Among the DNMT partners, we focused our analysis on the impact of tumorigenesis on the disruption of the DNMT1/PCNA, DNMT1/HDAC1, DNMT1/HP1β, DNMT1/EZH2, DNMT1/Sp1, DNMT1/DMAP1 and DNMT1/DNMT3b complexes (Figure 1A) 5 6 7 8 9 10 11 12 13 14 .

To investigate the impact of the disruption of these interactions on tumorigenesis, we transfected an astrocyte cell line (Astro#40) and a glioma cell line (U87) with plasmid constructions encoding for peptides mimicking certain amino-acid regions implicated in the interactions existing between DNMT1 and DMAP1, DNMT3b, PCNA, EZH2, HDAC1, Sp1 and HP1β (Figure 1A and Additional file 1: Figure S1).

After 4 weeks of transfection/selection/amplification of cells, proximity ligation in situ assays (P-LISA) revealed that the 47–60, 197–212, 163–174, 430–444, 712–725, 791–802 and 885–900 peptides were specific to disrupting the interactions between DNMT1 and DMAP1, DNMT3b, PCNA, EZH2, HDAC1, Sp1 and HP1β, respectively. For example, we observed in Astro#40 that the 197–212 peptide disrupted the DNMT1/DNMT3B interaction, but not the DNMT1/HP1β, DNMT1/HDAC1, DNMT1/Sp1, DNMT1/EZH2 and DNMT1/DMAP1 interactions (Figure 1B). More generally, we noted that peptides designed to specifically disrupt an interaction disrupted only the targeted and expected interactions in Astro#40 and U87 cells ( 2: Figure S2).

We then investigated the impact of these disruptions on the global level of DNA methylation, i.e., on the 5-me-thylcytosine level. Two DNMT inhibitors (5aza-2deoxy-cytidine and procainamide) were used as these drugs induced global DNA hypomethylation (Figure 1C). In agreement with our previous reports, the 163–174 plasmid was used as a peptide that induced global DNA hypomethylation 3 4 (Figure 1C). Thus, we noted that the disruption of DNMT1/HDAC1 and DNMT1/HP1β interactions promoted a global decrease in the 5-me-thylcytosine level in Astro#40 and U87 cells, while the disruption of the DNMT1/EZH2, DNMT1/Sp1 and DNMT1/DMAP1 interactions did not affect the 5-me-thylcytosine level in these cells (Figure 1C).

Tumorigenic assays performed in Swiss nude mice revealed that the subcutaneous injection of Astro#40 cells treated 4 weeks with 5aza-2deoxycytidine and procainamide or transfected with plasmid encoding for the 163–174, 712–725 and 885–900 peptides generated tumor formation in 3/5, 2/5, 5/5, 2/5 and 2/5 cases, respectively (Figure 2A). Thus, we observed that gliomagenesis is induced by the loss of DNMT1/PCNA, DNMT1/HDAC1 and DNMT1/HP1β interactions, i.e., under conditions that also induced a decrease in the global DNA methylation level. In parallel, we noted that the subcutaneous injections of U87 cells treated with 5aza-2deoxycytidine and procainamide or transfected with plasmid encoding for the 163–174, 712–725, 430–444, 885–900 and 197–212 peptides generated the formation of larger tumors as compared to untreated U87 cells (Figure 2B). These data suggest that the disruption of DNMT1/PCNA, DNMT1/EZH2, DNMT1/DNMT3B, DNMT1/HDAC1 and DNMT1/HP1β interactions acts such as an enhancer of tumorigenesis. Thus, our work stratifies the specific disruption of certain DNMT1/protein-x interactions into two categories (Figure 2C). The first category includes the specific disruption of certain DNMT1/protein-x interactions acting as inducers and/or enhancers of gliomagenesis (DNMT1/PCNA, DNMT1/HDAC1 and DNMT1/HP1β, DNMT1/EZH2, DNMT1/DNMT3b), i.e., acting as pro-tumorigenic-specific disruption of DNMT1/protein-x interactions. The second category includes the specific disruption of DNMT1/protein-x interactions devoid of any action on gliomagenesis. This category is called neutral-tumorigenic-specific disruption of DNMT1/protein-x interactions. Although absent in our data, we added a third category called anti-tumorigenic-specific disruption of DNMT1/protein-x interactions including the specific disruption of DNMT1/protein-x interactions promoting the decrease in tumorigenesis (Figure 2C).

In the absence of any observations of anti-tumorigenic-specific disruptions of DNMT1/protein-x interactions, we next analyzed whether the specific disruption of a DNMT1/protein-x interaction devoid of action on glioma-genesis could have an impact on the response to the standard GBM treatment combining temozolomide (TMZ) and irradiation. For this, we measured the percentage of cell death induced by a TMZ + irradiation treatment of U87 cells transfected with the plasmids encoding for the 47–60 and 791–802 peptides (Additional file 3: Figure S3). We noted that the 47–60 peptide, which induced the disruption of the DNMT1/DMAP1 interaction, increased the percentage of TMZ + irradiation-induced cell death, while the 791–802 peptide, which induced the disruption of the DNMT1/Sp1 interaction, had no effect on the percentage of TMZ + irradiation-induced cell death (Figure 2D). In addition, our data also indicated that the 5aza-2deoxycytidine and procainamide treatments and the 163–174 peptide induced a phenotype resistant to TMZ + irradiation cell death since the percentage of TMZ-irradiation-induced cell death significantly decreased under these conditions (Figure 2D). Taken together, these data indicate that, among the considered peptides and DNMT1 inhibitors, only the specific 47-60-induced disruption of the DNMT1/DMAP1 interaction increased the percentage of TMZ + irradiation-induced cell death without promoting or increasing the tumorigenesis. Thus, we conclude that the specific inhibition of the DNMT1/DMAP1 interaction is a neutral-tumorigenic-specific inhibition of DNMT1/protein-x interaction harboring the capacity to sensitize cells to TMZ + irradiation-induced cell death.

To investigate this point, established tumors were treated with TMZ, TMZ + 47-60^DNMT1 and mutated 47-60^DNMT1 (m47-60^DNMT1) (Additional file 4: Figure S4). Mice received 6 weeks of treatment as in treatment of human GBM, and tumor weight was analyzed 2 weeks after the end of the treatment. As illustrated in Figure 2E, we first noted that TMZ treatment reduced the tumor weigh of U87-induced glioma. Second, we observed that the addition of 47-60^DNMT1 increased the efficiency of TMZ treatment and decreased tumor growth, while m47-60 ^DNMT1 had no effect on the efficiency of TMZ treatment (since this peptide does not disrupt the DNMT1/DNMAP1 interaction; Additional file 5: Figure S5). Thus, our data identify 47-60^DNMT1 as an enhancer of TMZ treatment.

P-LISA is a technology permitting the visualization of stable and transient interactions at endogenous protein levels directly in situ 20 . Briefly, two primary antibodies raised in different species recognize the target antigen or antigens of interest. Species-specific secondary antibodies, called PLA probes, each with a unique short DNA strand attached to it, bind to the primary antibodies. When the PLA probes are in close proximity (< 40 nm), the DNA strands can interact through a subsequent addition of two other circle-forming DNA oligonucleotides. After ligating the two added oligonucleotides, creating a circle DNA molecule, they are amplified via rolling circle amplification. After amplification, several-hundredfold replication of the DNA circle has occurred, and labeled complementary oligonucleotide probes highlight the product. The resulting high concentration of fluorescence in each single-molecule amplification product is easily visible as a distinct bright dot when viewed with a fluorescence microscope.

Cells were cultured for 24 h on cover slips. Cells were then fixed with 4% paraformaldehyde in PBS, pH 7.4, for 15 min at room temperature. Permeabilization was performed with PBS containing 0.5% Triton X-100 for 20 min at room temperature. Blocking, staining, hybridization, ligation, amplification and detection steps were realized according to the manufacturer’s instructions (Olink Bioscience). All incubations were performed in a humidity chamber. Amplification and detection steps were performed in a dark room. Fluorescence was visualized by using the Axiovert 200M microscopy system (Zeiss, Le Pecq, France) with the ApoTome module (X63 and numerical aperture 1.4). Preparations were mounted using ProLong® Gold antifade reagent with DAPI (InVitrogen, France). Picture acquisition was realized in structured illumination microscopy 24 . Finally, the images were analyzed using the freeware “BlobFinder” available for download from http://www.cb.uu.se/~amin/BlobFinder. Thus, we obtained the number of signals per nuclei since nuclei can be automatically identified.

To express peptides in cells, we subcloned in pcDNA3.3 (Life Technology, France) the sequences encoding for the indicated peptides. In addition, the NLS sequence (PKKKRKV) was added to the sequences encoding for the indicated peptides in order to address the peptides in the nucleus. Tranfections were next realized by using 2.10⁵ cells, 5 μg of plasmid and Lipofectamine™ 2000 reagents (Life Technology, France). Selection was realized by adding 500 μg/ml of Geneticin selective antibiotic in complete medium of cell culture for 3 weeks. Next, 1 week was used to amplify the cells. Veracity of transfection was determined by PCR analyses using primers directed against pcDNA3.3 (ACGTTGTCACTGAAGCGG and CCTGATGCTCTTCGTCCA) and by the fact that each plasmid affects the DNMT1/protein-x interaction of interest.

DNA was extracted using the QiaAmp DNA mini Kit (Qiagen, France). The quantification of 5-methylcytosine is performed by using the Methylamp Global DNA methylation Quantification kit (Euromedex-Epigenetiek, France).

Cultured cells were harvested by trypsinization, washed and resuspended in saline buffer. Cell suspensions were injected s.c. as 10⁶ cells in 0.2 ml volume in the flank of 7-/8-week-old nude NMRI-nu female mice (Janvier, France).