|Home | About | Journals | Submit | Contact Us | Français|
The Bloom’s syndrome protein, BLM, is a member of the conserved RecQ helicase family. Although cell lines lacking BLM exist, these exhibit progressive genomic instability that makes distinguishing primary from secondary effects of BLM loss problematic. In order to be able to acutely disable BLM function in cells, we undertook a high throughput screen of a chemical compound library for small molecule inhibitors of BLM. We present ML216, a potent inhibitor of the DNA unwinding activity of BLM. ML216 shows cell-based activity, and can induce sister chromatid exchanges, enhance to the toxicity of aphidicolin and exert anti-proliferative activity in cells expressing BLM, but not in those lacking BLM. These data indicate that ML216 shows strong selectively for BLM in cultured cells. We discuss the potential utility of such a BLM-targeting compound as an anticancer agent.
DNA helicases are ATP-dependent enzymes that separate the complementary strands of duplex DNA (Soultanas and Wigley, 2001). Several families of helicases have been identified, including the RecQ helicases, which are conserved between bacteria, yeasts and all higher eukaryotes (Bachrati and Hickson, 2003). In human cells, there are five RecQ family members, designated RECQ1, RECQ4, RECQ5, BLM and WRN, all of which contain a centrally located domain with seven signature motifs found in the so-called Superfamily II helicases (Chu and Hickson, 2009). These motifs include the Walker A and B box sequences required for the binding and hydrolysis of ATP that drives DNA strand separation. Mutation of any of the three genes encoding human RecQ proteins (BLM, WRN and RECQ4) gives rise to a defined genetic disorder associated with premature ageing and/or cancer predisposition (Chu and Hickson, 2009). Cells from affected individuals in all cases show excessive chromosomal instability. One of these conditions, Bloom’s syndrome (BS) is caused by mutation of the BLM gene, which is located on chromosome 15q26.1 in humans (Ellis, et al., 1995).
BS is characterized by predisposition to the development of cancer, growth retardation, immunodeficiency, sunlight sensitivity, and fertility defects. Leukemia and lymphoma are the most commonly observed cancer types, but the full spectrum of cancers in the normal population is evident in BS (German, 1997). Although the cells from many BS individuals lack expression of the BLM protein, it is clear that BLM function is essential for viability in mice, with homozygous deletion of the gene leading to embryonic lethality by day 13.5 (Chester, et al., 1998). Cells from BS patients exhibit extensive chromosome instability. This is manifested as increased numbers of chromatid gaps and breaks, as well as chromosome structural rearrangements, including symmetrical quadriradials, micronuclei, anaphase bridges, and lagging chromosomes (Chu and Hickson, 2009).The characteristic cellular feature, used in the molecular diagnosis of BS, is an increase in the frequency of sister chromatid exchanges (SCEs) (Chaganti, et al., 1974).
BLM protein displays an ATP-dependent, 3’-5’ DNA helicase activity that can unwind a variety of DNA substrates that can arise during DNA replication and repair. These include various types of partial duplex DNA substrates, which are unwound in a DNA structure-specific manner (Mohaghegh et al., 2001). BLM also exhibits an activity that can branch migrate 3- and 4-way DNA junctions that model a DNA displacement loop (D-loop) and a Holliday junction recombination intermediate, respectively (Bachrati, et al., 2006; van Brabant, et al., 2000). Finally, BLM can disrupt some non-B-form DNAs, such as G-quadruplexes (Sun et al., 1998). To perform these functions, BLM associates with conserved partner proteins, including topoisomerase IIIa, and RMI1 and 2 (Brosh, et al., 2000; Chang, et al., 2005; Meetei, et al., 2003; Singh, et al., 2008; Wu, et al., 2001; Wu, et al., 2000; Yin, et al., 2005).
Although it is possible to study BLM function using BS cell lines, these cells exhibit ongoing genomic instability making it difficult to distinguish primary from secondary consequences of BLM loss. Attempts to overcome this problem have focused on the use of siRNA/shRNA-mediated depletion of BLM, which suffers from an inability to acutely disable BLM and requires a time scale of several days to be effective. We reasoned, therefore, that a small molecule inhibitor of BLM would permit a number of studies to be undertaken that cannot be conducted with existing cell lines. Moreover, it was our intention to identify ‘molecular probes’ of BLM function that could be used not only to conduct synthetic lethality type screens in human cells, but also to serve as lead compounds in the search for potential anticancer agents.
In this study, we report the characterization of the first small molecule inhibitor of BLM, identified through high throughput screening (HTS) and medicinal chemistry optimization. This compound, designated ML216, is potent and acts through competitive inhibition of the DNA binding activity of BLM. In cell-based assays, ML216 induces SCEs in cells containing BLM, but not in otherwise isogenic, BLM-deficient cells. We discuss potential therapeutic uses of BLM-targeting compounds derived from ML216 in the treatment of certain tumor types.
We developed a fluorescence-quenching assay (Fig. S1) to monitor DNA strand separation catalyzed by BLM that was adapted for quantitative, high-throughput screening (qHTS) for inhibitors of BLM (see Methods). The assay utilized a dual-labeled, partial duplex DNA substrate (‘forked’ duplex) in which the fluorescence of a rhodamine type moiety is normally suppressed by a non-emitting dark quencher (BHQ-2) present on the opposite strand. This quenching is lost following the ATP-dependent unwinding of the two DNA strands by the BLM helicase and hence a fluorescent signal is produced (Fig. S1A and S1B). This assay was miniaturized to a 1536-well format and used to screen the Molecular Libraries Small Molecule Repository (MLSMR) library of approximately 350,000 compounds (Fig. S1C). The inhibitory activity of the compounds was monitored as a diminution in the fluorescent signal compared to that from the negative control sample containing DMSO alone. Dose-response curves of potential BLM inhibitors from the primary screen are shown in Table S2. Primary screening hits were analyzed in a more traditional helicase assay, where separation of a radiolabeled, forked duplex substrate into ssDNA products is measured using PAGE. Of the screening hits that showed activity in both the fluorescence- and radioactivity-based assays, compound MLS000559245 was selected for further medicinal chemistry optimization directed primarily at improving physicochemical properties such as solubility and cell permeability. An optimized analogue, ML216 (Fig. 1A), emerged as the lead molecule for the present study. Further details of the optimization process will be published elsewhere.
In the qHTS assay, we used a truncated variant of BLM that lacks the N-terminal 635 residues and the C-terminal 119 residues of BLM (hereafter called BLM636–1298) in order to meet the requirement for milligram quantities of recombinant BLM in the large-scale screen, which cannot be generated for the full-length BLM protein. Although the truncated protein is fully functional as a helicase, we wanted to confirm that ML216 is also able to inhibit the activity of full length BLM, and to do so in a concentration-dependent manner. We found that ML216 is a potent inhibitor of both full length BLM and BLM636–1298, showing similar IC50 values of 3.0 and 0.97 µM, respectively (Fig. 1B and 1C; Table 1).
BLM has a broad substrate range in addition to its ability to unwind the forked duplex DNA used in the initial screen. Additional substrates include DNA displacement loops (D-loops), Holliday junctions and G-quadruplexes (Bachrati, et al., 2006; Ellis, et al., 1995; Mohaghegh, et al., 2001; Sun, et al., 1998). In at least the case of the Holliday junction, BLM exerts its effect on the DNA via promoting branch migration of the 4-way DNA junction, and not by catalyzing DNA strand separation. Moreover, we have provided evidence previously that BLM might disrupt the mobile D-loop substrate used here via migration of the 3-way junction (Bachrati, et al., 2006). We asked, therefore, whether ML216 is able to inhibit BLM’s action on a D-loop, a G-quadruplex or a synthetic Holliday junction. We found that ML216 could inhibit disruption of the D-loop and Holliday junction substrates, although only at high concentrations (50 µM) (Fig. S1D and S1F). Moreover, only very modest inhibition of BLM’s G-quadruplex disrupting activity was observed (Fig. S1E). These data suggest that ML216 is much more potent as an inhibitor of BLM’s strand separation activity on a forked duplex substrate than it is as an inhibitor of either branch migration of DNA junctions or G-quadruplex DNA disruption.
Next, we asked whether ML216 could inhibit other human RecQ family helicases, which are closely homologous to BLM in their helicase domains. In parallel, we also tested a putative negative control, E. coli UvrD helicase, which is a very distantly related enzyme belonging to helicase Superfamily I. ML216 showed only very modest inhibition of the helicase activity of E. coli UvrD or human RECQ1 and RECQ5, even at a concentration of 50 µM, whereas BLM activity was inhibited by 99% at 50 µM (Table 1). ML216 did, however, show inhibition of the WRN helicase. ML216 inhibited both the full length WRN and a truncated WRN derivative (WRN500–946), with the former being 2.5-fold more sensitive to inhibition (based on IC50 values). BLM was a little more sensitive than WRN to inhibition by ML216 (1.7-fold based on IC50 values; Table 1). Despite the detectable inhibition of WRN by ML216, this compound appears selective for BLM in human cells, as we discuss below.
ML216 could, in principle, inhibit BLM via a variety of different mechanisms such as by competing with ATP binding, the driving force behind its DNA unwinding, or by preventing BLM from binding to DNA. To investigate the mechanism of action of ML216, we analyzed its effects on DNA binding by BLM using an electrophoretic mobility shift assay (EMSA). This analysis provided clear evidence of displacement of either ssDNA or forked duplex DNA from a complex with BLM with increasing concentrations of ML216 (Fig. 2A and Fig. S2A/B). To confirm this finding, fluorescence polarization analysis was undertaken, where the binding of BLM to a fluorescently labeled oligodeoxynucleotide corresponding to the upper strand of the forked duplex substrate (Table S1; A1) was monitored in the presence of a dilution series of ML216. This test also showed that ML216 inhibits the ability of BLM to bind DNA (Fig. 2B). Taken together, these data strongly suggest that ML216 competes with the DNA for binding to BLM.
BLM requires ssDNA binding for its ATPase activity. Consistent with ML216 being able to disrupt BLM’s binding to DNA, rather than being an inhibitor of ATP binding by BLM, the inhibition of ssDNA-dependent ATPase activity by ML216 (Ki of 1.76±0.26 µM) was non-competitive with respect to ATP (Fig. S2C).
For ML216 to be an inhibitor with broad utility, it would have to be cell membrane permeable and capable of inhibiting the function of BLM in living cells. To investigate this, we compared the effects of ML216 in two cell lines that were derived from GM08505, a fibroblast cell line isolated from a patient with BS. One cell line, PSNG13, has been transfected with an empty vector alone, and therefore lacks BLM. The other, PSNF5, is isogenic to PSNG13, but has been transfected with the BLM cDNA. Previous studies showed that such re-expression of BLM in the PSNF5 cells corrects the hyper-SCE phenotype of the parental GM08505 cells (Gaymes, et al., 2002). Using this isogenic pair of cells, we sought to identify whether ML216 could influence cell growth and SCE induction in a BLM-specific manner. We found that ML216 was an inhibitor of the proliferation of PSNF5 cells, but not of PSNG13 cells, and did so in a concentration-dependent manner (Fig. 3 A/B). Moreover, ML216 treatment led to a statistically significant increase in the frequency of sister chromatid exchanges (SCEs) in PSNF5 cells, but not in PSNG13 cells (Fig. 3 C/D). In each experiment conducted thus far, the exposure time of cells to ML216 was 24–72 hours. However, for ML216 to have broad utility in cellular studies, it would ideally have to inhibit BLM during only a short treatment period. Hence, we asked whether ML216 could induce SCEs in PSNF5 cells if present only during the last 8 hours of the experiment prior to collection of cells in metaphase. That way, BLM would be targeted by ML216 only during the second S-phase of exposure to BrdU, because SCE analysis requires that BrdU be present during two consecutive S-phases before cells are arrested with colcemid and metaphases are harvested. We found that this shorter exposure led to a statistically significant increase in the frequency of SCEs in PSNF5 cells expressing BLM (Fig. S3A).
PSNG13 cells are hypersensitive compared to PSNF5 cells to inhibitors of DNA replication, such as aphidicolin, an inhibitor of replicative DNA polymerases (Davies et al., 2004). We asked, therefore, whether ML216 might selectively sensitize PSNF5 cells to aphidicolin, presumably by inhibiting BLM. As shown in Figure 4A, ML216 increased the sensitivity of PSNF5 cells to aphidicolin (p<0.01) but had no sensitizing effect on isogenic PSNG13 cells devoid of BLM.
Previous studies (Chu, et al., 2010) using mouse embryonic stem (ES) cells have indicated that BLM is required to convert mitomycin C-induced DNA damage (likely DNA inter-strand cross-links) into double strand breaks (DSBs), which can be detected by immunofluorescent staining of cells for the phosphorylated version of histone H2AX (termed γ-H2AX). We found that PSNF5 cells showed a substantially higher frequency of γ-H2AX nuclear foci in response to mitomycin C (MMC) exposure than did PSNG13 cells, in agreement with the previous study of mouse ES cells (Chu, et al., 2010). Consistent with ML216 targeting BLM in living cells, we observed a statistically significant reduction in γ-H2AX focus formation in PSNF5 cells treated with the combination of ML216 and MMC compared to PSNF5 cells treated with MMC alone (Fig. 4B and S3B; p<0.05). As expected, ML216 showed no significant effect on the control PSNG13 cells, with or without MMC treatment (Fig. 4B and S3B; p>0.05) because these cells display no γ-H2AX foci following this MMC treatment regimen.
The cellular data presented thus far indicate that ML216 can target BLM in human cells, but do not rule out the possibility that it also targets WRN, given that WRN protein is inhibited by ML216 in vitro. To investigate this, we tested whether fibroblasts lacking WRN were differentially sensitive to ML216 compared to wild-type control fibroblasts. We found that ML216 inhibited proliferation of WRN+ and WRN− cells equally well, and similarly sensitized both cell types to aphidicolin (Fig S4). Taken together, these data indicate that ML216 can specifically inhibit the function of BLM in human cells.
In this study, we report the identification and characterization of the first small molecule inhibitor of the human BLM helicase. This compound, designated ML216, resulted from medicinal chemistry-based optimization of compound MLS000559245, which was identified via high throughput screen of a library of ‘drug-like’ molecules. Based on our analysis of ML216 in human cell lines, we conclude that the compound is membrane permeable and can target BLM in living cells. The general lack of any effects of ML216 on a BS cell line, which does not express BLM, is strongly suggestive that the primary target for ML216 in human cells is BLM.
The initial screen was conducted using a truncated variant of BLM, BLM636–1298, which was prepared from bacteria. Nevertheless, we confirmed that full-length recombinant BLM (isolated from yeast cells) was inhibited to a similar degree based on IC50 values. This strongly suggests that ML216 targets the central ‘catalytic core’ of BLM contained in both recombinant BLM preparations comprising the helicase, RQC and HRDC domains. Given the strong selectivity of ML216 for BLM compared to human RECQ1 and RECQ5, we believe that it must target a structural feature of BLM that is poorly conserved, even amongst many human RecQ family helicases. Although the inhibition of WRN seen in helicase assays indicates that ML216 can target this closely related helicase (albeit, with a somewhat lower potency), the cell-based assay results probing many different functions of BLM indicate strongly that ML216 does not interfere significantly with WRN function in vivo. Despite this contention, it is still possible that ML216 targets WRN in vivo, but thus does not have any effect on the cellular endpoints analyzed here, including proliferation and aphidicolin sensitivity. However, the inhibition of WRN did also potentially reveal an interesting feature of the mode of action of ML216. We found that a truncated (but catalytically active) WRN was inhibited weakly by ML216, while full length WRN was more sensitive. In contrast, both truncated BLM (residues 636–1298) and full length BLM were inhibited with similar IC50 values. Given that the truncated WRN retains the helicase domain, these data suggest that residues outside of the helicase core of WRN might participate in DNA binding/unwinding and be targeted by ML216. A likely candidate is the RQC domain. Unfortunately, we currently only have access to the high resolution structure of human RECQ1 (Pike, et al., 2009), and not of BLM, WRN or RECQ5, and therefore more detailed analysis of the binding region for ML216 on BLM will have to await new structural studies.
We have provided mechanistic insight into how ML216 targets BLM. ML216 could inhibit the DNA-dependent ATPase activity of BLM, but did so in a non-competitive manner with respect to ATP. The use of both gel retardation and fluorescence polarization assays revealed that ML216 inhibits DNA binding by BLM, which would be sufficient to block its helicase activity. DNA binding, particularly of partial duplex DNA structures such as the forked duplex studied extensively here, seems to require both the helicase and RQC domains of RecQ helicases (Pike, et al., 2009). Moreover, the HRDC domain is known to influence DNA substrate specificity of BLM by altering its affinity for different DNA structures (Wu, et al., 2005). ML216 showed preferential inhibition of the unwinding of a forked duplex by BLM. Its ability to inhibit disruption of D-loops, 3/4-way junctions, and G-quadruplexes required approximately 10-fold higher drug concentrations. This pronounced DNA substrate-structure dependency of inhibition, combined with the EMSA and FP studies, argue strongly for a selective engagement of ML216 into the BLM-nucleic acid substrate binding site.
During the cellular analyses of the effects of ML216, we analyzed whether this compound, through targeting BLM, might activate cell cycle checkpoint responses. However, we obtained no evidence to suggest that ML216 did so. Exposure of cells for 24 hours to 50µM ML216 did not cause cell cycle arrest in G1/S or G2/M, and did not appear to cause phosphorylation of well-established ATM or ATR targets, including CHK1 and CHK2 (data not shown). This is consistent with a lack of requirement for BLM to activate these checkpoints in human cells.
We have presented evidence that ML216 modulates the role of BLM in genome maintenance in human cells. The most revealing phenotypic effect of ML216 treatment of cells was an increase in the frequency of SCEs. Crucially, this induction of SCEs was seen only in cells expressing BLM and not in otherwise isogenic cells lacking BLM. Given the importance of BLM in suppression of SCEs in human cells (Chaganti, et al., 1974; Chu and Hickson, 2009), this increase in SCE argues strongly for on-target effects of ML216. An interesting feature of the action of ML216 is its ability to partially inhibit proliferation of cells expressing BLM but not those lacking BLM. It is clear from analysis of cell lines isolated from BS patients that BLM is not an essential gene in humans (unlike in mice) (Chester, et al., 1998). Nevertheless, as has been observed in numerous other cases, acute inactivation of enzyme function often has more deleterious effects than does removing the protein altogether. This may be due to secondary effects on protein complexes in which the target protein participates. Our in vitro biochemical data indicated that ML216 could inhibit WRN nearly as efficiently as it does BLM. Nevertheless, our cellular studies using cell lines lacking either BLM or WRN indicated that the effects of ML216 on cell proliferation, aphidicolin sensitivity and SCE induction were consistent with BLM being the primary target of the compound, and not WRN.
Very recently, Aggarwal et al. (Aggarwal, et al., 2011) characterized a set of small molecule inhibitors of WRN. The compound analyzed in most detail in that study (NSC 19630) was highly toxic to human cells in culture at low micromolar concentrations. We currently do not know why we did not see any anti-proliferative effects of ML216 on cells lacking BLM, even though ML216 inhibits WRN with a similar potency to that of NSC 19630. Of note, NSC 19630 is a maleimide derivative and therefore likely to be chemically reactive to the free cysteine thiol groups present on several proteins, as well as on non-protein cellular components, such as intracellular glutathione, a recognized mediator of cellular response to oxidative stress. It is possible, therefore, that these effects combined could conspire to produce the toxicity observed with NSC 19630.
We have identified the first small molecule inhibitor of the BLM helicase, the protein defective in, the cancer predisposition disorder, Bloom’s syndrome. This inhibitor, ML216, is cell membrane permeable and appears to be highly specific for inhibition of BLM in human cells. We suggest that ML216 is a suitable ‘molecular probe’ for detailed analysis of the role of BLM in the maintenance of genome stability, including the identification of factors that are required to compensate for the loss of BLM function. A major future use of BLM-targeting compounds lies in the testing of the hypothesis that tumor cells that depend on the so-called ALT (alternative lengthening of telomeres) mechanism for maintenance of chromosome ends (telomeres), rather than on telomerase, might be susceptible to BLM inhibition. Work in yeast has revealed that the BLM ortholog, Sgs1, is required for telomere maintenance in the absence of telomerase, at least under certain circumstances (Cohen and Sinclair, 2001; Huang, et al., 2001; Johnson, et al., 2001). Moreover, depletion of the BLM partner protein, Topoisomerase IIIa, from human ALT cells strongly inhibits proliferation (Temime-Smaali, et al., 2008). The availability of ML216 now allows the testing of BLM inactivation in ALT cells. A successful proof of principle would open up the potential for analogs of ML216, or newly developed BLM inhibitors, to be used clinically to target the 5–10% of tumors that depend for their continued proliferation on the ALT pathway. Given that several of the ALT tumor types, such as osteosarcomas, are refractory to conventional treatments, this could open up a new therapeutic strategy for targeting these treatment-resistant tumors.
Recombinant human BLM, WRN and RECQ1 helicases were expressed and purified to near homogeneity as described previously (Mohaghegh, et al., 2001; Popuri, et al., 2008). E. coli UvrD, the human RECQ5 and truncated WRN (residues 500–946 fused to GST) helicases were obtained from Drs S. Matson, P. Janscak and R. Brosh, respectively. Oligonucleotides were synthesized by Sigma-Aldrich (Haverhill, UK), Biosearch Technologies, Inc. (Novato, CA), and the oligonucleotide synthesis service of Cancer Research UK (CRUK) (see Table S1 for sequences).
The high-throughput screening for inhibitors of BLM activity was conducted using a forked duplex DNA substrate, which harbored a tetramethylrhodamine fluorophore at the 3'-end of one strand (Table S1; A1) and a non-fluorescent, Black Hole Quencher-2 at the 5'-end of the complementary strand (Table S1; A2). An increase in the fluorescence intensity as a result of the ATP-dependent unwinding of the duplex region by BLM was used to measure helicase activity. For details, see SI Methods.
For each substrate, a single oligonucleotide was 5′-end-labelled with [γ-32P] ATP using T4 polynucleotide kinase. The forked duplex, X12 four-way junction (Holliday junction), D-loop, and G4 DNA substrates were generated as described previously (Bachrati, et al., 2006; Mohaghegh, et al., 2001; Sun, et al., 1998; van Brabant, et al., 2000). Prior to use, each substrate was gel-purified and dialyzed against the buffers described in SI Methods. The concentration of the labeled substrate was determined using scintillation counting. For details, see SI Methods.
Helicase assays were carried out in a 100 µL reaction volume containing no more than 1% DMSO, with the inhibitors being added in a 1 µL volume. Results were expressed as percent unwinding normalized to the untreated (taken as 0%) and denatured DNA (100%) control samples. Enzyme concentrations required for 50% substrate unwinding, and inhibitor concentrations required for 50% enzyme inhibition (IC50 values) were determined using a non-linear regression model using GraphPad software. For detailed information, see SI Methods.
The forked duplex substrate was prepared by annealing the 3’-TAMRA-labeled A1 oligonucleotide (Table S1) with the unlabeled complementary strand, A2 (Table S1). A standard 20 µl reaction contained 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 2 mM MgCl2, 2 mM ATP, 1 mM DTT, 0.1% Tween-20 and 200 nM TAMRA-labeled forked duplex substrate. Following electrophoresis, DNA was visualized using the Bio-Rad ChemiDocTM XRS Gel Imager.
Equilibrium binding of DNA helicases to TAMRA-labeled oligonucleotides was monitored by fluorescence polarization. Serial dilutions of selected inhibitors were mixed with the indicated amounts of DNA helicase and the TAMRA-labeled oligonucleotide (Table S1; A1). Following incubation at room temperature for ~2 hr, plates were read on a ViewLux High-throughput CCD imager. For a detailed protocol, see SI Methods.
GM08505 is an SV40-transformed skin fibroblast cell line established from a patient with BS. PSNF5 is a stable transfectant of GM08505 expressing Flag epitope-tagged BLM protein, while PSNG13 is an empty vector control transfectant of the same cell line. These transfected clones have been described previously (Gaymes, et al., 2002). GM00637 and AG11395 are SV40-transformed fibroblasts from a normal individual and a patient with WS, respectively. Details of culture conditions for these lines are provided in SI Methods.
Cells were treated with BLM inhibitors at the indicated concentrations for up to 72 hr. At each time point, the WST-1 reagent (a pale tetrazolium derivative converted to an intensely-colored formazan product by the action of mitochondrial dehydrogenases) was added, and after a 4 hr incubation the plates were analyzed optically at 450 nm with a reference wavelength of 690 nm. For details, see SI Methods.
Clonogenic survival assays and SCE analysis were performed as described previously (Davies, et al., 2004).
PSNG13 and PSNF5 cells were treated with either DMSO or ML216 at the concentrations described for 48 hours before exposure to mitomycin C for a further 12 hours. Cells were then fixed and permeabilized using 4% paraformaldehyde and ice-cold 70% ethanol. Fixed cells were stained with a primary antibody for γ-H2AX (Upstate anti-mouse; 05-636). 4′,6-diamidino-2-phenylindole (DAPI) staining was used to define the location of cell nuclei.
Synthetic schemes and protocols describing the preparation of ML216 and other compounds, as well as supporting analytical data, are provided in SI Methods.
We are grateful to the members of the Hickson laboratory for their helpful discussions. IDH was funded by Cancer Research UK and the Nordea Foundation (Denmark). GHN was funded by the NIH-Oxford Graduate Partnership Program. This research was supported in part by Award No. R03MH087284 from the National Institute of Mental Health (to OG), the Molecular Libraries Initiative of the NIH Roadmap for Medical Research (grant U54MH084681), and the Intramural Research Programs of NIA and NHGRI, NIH. We thank Sam Michael for assistance with the robotic HTS. The SGC is a registered charity (number 1097737) that receives funds from the Canadian Institutes for Health Research, the Canada Foundation for Innovation, Genome Canada, GlaxoSmithKline, Lilly Canada, the Novartis Research Foundation, the Ontario Ministry of Research and Innovation, Pfizer, and the Wellcome Trust.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.