Interference with DNA repair after ionizing radiation by a pyrrole-imidazole polyamide

Pyrrole-imidazole (Py–Im) polyamides are synthetic non-genotoxic minor groove-binding small molecules. We hypothesized that Py–Im polyamides can modulate the cellular response to ionizing radiation. Pre-treatment of cells with a Py-Im polyamide prior to exposure to ionizing radiation resulted in a delay in resolution of phosphorylated γ-H2AX foci, increase in XRCC1 foci, and reduced cellular replication potential. RNA-sequencing of cell lines exposed to the polyamide showed induction of genes related to the ultraviolet radiation response. We observed that the polyamide is almost 10-fold more toxic to a cell line deficient in DNA ligase 3 as compared to the parental cell line. Alkaline single cell gel electrophoresis reveals that the polyamide induces genomic fragmentation in the ligase 3 deficient cell line but not the corresponding parental line. The polyamide interferes directly with DNA ligation in vitro. We conclude that Py-Im polyamides may be further explored as sensitizers to genotoxic therapies.


Introduction
Half of all cancer patients undergo radiotherapy [1]. A primary mechanism of action of radiotherapy is induction of DNA damage. Combinations of radiation with radiosensitizing drugs is a fundamental treatment paradigm in clinical oncology [2]. Most systemic therapies currently used as radiosensitizers (e.g., flurouracil, cisplatin, mitomycin, gemcitabine, topoisomerase poisons) interfere with DNA repair as part of their mechanism of radiosensitization [3]. However, all of these drugs are genotoxic. Non-genotoxic small molecules that potentiate the effects of ionizing radiation in malignant cells could be used to augment radiotherapy for solid tumors.
Pyrrole-imidazole (Py-Im) polyamides comprise a class of small molecule minor groovebinders that are non-genotoxic [4]. Py-Im polyamides, oligomers of aromatic amino acids linked in series, fold into an antiparallel hairpin structure upon binding DNA [4]. The side-byside pairings of the Py and Im subunits determine DNA target sequence specificity. The ringpairs conform to steric and hydrogen bonding pattern differences in the floor of the minor groove presented by the edges of the base pairs [5]. DNAse I footprinting titrations and other methods have established the binding affinity and specificity of these molecules [6]. Protein-DNA interactions can be inhibited by allosteric changes in the DNA minor and major grooves induced by Py-Im polyamide-DNA binding [7]. Py-Im polyamides composed of eight aromatic rings localize to the nucleus in live cells [8]. Py-Im polyamides fail to activate a canonical DNA damage response [9], are not genotoxic on their own [10,11], and do not significantly alter cell cycle distribution at concentrations used for gene expression studies [9]. Py-Im polyamides have been used as molecular probes in cell culture to modulate gene-expression pathways [12,13] and interfere with RNA Polymerase II elongation [10,14]. We hypothesized that a Py-Im polyamide could modulate the cellular response to ionizing radiation. Pre-treatment of LNCaP and VCaP cells with a Py-Im polyamide prior to exposure to ionizing radiation resulted in a delay in resolution of phosphorylated γ-H2AX foci indicative of delayed repair of double strand breaks, and increased induction of XRCC1 foci consistent with a higher frequency of single strand breaks. RNA-sequencing of cell lines treated with the polyamide showed induction of genes related to the ultraviolet radiation response. We observed the polyamide is almost 10-fold more toxic to a LN-428 cell line deficient in DNA ligase 3 as compared to its parental cell line. Alkaline comet assay reveals that the polyamide induces genomic fragmentation in the Ligase 3 deficient but not the parental line. The polyamide interferes directly with DNA ligation in vitro. We conclude that Py-Im polyamides may be further explored as sensitizers to genotoxic therapies.

Cell viability measured using xCELLigence
The xCELLigence system noninvasively monitors viability of cultured cells by impedance, quantified as cell index (CI), representing cell number, viability, morphology. Assays were in 96 well plates with KD-BER-LN428-control, KD-BER-LN-428-LIG3 cells at 6000 cells/well.

Cell viability assay with PrestoBlue
Exponentially growing LNCaP cells on 6 well plates were treated with 10 μM 1 for 24 hours and then irradiated. The cells were pelleted and resuspended twice to remove polyamide from the media, plated at 4000 and 8000 cells/mL in 96 well plates. The plates were incubated at 37˚C for 2 weeks, PrestoBlue (Invitrogen) was added and incubated 30 min at room temperature, fluorescence read at 560 nm by spectrophotometry (SPECTRMax).

Alkaline comet assay
Alkaline comet assay was perform as described [15] using a Trevigen kit (4250-050-K). After treatment with 1 or vehicle, cells were centrifuged. Cells suspension of 1x10 5 /ml in PBS (Ca + + and Mg ++ free) were embedded in 300 μl of 1% low-melt agarose at 37˚C and 50 μL were mounted on CometSlides (Trevigen) pre-incubated at 37˚C. Embedded cells were lysed at 4˚C for 60 minutes in the dark, treated 20 min in alkaline unwinding solution (200mM NaOH, 1 nM EDTA pH>13.3) at room temperature in the dark, and electrophoresed (21V for 30 minutes) in a pre-chilled apparatus with fresh un-winding buffer as previously described. Slides were fixed in 70% ethanol for 5 minutes, dried at 37˚C, stained with 100 l of 2X diluted SYBR Gold (Invitrogen) for 30 minutes. Slides were imaged with a Leica DMR fluorescent microscope and quantified using OpenComet in ImageJ.

DNA T4 Ligase experiments
Sequences of oligonucleotides used in ligation assay are: 1a FAM, FAM-GACGCAAGTTCAGCT CGA; 1b CAAGTTCAGACGC; 2a CTGCGTTCAAGTCGAGCTGTTCAAGTCTGCG (Integrated DNA Technology). Ligation was performed in presence of varying concentrations of 1 and 2, 100 nM of the annealing oligos and 4U of T4 ligase for 1 hour at room temperature. The ligation was stop by adding 5 μl of TBE-Urea sample buffer and incubation at 70˚C for 3 minutes. The ligation products were analyzed by acrylamide electrophoresis on a 15% TBE-Urea gel [16] at 180 V for 1 hour. Image acquisitions of the gels were done by Typhoon Imaging System and image quantification by Quant Software.

RNA sequencing and analysis
LNCaP and VCaP cells were plated at 5 × 10 4 cells/mL in 10-cm 2 dishes, treated with or without 10 μM 1 in RPMI 1640 and DMEM supplemented with 10% FBS, respectively, for 24 hours. Total RNA was TRIzol extracted, sequenced (Illumina HiSeq2000), and mapped against the human genome (hg19) with Tophat2 using Ensembl GRCh37 gene annotations. Htseqcount was used for exon alignment and DESeq2 for differential expression. Pathway analysis was performed with the gene set enrichment analysis (GSEA) software on genes with padj<0.05 and p<0.05 for LNCaP and VCaP, respectively.

Cell cycle
Cells cultured at 70% confluence were treated with 10 μM 1 for 48 hours and cell cycle distribution assessed by monoparametric propidium iodide flow cytometry, analyzed by FacScan I (Becton Dickinson) and ModFit software.

Polyamide 1 slows resolution of histone γ-H2AX foci after irradiation
Polyamide 1 (Fig 1) does not cause genomic fragmentation by alkaline comet assay [11]. However, we hypothesized that 1 may interfere with repair of DNA after genotoxic insult. We examined the effect of 1 on the DNA double strand break repair dynamics in LNCaP and VCaP lines exposed to ionizing radiation. Phosphorylated γ-H2AX was used as a marker for double strand breaks [17]. LNCaP and VCaP cell lines grown with 5 and 10 μM 1 for 24 hours were irradiated (10 Gy) and immune-stained at baseline and after 1 and 24 hours with anti-γ-H2AX antibody. Phosphorylated γ-H2AX foci increased dramatically from baseline and remained elevated at 24 hours. Cells pre-treated with 1 had higher levels of phosphorylated γ-H2AX at both time points (Fig 2A and 2B). LNCaP cells subsequently exhibited reduced longterm proliferation after irradiation if pre-treated with 1 (S1 Fig). Because the radiosensitivity can be cell cycle dependent [18],

Polyamide 1 increases foci of single strand break repair rapidly after irradiation
Ionizing radiation induces at least an order of magnitude greater number of DNA single strand breaks than double strand breaks [19], although the latter are thought to be the lethal Interference with DNA repair after ionizing radiation by a pyrrole-imidazole polyamide lesion, as single strand breaks are rapidly repaired [20]. XRCC1 was used as a marker for single strand breaks. XRCC1 is recruited to sites undergoing single-strand break repair by poly (ADP-ribose) polymerase 1 (PARP-1), responsible for the initial recognition of the break [21,22]. Once XRCC1 is bound to the single strand break, it serves as a scaffolding platform to recruit, activate, regulate downstream repair enzymes. In order to assess the effect of 1 on the formation of single strand breaks after ionizing radiation, we measured the nuclear recruitment of XRCC1 (Fig 2C and 2D). LNCaP and VCaP were grown with 5 and 10 μM 1 for 24 hours, irradiated (1 Gy), and immuno-stained after 4 min with anti-XRCC1 antibody. Treatment of LNCaP with 1 alone resulted in no increase in XRCC1 foci, while VCaP cells had a small increase over baseline. Pretreatment of both cell lines with 1 prior to irradiation resulted in a large increase in XRCC1 foci as compared to irradiation alone. . Cytotoxicity of 1 was measured in both lines using the xCEL-Ligence system. LN428-LIG3 was 8.6-times more sensitive (Fig 3A) to 1 compared to LN428-control. To determine if reduced levels of LIG3 are associated with increased genotoxicity upon treatment with 1, both cell lines were treated with 10 μM 1 or vehicle for 24 hours and assessed by alkaline comet assay. We observed that 1 increases genomic fragmentation in LN-428-LIG3 but not LN428-control (Fig 3B).

Sequence dependent interference with DNA ligation in vitro
The ATP-dependent DNA ligases catalyze the joining of single-stranded breaks (nicks) in the phosphodiester back-bone of double-stranded DNA. We investigated if 1 can interfere with the DNA ligation in vitro using a fluorescent labeled 30-mer oligonucleotide with a nick in a top strand adjacent to flanking binding sites for 1. Ligation by T4 Ligase was inhibited by polyamide 1 but not 2, a polyamide which targets an unrelated sequence (Fig 4).

Polyamide 1 induces a transcriptional response in cells associated with the UV response
We performed gene expression profile analysis in both LNCaP and VCaP cells treated with 1 for 24 hours [24]. GSEA of affected genes for pathways in the Molecular Signatures Database revealed, for both cell lines, DAcosta_UV_response_via_ercc3_dn was the most negatively enriched, and Dacosta_UV_response_via_ercc3_up was most enriched in VCaP (S3 Fig).
Ercc3 codes for xeroderma pigmentosum type B, which is involved in basal transcription and single strand break repair. UV irradiation generates a number of DNA damage lesions including cyclobutane-pyrimidine dimers and 6-4 photoproducts [25]. These lesions are repaired through single strand break repair pathways that require the activity of DNA ligases [25]. The downstream transcriptional consequences of interference with these repair pathways by 1 may resemble those induced by UV irradiation. Similar GSEA results for polyamides have been reported in other cell and tumor samples [24].

Discussion
We have previously shown that polyamides can modulate a variety of DNA-dependent processes, including transcription factor-DNA binding [12,13], RNA polymerase II elongation [10,14], DNA helicase activity [9], and integration of viral DNA into mammalian cells by integrase [26]. We now report in vitro data that a polyamide can interfere with DNA ligation when bound adjacent to sites of single strand breaks, in vivo data that a polyamide can inhibit DNA repair following genotoxic insult, and that siRNA knockdown of DNA ligase 3 increases polyamide cytotoxicity by >8-fold, associated with evidence of genomic fragmentation. Similar to prior reports [11], we find that the polyamide is not inherently genotoxic in cells with intact DNA repair mechanisms. However, the polyamide can potentiate the genotoxicity of ionizing radiation. Polyamide treatment of cells followed by exposure to ionizing radiation immediately increases foci of XRCC1 representative of single strand breaks and increases persistent foci of phosphorylated γ-H2AX representative of unrepaired double strand breaks. Overall, the data suggest a model where a polyamide may interfere with repair of single strand Interference with DNA repair after ionizing radiation by a pyrrole-imidazole polyamide breaks induced after ionizing radiation. Unrepaired single strand breaks in close proximity may become double strand breaks, which can cause cell death and reduced replication potential.
A prior study from our group showed that a related polyamide could exert low-level replication stress in cells accompanied by activation of ATR but not ATM, and at high concentrations (30-100 μM), led to accumulation of cells in S phase without detectable genotoxicity [9]. In vitro data showed that the polyamide could slow DNA unwinding by T7 gp4A helicase. Direct interference with DNA ligation of Okazaki fragments on the lagging strand by local polyamide binding offers a complementary explanation for these observations, and is consistent with our current results.
Cancers deficient in DNA repair pathways are often selectively sensitive to DNA damaging drugs and inhibitors of DNA repair [27]. Examples of this strategy include the use of PARP inhibitors or platinating agents for patients with DNA repair deficient breast [28], ovarian [29], and prostate cancers [30,31]. In LN428 cells stably expressing siRNA against DNA Lig3, polyamide treatment results in genomic fragmentation and increased cytotoxicity that is absent when DNA Lig3 is expressed at basal levels. To our knowledge, this is the first observation of a DNA repair deficient cell line with increased sensitivity to a polyamide. Although DNA ligase deficiency is a rare feature of cancers, the observation that reducing a cancer cells ability to enzymatically manipulate its DNA may selectively increase the cytotoxicity of a polyamide raises the question if polyamides may act synergistically with therapeutics that cause replications stress or interfere with DNA repair, in addition to therapies that induce genotoxic stress such as ionizing radiation.