., Chandrakumar, A., Deckert, M., Krzyzanowski, P. M., Scotter, A., Gu, S., Janmohamed, S., Cong, F., Simoncic, P. D., Ueki, Y., La Rose, J. & Rottapel, R. (2011). Cell, 147, 1324–1339. Web of Science CrossRef CAS PubMed Google Scholar
Liang, H. & Tan, A. R. (2010). IDrugs, 13, 646–656. Web of Science CAS PubMed Google Scholar
Lyons, R. J., Deane, R., Lynch, D. K., Ye, Z. S., Sanderson, G. M., Eyre, H. J., Sutherland, G. R. & Daly, R. J. (2001). J. Biol. Chem. 276, 17172–17180. Web of Science CrossRef PubMed CAS Google Scholar
McDonald, M. C., Mota-Filipe, H., Wright, J. A., Abdelrahman, M., Threadgill, M. D., Thompson, A. S. & Thiemermann, C. (2000). Br. J. Pharmacol. 130, 843–850. Web of Science CrossRef PubMed CAS Google Scholar
Melisi, D., Ossovskaya, V., Zhu, C., Rosa, R., Ling, J., Dougherty, P. M., Sherman, B. M., Abbruzzese, J. L. & Chiao, P. J. (2009). Clin Cancer Res. 15, 6367–6377. Web of Science CrossRef PubMed CAS Google Scholar
Murcia, J. de, Niedergang, C., Trucco, C., Ricoul, M., Dutrillaux, B., Mark, M., Oliver, F. J., Masson, M., Dierich, A., LeMeur, M., Walztinger, C., Chambon, P. & de Murcia, G. (1997). Proc. Natl Acad. Sci. USA, 94, 7303–7307. PubMed Web of Science Google Scholar
Nakajima, H., Kakui, N., Ohkuma, K., Ishikawa, M. & Hasegawa, T. (2005). J. Pharmacol. Exp. Ther. 312, 472–481. Web of Science CrossRef PubMed CAS Google Scholar
Narwal, M., Venkannagari, H. & Lehtiö, L. (2012). J. Med. Chem. 55, 1360–1367. Web of Science CrossRef CAS PubMed Google Scholar
Oka, S., Kato, J. & Moss, J. (2006). J. Biol. Chem. 281, 705–713. Web of Science CrossRef PubMed CAS Google Scholar
Oka, J., Ueda, K., Hayaishi, O., Komura, H. & Nakanishi, K. (1984). J. Biol. Chem. 259, 986–995. CAS PubMed Web of Science Google Scholar
Ossovskaya, V. & Sherman, B. M. (2009). US Patent 20090149417 A1. Google Scholar
Patel, A. & Kaufmann, S. H. (2010). Oncology (Williston Park), 24, 66–68. Web of Science PubMed Google Scholar
Pleschke, J. M., Kleczkowska, H. E., Strohm, M. & Althaus, F. R. (2000). J. Biol. Chem. 275, 40974–40980. Web of Science CrossRef PubMed CAS Google Scholar
Rouleau, M., Patel, A., Hendzel, M. J., Kaufmann, S. H. & Poirier, G. G. (2010). Nature Rev. Cancer, 10, 293–301. Web of Science CrossRef CAS Google Scholar
Ruf, A., de Murcia, G. & Schulz, G. E. (1998). Biochemistry, 37, 3893–3900. Web of Science CrossRef CAS PubMed Google Scholar
Sandhu, S. K., Yap, T. A. & de Bono, J. S. (2011). Curr. Drug Targets, 12, 2034–2044. Web of Science CAS PubMed Google Scholar
Sbodio, J. I. & Chi, N.-W. (2002). J. Biol. Chem. 277, 31887–31892. Web of Science CrossRef PubMed CAS Google Scholar
Schreiber, V., Dantzer, F., Ame, J. C. & de Murcia, G. (2006). Nature Rev. Mol. Cell Biol. 7, 517–528. Web of Science CrossRef CAS Google Scholar
Schüttelkopf, A. W. & van Aalten, D. M. F. (2004). Acta Cryst. D60, 1355–1363. Web of Science CrossRef IUCr Journals Google Scholar
Smith, S., Giriat, I., Schmitt, A. & de Lange, T. (1998). Science, 282, 1484–1487. Web of Science CrossRef CAS PubMed Google Scholar
Vagin, A. & Teplyakov, A. (2010). Acta Cryst. D66, 22–25. Web of Science CrossRef CAS IUCr Journals Google Scholar
Veres, B., Radnai, B., Gallyas, F., Varbiro, G., Berente, Z., Osz, E. & Sumegi, B. (2004). J. Pharmacol. Exp. Ther. 310, 247–255. Web of Science CrossRef PubMed CAS Google Scholar
Wahlberg, E., Karlberg, T., Kouznetsova, E., Markova, N., Macchiarulo, A., Thorsell, A. G., Pol, E., Frostell, Å, Ekblad, T., Öncü, D., Kull, B., Robertson, G. M., Pellicciari, R., Schüler, H. & Weigelt, J. (2012). Nature Biotechnol. 30, 283–288. Web of Science CrossRef CAS Google Scholar
Wang, Z., Michaud, G. A., Cheng, Z., Zhang, Y., Hinds, T. R., Fan, E., Cong, F. & Xu, W. (2012). Genes Dev. 26, 235–240. Web of Science CrossRef PubMed Google Scholar
Wang, Z. Q., Stingl, L., Morrison, C., Jantsch, M., Los, M., Schulze-Osthoff, K. & Wagner, E. F. (1997). Genes Dev. 11, 347–2358. Google Scholar
Weil, M. K. & Chen, A. P. (2011). Curr. Probl. Cancer, 35, 7–50. Web of Science CrossRef PubMed Google Scholar
Yeh, T.-Y. J., Sbodio, J. I., Tsun, Z.-Y., Luo, B. & Chi, N.-W. (2007). Biochem. J. 402, 279–290. Web of Science PubMed CAS Google Scholar
This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.