• Open Access

DNA ligand designed to antagonize EBNA1 represses Epstein–Barr virus-induced immortalization


To whom correspondence should be addressed.
E-mail: noguchi-kj@pha.keio.ac.jp


Epstein–Barr virus (EBV) transforms human B lymphocytes into immortalized cells in vitro and is associated with various malignancies in vivo. EBNA1, which is expressed in the majority of EBV-infected cells, recognizes specific DNA sequences at the cis-acting latent origin of plasmid replication (oriP) element of the EBV genome. EBNA1 plays a critical role in the viral episome maintenance and transactivates viral transforming genes in latently infected cells. Therefore, DNA-targeting agents that can disrupt the EBNA1–oriP interaction will offer novel functional inhibitors of EBNA1. Pyrrole–imidazole polyamides, sequence-specific DNA ligands, can be designed to interfere with the binding of various transcriptional factors. Here, we synthesized pyrrole–imidazole polyamides targeting EBNA1-bound DNA sequences and developed an inhibitor for the EBNA1–oriP interaction. A pyrrole-imidazole polyamide, designated as DSE-3, bound adjacent to the EBNA1 recognition sequences located in the dyad symmetry element of oriP, and selectively inhibited EBNA1–oriP binding both in vitro and in vivo. DSE-3 also inhibited the proliferation of established lymphoblastoid cell lines by eradicating EBV episomes from the cells. In addition, DSE-3 repressed the expression of viral transforming genes after infecting human peripheral blood mononuclear cells with EBV and, as a consequence, inhibited EBV-mediated B-cell immortalization. These results suggest that EBNA1 functions will be an attractive pharmacological target for EBV-associated diseases. (Cancer Sci 2011; 102: 2221–2230)

Epstein–Barr virus (EBV) is a ubiquitous human γ-herpesvirus. In vitro, EBV infection causes immortalization of lymphocytes to continuously proliferating lymphoblastoid cell lines (LCL) and this transforming ability of EBV appears to be involved in the pathogenesis of its associated diseases.(1,2)

During latent infection, including malignancies, EBV persists as a multicopy episome and typically expresses a restricted set of viral genes. Among the latent viral genes, EBNA1 is the one expressed in latencies of type I–III.(1) EBNA1 is a dimeric DNA-binding protein that binds to the oriP region of the EBV episome, which carries multiple copies of the EBNA1-binding sites, called the family of repeats (FR), and the dyad symmetry (DS) element.(3) The DS contains four EBNA1-recognition sites (site 1, 2, 3 and 4) and binding of EBNA1 to the DS initiates loading of DNA replication components, including the cellular origin recognition complex (ORC) and the mini chromosome maintenance (MCM) complex.(4,5) The FR consists of an array of imperfect 30-bp units, and each unit contains one EBNA1-binding site.(6) Binding of EBNA1 to the FR mediates tethering of the oriP-containing plasmid to the cellular chromosome(7) and ensures non-random segregation of EBV episomes in proliferating cells.(8) Thus, EBNA1 is a critical component of oriP-dependent viral DNA replication and segregation in dividing cells.

The consensus DNA sequences recognized by EBNA1 are palindromic sequences (5′-TAGCATATGCTA-3′) that are highly conserved at the oriP region.(3) Given that DNA-binding is essential for the pathogenic roles of EBNA1,(9,10) DNA ligands that selectively recognize EBNA1-bound sequences might competitively disrupt EBNA1–DNA binding, and are interesting candidates for EBNA1 inhibitor.

Pyrrole–imidazole polyamides containing N-methylpyrrole (Py) and N-methylimidazole (Im) amino acids are synthetic cell-permeable molecules that can bind in the minor groove of DNA.(11) The side-by-side pairings between Im and Py prefer G-C base pairs, and Py–Py pairing effectively recognize T-A/A-T base pairs.(12) These properties, called the “Dervan pairing rules,” enable proper recognition of specific DNA sequences by the designed polyamide. These designed polyamides can interfere with DNA binding of specific proteins and inhibit their functions.(13)

In this study, we synthesized a series of Py–Im polyamides targeting specific EBNA1-binding sequences, and examined their biological activities as inhibitors of EBNA1. The polyamide DSE-3 is hairpin-shaped Py–Im polyamide designed to target EBNA1-binding to a 6-bp sequence in site 3 of the DS. DSE-3 showed significant inhibitory activity against EBNA1-oriP binding in vitro and in vivo experiments. Notably, DSE-3 repressed EBNA1-dependent viral gene expression and immortalization of EBV-infected lymphocytes. These results indicate that DSE-3 might be a useful inhibitor of EBNA1, and EBNA1–DNA binding might be a promising molecular target for EBV therapeutics.

Materials and Methods

Synthesis of Py–Im polyamides.  Py–Im hairpin polyamides were designed to span the EBNA1 binding site at DS sites 1–4 (Fig. 1A,B). Polyamide was synthesized essentially as previously described.(14,15) Briefly, polyamides were synthesized by a machine-assisted 9-fluorenylmethyloxycarbonyl (Fmoc) chemistry solid-phase synthesis method, starting with Fmoc-β-Ala-Wang resin (Merck KGaA, Darmstadt, Germany), and the products were separated by reverse-phase high-performance liquid chromatography with a Chemcobond 5-ODS-H reversed-phase column (Chemco Scientific, Osaka, Japan). The structures of polyamides were confirmed by Electrospray Ionization-Mass (BioTOF II; Bruker Daltonics, Bremen, Germany) and the collected fractions were lyophilized to obtain the final polyamides as white/yellow powders. A stock solution of hairpin polyamide was dissolved in DMSO at 10–20 mM.

Figure 1.

 Schematic diagram of the EBNA1-binding sites and hairpin-shaped polyamides. (A) Schematic diagram showing the organization of the Epstein–Barr virus (EBV) linear genomes the target DNA for polyamides. The OriP region containing family of repeat (FR) and dyad symmetry (DS) are magnified, and EBNA1-binding sites 1–4 at the DS and expected binding of DSE polyamides are illustrated in the boxes. (B) Chemical structure of the hairpin-shaped polyamide. N-methylimidazole is indicated in blue. In the sequences presentation in each polyamide, β represents β-alanine, γ represents γ-aminobutyric acid and Dp represents dimethylaminopropylamide. Ball-and-stick models of the polyamides are shown between the two DNA strands. The closed and open circles represent Im and Py rings, respectively. The diamonds represent β-alanine, and the curved lines represent γ-aminobutyric acid. (C) Polyamide-mediated inhibition of EBNA1–DS probe binding assessed by EMSA. The biotin-labeled DS (position 8841–9143 in GenBank V01555.2) probe bound with GST–EBNA1 (458–607)–FLAG and free DS probes are indicated by arrows. The indicated concentration of polyamide was premixed with the biotin-labeled DS probe before mixing with GST–EBNA1–FLAG. The signals from the biotin-labeled probe were visualized using chemiluminescent detection reagent.

Electrophoretic mobility shift assay (EMSA).  Recombinant GST–EBNA1–FLAG and biotin-labeled DNA probes were prepared as described (Data S1). Biotin-labeled DS (8841–9143) and probes for sites 1–4 were used. Recombinant GST–EBNA1–FLAG protein (40 fmole) and biotin-labeled DNA probe (50 fmole) were incubated in 20 μL of reaction buffer containing 20 mM Tris–Cl (pH 7.5), 150 mM NaCl, 5% glycerol, 0.05% Tween-20, 100 ng/mL poly (dI-dC) and 10 mM MgCl2. After incubation at room temperature for 10 min, the samples were mixed with loading dye and loaded onto 4% (for the DS probe) or 8% (for sites 1–4 probes) native polyacrylamide gels with 0.5× Tris-borate-EDTA (TBE) running buffer (44.5 mM Tris, 44.5 mM boric acid, 1 mM EDTA, pH 8.3). For the polyamide inhibition assay, polyamide and biotin-labeled DNA probes were pre-incubated for 10 min at room temperature before mixing with GST–EBNA1–FLAG. After electrophoresis, the probes were electrophoretically transferred to a nylon membrane (Hybond N+; Amersham Biosciences, Piscataway, NJ, USA) at 380 mA for 30 min in 0.5× TBE buffer. Transferred DNA was crosslinked by ultraviolet C (UVC) at 120 mJ/cm2 and the protein–DNA complexes were detected using a Chemiluminescent Nucleic Acid Detection Module (PIERCE, Rockford, IL, USA).

DNase I footprint experiments.  The DS region (8988–9143) was amplified by PCR (primers: forward 5′-GGGAGATATCGCTGTTCCTTAGGAC-3′ and reverse 5′-GGAATTCCCCTTGTTAACCCTAAACGGGTAGCATATGC-3′) with genomic DNA as a template. One microgram of the 160-bp DS DNA product was digested by EcoRI and its 3′-end was labeled with [32P-α] dATP using Klenow DNA polymerase. DSE binding was carried out at room temperature for 20 min in the binding buffer used for the EMSA assay containing 10 ng of labeled DS probe and the indicated concentration of polyamide. DNase I digestion done in the presence of 2.5 mM CaCl2 and 5 mM MgCl2 for 10 min at room temperature with 4 × 10−3 units of DNase I (Roche Molecular Biochemicals, Indianapolis, IN, USA). Reactions were stopped by adding 20 mM EDTA and precipitation with ethanol. Probe DNA was dissolved by formamide-containing loading dye, heat-denatured, and electrophoresed on a 6% denaturing polyacrylamide gel containing 8.3 M urea and 1× TBE buffer at 1000 V for 120 min. The dried gels were analyzed using an FLA 7000 bio-image analyzer (Fujifilm, Tokyo, Japan).

Epstein–Barr virus infection and lymphocyte transformation assay.  Virus infection was done essentially as described before.(16) Briefly, B95-8 cells were transfected with OH2 plasmid expressing BZLF1 (kindly provided by Dr Maruo, Hokkaido University), and 5 days after transfection, the filtrated culture supernatants were collected as a source of EBV. Peripheral blood mononuclear cells (purchased from Lonza Walkersville, Walkersville, MD, USA) were infected with serially diluted (100 to 10−4) virus supernatant at the volume ratio of 1:1, and the infected cells were plated at a density of 105 cells/0.2 mL per well in 96-well plates, with 15 wells for each dilution. Half of the culture medium was replaced with fresh medium every 5 days. The number of wells with growing transformed cell colonies was counted 5 weeks after infection.

Quantitative real-time PCR assay.  To calculate EBV genome copy numbers after polyamide treatment, LCL were treated with 40 μM DSE-3 or 100 μM hydroxyurea (HU) and, every 3 days, half of the medium was replaced with fresh medium containing reagent for 6 days. Genomic DNA were extracted using a DNeasy Blood and Tissue kit (QIAGEN Sciences, Germantown, MD, USA) in two independent experiments. Equal amounts of DNA sample extracted from an equal number of cells were subjected to quantitative real-time PCR at least in triplicate for each sample in the same instrument at the same time. Quantitative real-time PCR was performed using a QuantiTect Probe PCR kit (QIAGEN) with EBV primers and probes labeled with 6-carboxyfluorescein and quenched with Black-Hole-Quencher 1a (BHQ1a) targeting BALF5, as previously described.(17) DNA amplification and real-time fluorescence detection of triplicate samples were performed using an Applied Biosystems model 7500 real-time PCR instrument (Life Technologies, Carlsbad, CA, USA). The amplification protocol consisted of an initial denaturation and a polymerase activation step for 15 min at 95°C, followed by 40 amplification cycles of 94°C for 15 s and 60°C for 1 min. Raji cells are reported to contain approximately 45 copies of the EBV genome and were used to prepare the standard curve of episome numbers.(18) The number of viral DNA copies in each cell was calculated using the standard curves.

To monitor the viral gene expressions after infection, hPBMC were infected with the B95-8 virus supernatant and simultaneously treated with 40 μM DES-3. At the indicated times, total RNA was extracted with ISOGEN reagent (Nippon Gene, Tokyo, Japan) and converted to cDNA using a PrimeScript 1st strand cDNA Synthesis Kit (TaKaRa Bio, Shiga, Japan). Quantitative real-time PCR was performed using a QuantiTect Probe PCR kit (QIAGEN) with primer and probe sets for EBNA2, EBNA1 and LMP2, as previously described.(19) The primer and probe set for GAPDH was purchased from TaqMan Gene Expression Assays (Applied Biosystems, Life Technologies, Carlsbad, CA, USA). We have examined change of relative expression levels of viral genes by DSE-3 treatment with a comparable Ct method. The Ct values of GAPDH were used as endogenous control for normalization, and target viral gene levels were measured on the same samples at least in triplicate for each sample. For each cDNA sample, the Ct for GAPDH was subtracted from the Ct for each viral gene at same time point to give the parameter ΔCt. The ΔΔCt is defined as ΔCt at indicated time − ΔCt of calibrator sample. Because expressions of viral gene EBNA2, EBNA1 and LMP2 were not detected at time 0 h, we used ΔCt values at time 12 h as a calibrator sample.

ChIP assay.  Recombinant EBV-infected Akata cells expressing EBNA1 tagged with triple-HA were established as described previously.(8) These cells were treated with 40 μM DSE-3 and harvested at the indicated times. DNA preparation and the ChIP assay were performed as previously described.(20) For immunoprecipitation of EBAN1–3 × HA, ORC2 and MCM3, 5 μg of anti-HA (clone 3F10; Roche Applied Science, Indianapolis, IN, USA), anti-Orc2 (H-300×; Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-MCM3 (ab4460; Abcam plc., Cambridge, MA, USA) or control antibodies were incubated with the same amounts of DNA overnight at 4°C. Then, equal volumes of collected DNA sample were subjected to quantitative real-time PCR assay with a SYBR Premix Ex Taq II Perfect Real Time kit (TaKaRa Bio) using synthetic SC5 (5′-TGTCATAGCACAATGCCACCAC-3′ and 5′-GGTCAGGATTCCACGAGGGTAG-3′) and Ori-Lyt (5′-GCCCGTTGGGTTTCATTAAG-3′ and 5′-CCAAATCTCGCGGACCTCTA-3′) primers.(4,21) Amplification of triplicate sample in a 50 μL volume was performed using an Applied Biosystems model 7500 real-time PCR instrument, and the amplification protocol consisted of an initial denaturation step for 30 s at 95°C, 45 amplification cycles of 95°C for 5 s and 60°C for 31 s, followed by a dissociation step of 95°C for 15 s, 60°C for 1 min, 95°C for 15 s and 60°C for 15 s. Some of the pre-cleared lysate was separated as an “input standard” and serially diluted input standards were also subjected to the same real-time PCR experiments to generate a standard curve.

Cell culture and growth inhibition assay.  The B95-8 cell line is a lymphoblastoid cell line derived from a marmoset cell immortalized by EBV. Raji and Namalwa cells are EBV genome-positive Burkitt’s lymphoma cell lines, and IB4 cells are EBV-infected lymphoblastic cell lines. Raji cells are estimated to have 45 copies of the EBV episome, whereas EBV genomes were integrated into cellular chromosome in Namalwa and IB4 cells.(22) The DG75 and RamosRA1 cell lines are EBV-negative Burkitt’s lymphoma cell line. These lymphoma cell lines were grown in RPMI 1640 medium (Sigma, St. Louis, MO, USA) supplemented with 10% heat-inactivated FBS (JRH Biosciences, Lenexa, KS, USA) and 50 μg/mL kanamycin (Meiji Seika Pharma, Tokyo, Japan) at 37°C in a 5% CO2 atmosphere. Human peripheral blood mononuclear cells (hPBMC) were grown in RPMI 1640 medium supplemented with 20% FBS and 50 μg/mL of kanamycin. EBV-infected hPBMC were grown in RPMI 1640 medium supplemented with 20% FBS, 50 μg/mL kanamycin and 0.1 μM cyclosporine A. Human lymphoblastic cell lines (LCL-1, -2 and -3) were established from EBV-infected hPBMC and expanded for <4 months. Human embryonic kidney transformed fibroblast HEK293, human cervical cancer HeLa, human colon cancer HCT-116 and human osteosarcoma U2OS cells were grown in Dulbecco’s Modified Eagle’s Medium (Sigma) supplemented with 7.5% FBS and 50 μg/mL kanamycin.

For the B95-8 cell growth inhibition assay, B95-8 cells (103 cells/100 μL per well, seeded the previous day) were treated with the indicated concentration of polyamide for 5 days. Surviving cells were stained with 0.4% crystal violet/methanol solution. For the cell growth inhibition assay (WST assay), cells were seeded (2 × 104 cells/mL) at triplicate in a 96-well plate without or with polyamide. After 5 days, the ratio of viable cells was determined by Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) and the IC50 values (the dosage of drug at which a 50% inhibition of cell growth was achieved) were determined from the growth inhibition curve. For the long-time (>5 days) growth inhibition assay, cells were expanded to fresh medium containing polyamide every 4 days. Viable cells were counted using a Coulter counter or using the trypan blue dye exclusion assay.


Py–Im polyamides targeting EBNA1–DNA binding.  EBNA1 binds to unique DNA sequences in the DS and FR elements of the viral oriP region (Fig. 1A). The crystal structure of DNA-bound EBNA1 has been resolved in detail (PDB entry 1B3T), and shows that the DNA-binding domain of EBNA1 recognizes several bases in the minor groove (Fig. S1A). The EBNA1-interacting sequences, typically 5′-TAGCA-3′, are located at sites 1–4 in the DS (Fig. S1), and the designed polyamides, DSE-1–3, were expected to target those bases (Fig. 1A,B). The inhibitory effect of polyamides on EBNA1–oriP binding was analyzed by EMSA, indicating that 3 μM DSE-3 significantly inhibited binding of EBNA1 to the DS (Fig. 1C). In addition, we confirmed that DSE-3 did not inhibit NF-kappaB-DNA binding, suggesting that the inhibitory activity of DSE-3 on the EBNA1–oriP binding would not be due to non-specific DNA-binding of DSE-3 (Data S1, and Fig. S2A–C).

DNA sequence preference of polyamides.  The DNA sequences of site 1, 2 and 4 in the DS show similarities with that of site 3 (Figs 1A, S3). We compared the inhibitory effects of DSE-3 on the binding of EBNA1 to each of these sites by EMSA and found that binding of EBNA1 to site 3 was most sensitive to DSE-3 (Fig. 2). Site 2 contains two base mismatches, as compared with site 3, and GST–EBNA1–FLAG binding to site 2 was particularly resistant to DSE-3 interference. Sites 1 and 3 (from B95-8 cells) and 4 have one base mismatch sequences compared with the most common site 3, and DSE-3 showed moderate inhibitory effects on GST–EBNA1–FLAG–binding to these sites (Figs 2, S3). We considered that mismatched DNA sequences would contribute to the relative resistance to DSE-3-mediated inhibition (Fig. S3).

Figure 2.

 Inhibitory effects of polyamides on EBNA1–DNA binding at sites 1–4. (A) EMSA experiments on EBNA1–oligonucleotide binding. GST–EBNA1–FLAG was incubated with a biotin-labeled double-strand oligonucleotide probe corresponding to sites 1–4 in the presence of 0, 0.1, 0.3, 1 or 3 μM DSE-3. GST-EBNA1–DNA binding was detected by chemiluminescence system, as described in Figure 1. Data were analyzed using ImageJ software developed at the National Institute of Health (MD, USA) and the reduction of EBNA1–DNA binding signals by DSE-3 are shown in (B). DNA sequences and inhibition ratio at 1 μM are shown in Figure S2.

Optimization of polyamides for sequence-specific binding.  We synthesized DSE-3-related polyamides (Fig. 3A). DSE-4 is a mismatched polyamide that was used as a negative control, and DSE-5 and DSE-6 contain modified amino residues at the hairpin turn. DSE-7–9 contain modified sequences of N-methylimidazole and N-methylpyrrole. Representative results of EMSA experiments showed that DSE-3 had the greatest inhibitory effects of all these molecules (Fig. 3B). Although DSE-3, DSE-5 and DSE-6 were expected to recognize the same DNA sequences, the inhibition of EBNA1–DS binding was weaker with DSE-5 and DSE-6 than with DSE-3. These observations suggest that the chiral amino residue at the α position of the hairpin turn plays additional and pivotal effects on the sequence selectivity of DSE-3. DSE-8 was expected to match sites 1–4; however, our observations suggest that DSE-8 was unstable in the DMSO solution because the inhibitory effects of DSE-8 were not reproducible or were very weak in most of the later experiments.

Figure 3.

 Effects of DSE-3 derivatives on EBNA1–DS binding. (A) Chemical structure of the hairpin-shaped polyamide. The orange dotted circles show the modifications relative to DSE-3. (B) Polyamide-mediated inhibition of EBNA1-binding to the DS probe shown by EMSA. The biotin-labeled dyad symmetry (DS) probe and polyamides (1 μM) were premixed and EBNA1–DNA binding complex were detected as in Figure 1. The complex of biotin-labeled DS probe with GST–EBNA1–FLAG and the free DS probe are indicated by arrowheads.

We next performed a DNase I footprint assay to examine the DSE-3-binding area on the DS. The four DSE-3 binding sites were found in the DS and these regions overlap sites 1–4 (Fig. 4A,B). DSE-5 and -6 bound similarly to the EBNA1-binding sites, except to site 4 at a concentration of 1 μM. We calculated their affinities to sites 1–4 using quantitative DNase I footprint experiments (Fig. 4C), and the Kd values indicate that DSE-3 had higher affinity to the DS than did the other polyamides (Fig. 4D).

Figure 4.

 Polyamide–dyad symmetry (DS) binding. (A) DNase I footprinting assay with the DS fragment (8988–9143) labeled with 32P-dATP at the 3′-end. A+G (Lane 1) is chemical digestion of the A+G reaction. Control lane is digestion with DNase I without polyamide. The concentration of each DSE polyamide was 1 μM. The DSE-binding sites were protected from DNase I cleavage. (B) The digestion ratio of (A) was quantified using a FLA 7000 bio-image analyzer. An amount of DNA fragment generated by DNase I treatment without polyamide was determined as digestion of control, and ratio of digestion at each site in the presence of polyamide were shown in a bar graph. (C) Quantitative DNase I footprint titration with DSE-3, -5 and -6 on the DS. The DS probes labeled with 32P were incubated with increasing concentrations of each polyamide (0, 0.03125, 0.0625, 0.125, 0.25, 0.5 and 1 μM), and DNase I digestion was performed as in (A). The results were analyzed as in (B) and the calculated DNA binding affinities of each polyamide at each site are shown in (D). ND, not determined.

Additionally, DSE-9 bound to the DS probe, but the binding profile of DSE-9 differed from that of DSE-3 because DSE-9 appeared to bind to a broad region between sites 2 and 3 (Fig. 4A). As DSE-9 contains a tandem N-methylimidazole structure that might recognize G-G sequences, we suspect that a tandem repeat of guanine, such as G-G, in the DS located adjacent to sites 1–4 might be the target of DSE-9 (Fig. S4). An EMSA showed that DSE-9 inhibition of EBNA1–DS was approximately 10-fold weaker than DSE-3 (Fig. S4), suggesting that the binding targets of DSE-9 might differ from those of DSE-3, or DSE-9 might simply bind non-specifically to DNA.

Polyamide inhibits growth of EBV-infected lymphoblastic cell lines.  Latently EBV-infected B95-8 cells were treated with the DSE series of polyamides for 5 days, and the surviving cell colonies were analyzed by crystal violet staining. DSE-3 had the greatest anti-proliferative effect of all DSE tested on B95-8 cells (Fig. 5A). An analogous growth inhibition assay with two EBV-infected human LCL (LCL-1 and -2) confirmed that DSE-3 had inhibitory effects on EBV-infected cells (Fig. 5B,C, filled symbols). Additionally, flowcytometric analysis showed that a longer duration of DSE-3 treatment perturbed the cell cycle profile and decreased the proportion of cells in the G2/M phases in particular (Fig. S5). In that experiment, the area corresponding to hypodiploid DNA content (apoptotic sub-G1 fraction) was increased in the DSE-3-treated cells, suggesting that cell death was induced in the DSE-3 treated cells. However, statistical analysis suggested that significant growth inhibition by DSE-3 was not observed in EBV episome-negative cell lines (Fig. 5D,E). DSE-9, which is not specific to the EBNA1-binding sequences, showed weak inhibition on B95-8 cell growth, and DSE-5 and -6, which are weaker inhibitors of EBNA1–DS binding in vitro, did show weaker inhibitory effect on LCL cell growth than DSE-3. Therefore, selective targeting to the EBNA1-binding sites seems to be important for the growth inhibition effects of DSE-3.

Figure 5.

 Effects of polyamide on growth inhibition of Epstein–Barr virus (EBV)-transformed cells. (A) EBV-transformed B95-8 cells were treated with the indicated concentrations of DSE-3 (in triplicate) for 5 days and the surviving cells were visualized by crystal violet staining. Dotted square indicates samples treated by DSE-3. (B) Two independently established lymphoblastoid cell lines (LCL) were treated with the indicated concentrations of each DSE for 5 days and the surviving cells were counted by a trypan blue dye exclusion assay. The ratio of viable cell number compared with that of the control sample (no polyamide) is shown (left panels). (C) Growth inhibition of LCL by DSE-3. Two independent LCL were treated with 40 μM of DSE-3 and the viable cell number was counted by trypan blue dye exclusion assay at each time point. Every 4 days, cells were expanded to fresh medium containing DSE-3. The cumulative viable cell numbers are shown. Filled symbols, DES-3-treated cells; empty symbols, control cells. (D) Effect of DSE-3 on growth of various cell lines. The DSE-3-mediated growth inhibition of various cell lines was tested by WST assay (left graph), and the values of IC50 of DSE-3 to each cell were shown (right tables). (E) Growth curve of cells in the presence of DSE-3. Cell lines were treated with DSE-3 (40 μM) or control vesicle, and viable cell numbers were counted by trypan blue dye exclusion assay at each time point. Open symbols are control (DMSO)-treated and filled symbols are DSE-3-treated cells. Statistic significances of inhibitory effect of DSE-3 at day 8 were analyzed by two-tailed paired Student’s t-test. **Values of calculated P < 0.01 are significant.

DSE-3 inhibits the cellular functions of EBNA1.  In terms of viral episome stability, treatment with HU, a ribonucleotide reductase inhibitor that suppresses DNA replication, for a significant period of time can reduce the EBV episome number in some cell types.(23) We tested the effects of DSE-3 on viral episome stability in LCL. Either treatment with 40 μM DSE-3 or 100 μM HU for 5 days suppressed LCL cell growth to a similar level (Fig. 6A), and the EBV episome number was analyzed by quantitative real-time PCR assay in two independent experiments. Using BALF5 as a primer probe target and Raji DNA (45 copies of viral episome in a cell) as a standard DNA, the viral genome copy numbers were calculated (Fig. 6B), and it was revealed that DSE-3 treatment for 5 days significantly reduced the EBV genome copy number in LCL (P < 0.01), whereas 5 days of HU treatment was insufficient (P > 0.05). It has been shown that EBNA1–DS binding recruits cellular replication components, such as ORC and MCM complexes, to initiate oriP-dependent DNA replication.(4) Therefore, we performed ChIP assays to investigate the effects of DSE-3 on the loading of EBNA1 and cellular factors to the DS in EBV-infected cells. After DSE-3 treatment, recombinant EBV-infected Akata cells expressing HA-tagged EBNA1 were crosslinked with formalin, and EBNA1-HA, ORC2 and MCM3 were immunoprecipitated to collect the bound DNA fragment. Quantitative real-time PCR targeting the oriP region using the SC5 primer set (Fig. 1A) indicated that EBNA1 binding was reduced by DSE-3, along with reduced accumulation of ORC2 and MCM3 at the oriP region (Fig. 6C, left, P < 0.05). In contrast, the accumulation of these components at another site, the ori-lyt region, a lytic replication origin not specific to EBNA1 (Fig. 1A), was not affected by DSE-3 (Fig. 6C, right). These data suggest that DSE-3 interferes with episomal DNA stability by inhibiting EBNA1–oriP binding in LCL.

Figure 6.

 Effects of polyamide on EBNA1–DNA binding in Epstein–Barr virus (EBV)-infected cells. (A) Cell growth was examined after 5 days of treatment with DSE-3 or hydroxyurea (HU). Cell numbers counted by trypan blue dye exclusion assay are shown. Open columns are control, filled columns are DSE-3 treated and gray columns are HU-treated cells. (B) Quantification of the EBV episome after polyamide treatment. Cells were treated as in (A) and genomic DNA was extracted 6 days later. EBV episome copy numbers were determined by quantitative real-time PCR using a BALF5 primer/probe set. The results from two independent experiments (Exp 1 and 2) are shown. (C) ChIP assay for oriP-bound proteins. EBV-infected Akata cells expressing HA-tagged EBNA1 were treated with 40 μM DSE-3 for 24 or 48 h, and the ChIP assay was performed with antibodies specific to HA, ORC2, MCM3 or control as described in Materials and Methods. Ratio of immunoprecipitated target DNA was determined by quantitative real-time PCR using a specific primer set for latent oriP (SC5, left) or ori-Lyt (right) region. Statistic significances of inhibitory effect of DSE-3 were analyzed by two-tailed paired Student’s t-test. ** indicates P < 0.01, and * indicates P < 0.05. LCL, lymphoblastoid cell lines.

Inhibition of Epstein–Barr virus-induced primary cell transformation by DSE-3.  Epstein–Barr virus latent genes EBNA2 and LMP-2 are induced after EBV infection in primary B cells during establishment of type III latency, and oriP plays a key role in viral gene induction in LCL.(24,25) Therefore, hPBMC were infected with B95-8 virus in the presence or absence of DSE-3, and viral RNA expression at the indicated times was analyzed by quantitative real-time PCR. The viral gene expression ratios (ΔCt) compared with cellular GAPDH mRNA at 12-h post-infection were determined as calibrators, and the ΔΔCt value at the indicated times shows induction of viral genes (Fig. 7A, open columns). This experiment showed that DSE-3 treatment suppressed the expression of viral genes (EBNA2, EBNA1 and LMP-2) after EBV infection (Fig. 7A, filled columns).

Figure 7.

 Inhibitory effects of DSE-3 on EBNA1-dependent events. (A) OriP-dependent viral gene expression in Epstein–Barr virus (EBV)-infected primary lymphocytes. Human peripheral blood mononuclear cells (hPBMC) were infected with EBV (strain B95-8) and RNA was extracted at the indicated times. The relative expressions of mRNA of EBNA2, EBNA1 and LMP-2 were examined by quantitative real-time PCR. As an internal control, GAPDH mRNA expression was also determined, and relative expression of viral gene during treatment is determined by ΔΔCt method. The ΔCt (Ct of viral gene − Ct of GAPDH) at 12-h post-infection was set as a calibrator sample, and the value of −ΔΔCtCt [indicated time] − ΔCt [calibrator sample]) is shown. An increase in viral gene expression is represented as an increase in the −ΔΔCt value. Statistic significances of inhibitory effect of DSE-3 were analyzed by two-tailed paired Student’s t-test. ** indicates P < 0.01, and * indicates P < 0.05. (B) Effects of DSE-3 on EBV-induced transformation. hPBMC were infected with EB virus supernatant in the presence of either vehicle (DMSO) control or 40 μM DSE-3 in a 96-well plate. Well with transformed cell colonies or DSE-3-treated were photographed 4 weeks after infection. (C) hPBMC were infected with serially diluted EBV with or without DSE-3 (each treatment tested on 15 wells) as in (B), and the number of wells with or without proliferating transformed lymphoblastoid cell lines (15 wells per treatment per virus dilution) was counted 6 weeks after infection. A well with transformed cell colonies was determined as “positive” and a well without it was “negative.”*Data were analyzed using two-tailed Fisher’s exact probability test to calculate statistic significances of DSE-3-mediated inhibition (P < 0.0001 at virus dilution 0 and P < 0.0005 at −1).

Deletion of EBNA1 represses EBV-induced immortalization of primary B-lymphocyte.(26) We examined whether DSE-3 could inhibit EBV-induced B cell transformation. After culturing EBV-infected hPBMC for 4 weeks, the transformed cell colonies grew in the absence of DSE-3, whereas DSE-3 treatment suppressed colony formation (Fig. 7B). Furthermore, 15 replicated wells were infected with lower concentrations of viruses to determine the inhibitory effect of DSE-3 on B cell transformation (Fig. 7C). This assay showed that 40 μM DSE-3 decreased EBV-mediated transformation efficacy in approximately 10% of the control for hPBMC. Collectively, these experiments confirmed that DSE-3 suppressed EBV-induced transformation.


In this study, we showed that a Py–Im hairpin polyamide, DSE-3, recognized EBNA1-binding sequences and inhibited EBNA1–DS binding. As expected, DSE-3 reduced oriP-dependent viral gene expression, episome stability in EBV-infected cells, and EBV-infected cell transformation. These data demonstrate the anti-EBV activity of DSE-3 by inhibiting EBNA1–oriP interactions.

While the EBNA1-binding sites 1, 2 and 4 contain one or two base-mismatch sequences, as compared with site 3, DSE-3 inhibited EBNA1–DS-binding to these sites with different affinities, suggesting that the mode of interaction between DSE-3 and its target DNA might be less rigid than for other published Py-Im polyamides. We presume that the DNA recognition mode of DSE-3 is not solely based on the “Dervan pairing rule.” It has been reported that chiral substitution of the γ-aminobutyric acid linker modulates the properties of polyamide hairpins with regard to DNA affinity and sequence specificity.(27) Consistently, the chiral amino substitution at the α position in the γ-aminobutyric acid linker of DSE-3 significantly affected the binding affinity of polyamide to the DS probe, as observed for DSE-5 and -6 (Fig. 4E). Therefore, some other factor(s) provided by the α-amino-substituted γ-aminobutyric acid linker would significantly stabilize DNA recognition by DSE-3.

Regarding the effect of DSE-3 on EBV-infected cells, we assume that DSE-3 also binds to the FR because the FR contains multiple EBNA1-binding sequences that are quite similar to those in the DS and the consensus sequence for EBNA1-binding (Fig. S6). Moreover, in terms of non-viral DNA-binding of DSE-3, there are likely to be multiple DSE-3 targeting sequences in the host cell’s genomic DNA. Binding of EBNA1 to the host cell’s genomic DNA might affect the expression of some host genes,(28,29) and DSE-3 will likely affect the expression of some non-viral genes. Thus, we cannot exclude a possibility that the polyamide exerts effects other than against EBNA1-binding. However, our assay showed that DSE-3 inhibited growth of EBV episome-positive cell lines but not of seven other cell lines, including Namalwa and another LCL, IB4 cells (Fig. 5D,E). Because viral genomes were integrated into cellular chromosomes in both IB4 and Namalwa cells, the suppressive effect of DSE-3 might be apparent to the viral episome containing LCL. Taken together, these observations suggest that the flexible but preferential targeting of DSE-3 to the EBNA1-binding sites in the episomal oriP region should contribute to the anti-EBV activity of DSE-3.

Various studies are attempting to develop inhibitors of EBNA1 functions as therapeutic options for EBV. Hsp90 inhibitors can inhibit the growth of EBV-transformed LCL and prevent EBV-induced primary B-cell transformation.(30) The G-quadruplex-interacting compound BRACO-19 preferentially inhibits the proliferation of EBV-positive cells.(31) In contrast, PARP-1 inhibitor enhances oriP-dependent plasmid maintenance.(21) Compounds that can bind to the EBNA1–DNA-binding domain have been identified as an inhibitor by in silico or in vitro high-throughput screening,(32,33) and cell-based high-throughput screening has revealed that roscovitine inhibited oriP-dependent transcription and oriP-episome persistence by suppressing CDK-mediated phosphorylation of EBNA1.(34) Although these studies focus on the compounds that bind directly to EBNA1 or regulate EBNA1, no previous study has described the effects of inhibitors targeting the sequences involved in EBNA1–DNA binding. Therefore, our study is the first to report a DNA ligand with anti-EBV activity. Notably, there are various types of EBV-associated diseases with different types of latencies. Although we do not have evidence on whether inhibition of EBNA1 is sufficient to treat EBV-positive malignant cancers, recent studies suggest that, other than EBNA1, transformation-associated viral or host cell gene products such as LMP-1, LFA-1 and NF-κB might offer additional targets for anti-EBV chemotherapy.(35,36) Developing agents with different pharmacological target will provide additional options from which to select the appropriate drug(s), and multidrug combination chemotherapy should offer better efficacy than monotherapy. Further research into anti-EBNA1 and anti-EBV agents will contribute to the development of molecular-targeted therapies for EBV-associated disease.


This work was supported by Grants-in-Aid for Scientific Research (C) from the Ministry of Education, Culture, Sports, Science and Technology, Japan (KAKENHI 20590068 and 23590134). We thank Dr Seiji Maruo (Hokkaido Univ) for providing the BZLF1-expressing plasmid.

Disclosure Statement

The authors declare no conflict of interest.