DLG2 impairs dsDNA break repair and maintains genome integrity in neuroblastoma

.


Introduction ¡ DNA repair in neuroblastoma
Neuroblastoma (NB) is a tumor arising from the embryonic neural crest, later presenting in the autonomous nervous system [1].It is one of the most common forms of pediatric malignancies and has a disproportionately high mortality rate [2].Clinical screening and eventual diagnosis of NB is difficult due to the vague appearance of symptoms which is compounded by the young age of the patients [3].NB can be defined by risk group, both low and intermediate NB have good treatment prospects, whereas high risk tumors are difficult to treat with current protocols [4].This difficulty to treat results in higher incidence of refractory NB [5] as well as higher mortality [6], indicating that new treatments are needed.High-risk NBs are highly aggressive, with specific genetic lesions, copy number variations (CNV), and structural chromosomal changes [7].Common genetic features of NB include chromosome 1p deletion [8][9][10] and 17q gain [11,12], whereas 11q deletion [7,13,14] and amplification of the proto-oncogene MYCN [15][16][17][18][19] account for approximately 30% and 20% of all NB cases, respectively.In NB, deletion of 11q has previously been shown to result in an increase in the number of chromosomes with DNA breaks as well as the total number of chromosomal breaks [7].
DNA strand breaks can be repaired using different mechanisms depending on the type of break.DNA repair of single-strand breaks will often be highly successful as the DNA template still remains.In contrast, double-strand break repair can be deleterious as there is no template remaining for use during DNA repair.There are two main mechanisms by which double-strand break repair occurs: (1) Homologous Repair (HR) is highly reliable and uses the sister chromatid as a template, but can only take place during the S and G2 cell cycle phases [20] and (2) non-homologous end joining (NHEJ) does not use a template and directly ligates the two ends of the DNA breaks, meaning that the repair mechanism can be active throughout the cell cycle, except during mitosis [21].

DNA repair genes are altered in 11q deleted neuroblastoma
In NB, the 11q-deleted region spans a number of candidate tumor suppressor genes such as ATM, H2AX, CHK1 and MRE11a [6], all of which are involved in DNA damage response (DDR) [27].Recently, 11q deleted NB was shown to have abnormally low expression of Discs Large Homologue 2 (DLG2; 11q14) [27,28].Low DLG2 expression has also been found in osteosarcoma [29] and ovarian cancer [30].Previous results show that DLG2 can regulate the cell cycle by altering cyclin A and B as well as potentially regulating DNA damage repair [27].All of the candidate tumor suppressors spanning the 11q region have been shown to lack the second hit as per the Knudson two-hit hypothesis [6].It has, therefore, previously been suggested that there is a coalescence of haploinsufficiency within a single pathway [28].However, this has not yet been proven.
In this study, we investigate the difference in expression of genes associated with NHEJ in primary NB of various stages, and can show that some of these changes may be induced by loss of DLG2.Furthermore, we can show that DLG2 loss increases the presence of DNA double-strand breaks, and that restoration of DLG2 results in an increase in overall genomic stability.

Plasmids, siRNAs, transfections, irradiation and treatment
The DLG2 isoform 7 (NM_001351274.2) over expression plasmid on a backbone of pcDNA3.1/C-(K)-DYK(OHuq102626D) vector was purchased from GenScript (Genscript Biotech Corporation).siRNA targeting DLG2 (cat.s4122) or Silencer™ Select Negative control No. 1 siRNA (cat.4390843) were purchased from Ambion (Thermo Fisher Scientific).Cell lines were grown to 80% confluence and subsequently transfected with DLG2 plasmid, empty vector "mock" (pCMV6-AC-GFP or pCMV6-Myc-DDK), si-DLG2 or scrambled control "mock".In brief, 500 ng plasmid DNA or 30 pmol siRNA was complexed with 12 μl Lipofectamine 2000 according to the manufacturer's protocol (Thermo Fisher Scientific).To induce DNA damage, the cells were uncovered and subjected to UVC radiation (ESCO; CRF/UV-30A; 253.7 nm) for 30 s at 1200 mm, immediately given fresh media, and allowed to recover for 2 h or 24 h prior to harvesting and compared to control cells.As a second method to induce DNA damage, cells were treated with 7.5 μM etoposide (Merck) for 24 h and subsequently harvested.

Comet assay
To determine DNA damage repair post UV exposure or etoposide treatment, comet assay was performed according to the manufacturer's protocol (Abcam, Cat.ab238544).Aliquots of 10 000 cells were mixed with low melting point agarose and added to comet assay microscope slides.The slides were allowed to harden for 15 min, lysed (pH 10; 4 • C) for 1 h in pre-chilled lysis buffer, subsequently washed 3 times in alkaline buffer and transferred into alkaline buffer for electrophoresis.The slides were run for 30 min at 1 V/cm, 300 mA, washed 3 times with H 2 O, incubated for 5 min with ice cold 70% ethanol, allowed to air dry, and incubated with Vista green for visualization.Measurements and analysis were performed using Image J software version 1.52a.Data were normalized to the control transfection and expressed as a percentage of the control.

Quantitative PCR (qPCR) analysis
Total RNA from NB cell lines and flies was extracted using the RNeasy plus mini kit® (Qiagen) according to the manufacturer's protocol.The RNA concentration was quantified by NanoDrop (NanoDrop Technologies) and 2 μg of RNA was reverse transcribed into double stranded cDNA on a T-professional Basic Gradient thermal cycler (Biometra) using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems).cDNA corresponding to 20 ng of RNA for each qPCR reaction was used.qPCR was performed on a Pikoreal qPCR System (Thermo Fisher Scientific) in triplicate for TaqMan Human and Drosophila transcripts (Table 1), using TaqMan™ Gene Expression Master Mix (Applied Biosystems, Cat.4318157).Quantitative gene expression data were normalized to the expression levels of the two human reference genes GAPDH and GUSB, or the fly reference gene Rpl32 (Table 1) using the Livak method [35,36].

Apoptosis assay
Apoptosis levels were determined in the transfected NB cells to confirm the previous qPCR results by the Apo-ONE® Homogeneous Caspase-3/7 Assay (Promega) by exciting at 485 nm and reading at 520 nm using a FLUOstar Omega multiplate reader (BMG Labtech), as per the manufacturer's instructions.After the addition of the caspase assay, the plate was incubated for 65 min at room temperature.Apoptosis was calculated by subtracting the fluorescence for the average blank background and then normalizing against the mean of the control cells.

NHEJ reporter cells
The relative level of NHEJ was determined using SHSY5y NHEJ reporter cells (DR5000-HNSH-Sce).The cells were transfected as previously described and the number of GFP positive cells quantified 48 h post transfection by microscopy at 20x magnification on the EVOS M7000 (Thermofisher).The number of positive cells was then normalized to the control and expressed as log2 Fold change.

Fly strains and crosses
Commercially available control white (w1118) flies (Drosophila melanogaster) and UAS-RNAi-dlg1 flies were crossed with da-GAL4 driver strain to silence gene expression.All strains were obtained from the Bloomington Drosophila Stock Center (Bloomington).Twenty female da-GAL4 flies were crossed with 10 male UAS-transgenic flies or control flies and the progeny incubated at 25 • C on standard fly media.10 flies per cross were subsequently harvested for subsequent RNA and DNA preparation.To determine the effect of UVC irradiation the flies were subject to 30 Seconds UVC and 3 h recovery on regular substrate.

Nanopore libraries and Sequencing
To prepare ONT libraries D. melanogaster gDNA was extracted with the Blood & Cell Culture DNA Kit (13323, Qiagen) using the man-ufacturer´s tissue protocol.Libraries were prepared with 1 μg gDNA using the Ligation Sequencing Kit SQK-LSK109 (Oxford Nanopore Technologies) combined with native barcoding expansion 1-12 PCR free EXP-NBD104 (Oxford Nanopore Technologies) according to the manufacturer's protocol.Two 24 h sequencing runs for D. melanogaster were performed with R9.4.1MinION flow cells.

Sequence alignment and quantification of DNA integrity
To quantify DNA breakage, a strategy of global alignment between the full length of the sequenced reads and the corresponding reference genome region, where the read best aligned was chosen.Initial alignment was performed with BLASTn version 2.9.0+.A custom BLAST reference database was created from FlyBase D. melonagaster version 6.38 (FB2021_02, released April 13, 2021) [37], filtered to only contain chromosomes.All reads passing quality control were subsequently BLASTed to this database using a gap opening and extension penalty of 1 and an e value cutoff of 1e-10.Only hits that uniquely mapped to the reference genome were considered for further downstream alignment.Global alignment was performed using the needle function from the Biopython module EMBOSS [38] on the full length of the read and the corresponding region of the same length in the reference genome.Every mismatch, insertion and deletion were counted in these alignments.To quantify fused sequences, previously discarded reads were counted that fit the following criteria: the read has exactly two BLAST hits and there is at least 10 kb between the two hits to account for larger deletions in the sequenced DNA.

Statistical analysis
All data are presented as Tukey's box and whisker plots showing IQR, line at the median, + at the mean with whiskers ± 1.5-fold of interquartile range for at least 3 independent experiments.For all multigroup analyses, differences were determined by one-way ANOVA test followed by Holm-Sidak's multiple comparison test.For comparisons between two groups, Mann-Whitney U test was performed.A p < 0.05 was considered to be statistically significant.All analyses were conducted using GraphPad Prism version 9.0.1 for Windows (GraphPad Software, www.graphpad.com).

DLG2 enhanced DNA repair in NB cells after UVC irradiation
To determine whether NB cells with or without 11q-deletion had more tendency to accumulate DNA breaks, we induced DNA breaks by exposure to UVC in combination with a forced change in DLG2 expression level.DNA integrity was assessed by exposing DLG2 transfected NB cells to UVC irradiation followed by comet assay.We could determine that silencing of DLG2 resulted in a smaller comet assay head size (68.7%,p < 0.001) and longer tail length (60.1%, p < 0.001) in 11q normal NB cells prior to UVC irradiation.Silencing DLG2 in 11q deleted cells resulted in no change in head size (15.0%,p = 0.21) but an increased tail size (33.3%, p < 0.001), indicating fewer intact DNA strands and more DNA breaks in cells with lower DLG2 levels (Fig. 2a, b).A decrease in head size (54.1%,p < 0.001) was also detected in the DLG2 over expressed cells before UVC irradiation in 11q normal NB cells.However, tail length was shown to be decreased in both 11q normal (43.2%, p < 0.001) and 11q-deleted (29.4%, p < 0.001) cells with DLG2 over expression prior to UVC irradiation, indicating a general effect on the amount of DNA breaks without inducing breaks (Fig. 2b).To determine the effect of DLG2 on DNA repair, we induced DNA breaks with UVC irradiation in 11q-deleted and 11q-normal NB cells and allowed the cells to recover for 4 h.Using comet assay, we could determine that there was a decrease in the head size in both 11q-normal and 11q-deleted cells after irradiation in DLG2 silenced cells (20.1%, p < 0.001% and 50.4%, p < 0.001) and control samples (36.1%, p < 0.001% and 38.2%, p < 0.001); no difference was seen between the irradiated and non-irradiated samples when DLG2 was over expressed (Fig. 2a, b).There was a corresponding increase in the length of the tail in these samples; DLG2 silenced cells (83.8%, p < 0.001 and 33.9.4%, p < 0.01) and control samples (42.7%, p < 0.001% and 51.6%, p < 0.001), with no increase in the tail length of the DLG2 over expressed cells (Fig. 2b).To verify our hypothesis that it was the level of non-homologous end joining (NHEJ) that was affected by DLG2 expression, we used the SH-SY5y NHEJ reporter cell line with over expressed DLG2, mock or DLG2 silencing.Silenced DLG2 resulted in an increase in NHEJ active cells compared to control (log2 FC = 0.727, p < 0.001) with the over expression of DLG2 resulting in a decrease in the activation of NHEJ (log2 FC = − 0.706, p < 0.001; Fig. 2c).
To further determine the effect of UVC irradiation on DNA breakage and DLG loss we compared the DNA integrity after UVC irradiation in a Drosophila model with RNAi-silenced DLG (dmDLG knockout) to controls.We evaluated the percentage of DNA-sequences that were multiple, unique and not mapped (Fig. 2d) in UVC treated and untreated control (white) and dmDLG knockout flies.We could ascertain that there were more unmapped sequences in the UVC treated, dmDLG knockout flies compared to the untreated dmDLG knockout (7.8% increase over the control, p < 0.05).There was no difference in the percentage of mapped sequences for any of the samples.DNA from the UVC treated dmDLG knockout samples had a higher percentage of non-mapped sequences compared to the UVC treated control (4.33% increase over the control, p < 0.01; Fig. 2d).We further we determined the number of multiple mapped sequence reads that contained two separate fragments from non-overlapping (greater then 10,000 bp gap) sequences.We could show that the untreated control and untreated dmDLG knockout resulted in no difference in the percentage of detected fusion sequences (0.18 fold, p = 0.61), there was a large increase in the number of fusion sequences when the control and dmDLG knockout UVC treated samples were compared against their untreated control (3.4 fold, p < 0.001 and 4.3 fold, p < 0.001, respectively).dmDLG knockout after UVC showed a large increase in the number of fusion samples when compared to the UVC treated control (7.9 fold increase, p < 0.001; Fig. 2e).
Finally, we determined the median sequence length for the multiple, unique and not mapped sequences for each of the four experiments.The UVC dmDLG knockout flies showed a longer multiple mapped sequences read length compared to the untreated dmDLG knockout flies and the treated control flies (836 bp longer, p < 0.001 and 868 bp longer, p < 0.001, respectively; Fig. 2f) with no significant differences between the other samples.The median read length for UVC dmDLG knockout flies was longer compared to the untreated dmDLG knockout flies (654 bp longer, p < 0.01) with no differences between the other samples.The median length for the unmapped sequences was longer for the untreated control compared against the UVC treated control (788 bp longer, p < 0.001) with no difference between the other samples.

DLG2 over expression induced apoptosis in NB cells
We evaluated the effect of over expression of DLG2 on the initialization of apoptosis in SKNBE (MYCN-amplified) and SKNAS (11qdeleted) NB cells 48 h post transfection.Overexpression of DLG2 in SKNAS cells resulted in increased mRNA expression of BAX (log2 FC = 1.47, p < 0.001) as well as decreased BCL2 expression (log2 FC = 0.713, p < 0.001); DLG2 silencing resulted in decreased mRNA expression of BCL2 (log2 FC = 0.744 p = 0.001) with no alteration in BAX expression (Fig. 3a, b).The ratio of BAX/BCL2 gene expression was determined and shown to be high in cells with DLG2 over expression (FC = 4.16, p < 0.01), with no alteration for DLG2 silenced cells (FC = 0.329, p = 0.931; Fig. 3c).To determine if this resulted in apoptosis initialization we determined caspase 3/7 activation over time between 0 and 70 h.There was no difference in caspase 3/7 activation between control cells and DLG2 transfected cells 0, 16, 20 or 24 h post transfection; there was an increase in activation at 30 h (27.8%, p < 0.001) and 54 h (15.6%, p < 0.001), with no difference in the level of activation again at 70 h (Fig. 3d).We could confirm that BAX gene expression changes after DLG2 over expression resulted in increased levels of BAX protein in both SKNBE and SKNAS cells, and that BCL2 protein levels were lowered in SKNAS cells (Fig. 3e).DLG2 over expression resulted in an increase in total p53 level and increased degree of p53 Ser46 phosphorylation in both SKNBE (log2 Fold change = 1.0, p < 0.05% and 19.3% increase compared to the control, p < 0.05; Fig. 3f-g) and SKNAS (log2 Fold change = 0.69, p < 0.01% and 9.4% increase compared to the control, p < 0.01; Fig. 3h-i) cell lines.These results indicated that DLG2 over expression could induce apoptosis in NB cells.

DLG2 over expression increases S46-p53 after exposure to etoposide
To confirm if UVC and etoposide resulted in the same effect on NB cells, we altered DLG2 expression level in cells with or without 11q-deletion to determine if DNA breaks accumulated.DNA integrity was assessed after exposing DLG2 transfected NB cells to 7.5 μM etoposide with no recovery period followed by comet assay.We could determine that there was a decrease in the head size in both 11q-normal and 11qdeleted cells after etoposide treatment in DLG2 silenced cells (16.6%, p < 0.01% and 41.8%, p < 0.001) and control samples (43.3%, p < 0.001% and 17.6%, p < 0.001); no difference was seen after etoposide treatment when DLG2 was over expressed (Fig. 4a).There was a corresponding increase in the length of the tail in these samples; DLG2 silenced cells (35.5%, p < 0.001% and 32.1%, p < 0.001) and control samples (45.4%, p < 0.001% and 16.6%, p < 0.01), with no increase in the tail length of the DLG2 over expressed 11q deleted cells, although an increase in the 11q normal cells (21.2%, p < 0.001) (Fig. 4b).
To confirm the post translational modifications caused by etoposide treatment we quantified the degree of phosphorylation of CHK1-S317, CHK2-T68, P53-S46, H2AX-S139 and the total protein, with the representative immunoblots from SKNBE (11q normal) and SKNAS (11q deleted) shown in Fig. 4c.Treatment of SKNBE with 7.5 μM etoposide resulted in an increased total p53 expression in control (Log2 Fold change 0.35 < 0.01 and siDLG2 Log2 Fold change 0.91 < 0.001; Fig. 4d) and a decrease in DLG2 (Log2 Fold change 0.18 < 0.05; Fig. 4d) The degree of S46 p53 phosphorylation decreased with etoposide treatment in DLG2 silenced cells (32.1% less phosphorylation, p < 0.001).The over expression of DLG2 resulted in a high degree of S46 phosphorylation of p53 that was slightly lowered after etoposide treatment (14.3% less phosphorylation, p < 0.05; Fig. 4e).The degree of S317-CHK1 phosphorylation was increased by etoposide treatment in DLG2 silenced cells (41.8% more phosphorylation, p < 0.001; Fig. 4f), this was not seen in control or DLG2 over expressed SKNBE cells.The degree of T68-CHK2 phosphorylation was not affected by etoposide in SKNBE cells (Fig. 4g).

High DLG2 levels enhanced CHK1 phosphorylation after UVC irradiation in NB cells
To determine the effect of DLG2 on DNA repair in NB, we investigated DNA damage response pathways (DDR) in the NB cell lines NB69 (MYCN and 11q normal), SKNBE (MYCN-amplified) and SKNAS (11q deleted) after DLG2 silencing or DLG2 over expression and UVC irradiation.We could show an increase in phosphorylated CHK1-S317 in all cells 2 h after UVC irradiation, with the lowest level in DLG2 silenced cells, in NB69 and SKNBE cells the degree of CHK1-phosphorylation was lowered after 24 h of recovery (Fig. 5a-b).There was no difference in basal CHK1 expression (Fig. 5a).Subsequent evaluation of CHK2-T68 phosphorylation demonstrated high basal expression in DLG2 silenced and control cells with equal phosphorylation after 2 h of recovery, decreasing after 24 h of recovery (Fig. 5a,c).No clear and consistent differences in the total amount of ATM or DNA-PKCs could be determined across the different cell lines.However, silenced DLG2 and 24 h recovery after UVC irradiation in NB69 showed produce an increase in ATM (Fig. 5a).To further investigate the presence of double-strand DNA breaks, we investigated the presence of γ-H2AX (Ser-139 phosphorylated H2AX, a biomarker of DNA double-strand breaks).We could determine that γ-H2AX levels were maintained at low levels in DLG2 over expression cells regardless of UVC exposure, indicating few remaining DNA breaks (Fig. 5a).γ-H2AX levels were increased after UVC exposure in all cells after DLG2 silencing, and at the highest level after 24 h of recovery, with the exception of SKNBE cells which showed a peek after 2 h (Fig. 5a).
We quantified the phosphorylation of S317 CHK1 and T68 CHK2 and could show that phosphorylation of CHK1-S317 increased from 0 h to two hours in all DLG2 transfections with the smallest magnitude observed in the DLG2 silenced cells (42% increased phosphorylation, p < 0.001) compared to the control and DLG2 over expressed cells (63.2% increased phosphorylation, p < 0.001% and 59.7% increased phosphorylation, p < 0.001; Fig. 5b).After 24 h recovery the DLG2 over expressed cells were back to pre UVC treatment levels (3.8%, p = 0.78) with the control and DLG2 silenced remaining at higher levels (43.3% more phosphorylation compared to pre-treatment levels, p < 0.001% and 39.5% increased phosphorylation, p < 0.001, respectively; Fig. 5b).The degree of phosphorylation of T68 CHK2 showed no significant differences following UVC exposure in the DLG2 over expressed cells.Between 2 h and 24 h recovery after UVC exposure the phosphorylation decreased in the DLG2 silenced and the control cells (25.1% decrease between 2 and 24 h, p < 0.001% and 24.57% decrease between 2 and 24 h, p < 0.001; Fig. 5c), with no difference in DLG2 over expressed cells.

Discussion
Previous attempts have been made to determine the illusive tumor suppressor gene located on 11q, with varying success [6].It has previously been reported that there are two tumor suppressor genes located on chromosome 11q that could act in a combined haploinsufficient manner [27].Here, we investigated the convergence of DLG2 (11q14.1)and the 11q23 deleted candidate TSGs within the DNA damage repair pathway.Previous studies have shown that some of the 11q23 SRO candidate TSGs function within this pathway [39][40][41], the role of DLG2 in DNA repair has only been previously suggested with bioinformatics [27].We have shown changes in the expression of DNA repair genes in primary NB samples with unfavorable prognosis and 11q deletion (Fig. 1a-d).As patients with 11q deleted NBs are classified as high risk, with poor survival and a high number of DNA breaks and chromosomal alterations in the tumor, we investigated how DLG2 alters these genes of interest.
We have recently shown that there are two isoforms of DLG2 expressed in NB, isoform 2 (ISO2) and isoform 7 (ISO7); and that DLG2 isoform 7 is lost in advanced staged NB [42].Here, we show that DLG2-ISO7 interrupts the expression of genes associated with the DNA damage repair machinery, i.e.PARP1, FEN1, PCNA as well as induce elevated MRE11a expression (Fig. 1e), another gene located in the 11q deleted region of NB.Inhibition of PARP1 [43,44], CHK1 [45] and PCNA [46] have previously been investigated in NB, while FEN1 is known to be up regulated in tumors with MYCN amplification [47].We have previously shown that MYCN amplification results in lower DLG2 expression [27].We now show that DLG2 represses FEN1 expression, potentially providing a mechanism as to why MYCN amplification results in increased FEN1 expression.To further investigate these findings, we silenced the fly orthologue gene dmDLG in a Drosophila model, which subsequently resulted in increased dmFEN and dmPARP, thereby reproducing the altered PARP1 and FEN1 expression found in NB patient samples and NB cell lines (Fig. 1g).Subsequently we silenced dmDLG in the fly and irradiated with UVC which resulted in increased DNA damage in the form of an increase in the number of sequences that could not be mapped to a reference sequence as well as the number of multiple mapped sequences, combined with an increased average read length compared to control, indicated that there were likely fusion-sequences present, indicative of low genome integrity (Fig. 2d and f).This was confirmed by looking at the number of multiple mapped sequences that were comprised of two distinct fragments (Fig. 2e).The loss of dmDLG without the UVC treatment did not alter the number of fusions, however when combined with UVC irradiation activated low fidelity DNA repair; taken together with the results in human NB non-homologous end joining (NHEJ) reporter cells showing increased NHEJ after DLG2 silencing (Fig. 2c), we conclude that loss of DLG results in increased NHEJ.
To further investigate the effects that DLG2 expression has on DNA repair, we induced DNA breaks by exposing transfected cells to UVC irradiation and allowed the cells to recover or exposed them to the double strand break inducing drug etoposide.Here we could show that DLG2 maintained genome stability and integrity, whereas loss of DLG2 resulted in high fragmentation even without the extra stimulus of the UVC irradiation or etoposide (Figs.2a-b and Fig. 4a-b).This increase in DNA fragmentation may be the direct result of aberrantly activated endonuclease activity such as that found in FEN1 [48] or loss of the G2/M checkpoint.The increase in head size observed in the 11q deleted DLG2 overexpressed cells (Fig. 4a) may be attributed to maintenance of the cells at the G2/M checkpoint as previously established [27].After either etoposide treatment or UVC stimulus, the NB cells with 11q-deletion or DLG2 silencing displayed a highly unstable genome with persisting DNA strand breaks.The 11q-deleted cells were not affected by DLG2 silencing, likely since they already have very low expression of DLG2.We have previously established that DLG2 silencing reduces cell viability, implying a basal function [27] and correlating with observed clinical data.Over expression of DLG2 however seems effective in maintaining DNA-integrity after UV-exposure in both 11q-normal and 11q-deleted cells (Figs. 2a, 4a).To show that double-strand breaks were induced, we investigated phosphorylation of S139 H2AX.Here we could show that double-strand breaks were still present 24 h after UVC irradiation when DLG2 was down regulated (Fig. 5a), indicating that DNA repair had not yet occurred or that FEN1-mediated genome fragmentation was actively occurring.It has previously been shown that loss of DLG2 removes the G2/M DNA damage checkpoint [27], maintenance of which requires CHK1 activation [49].Consistent with this, we here show that restoration of DLG2 expression produces a rapid S317 CHK1 phosphorylation, whereas there was a decrease and delay in phosphorylation in the DLG2 silenced cells (Fig. 5b) in response to UVC.Interestingly, there was a large increase in phosphorylated S317-CHK1 in the DLG2 silenced cells after etoposide treatment (Fig. 4f and j), not seen in the control or DLG2 over expressed cells despite the S139-H2AX marker present in all of the etoposide treated cells, and faintly present in the DLG2 over expressed untreated cells (Fig. 4c).The activation of S317 CHK1 occurs primarily through ATR but it can also occur through ATM [50].The increasing phosphorylation of T68 CHK2 over time was present in the DLG2 silenced NB69 cells but not in the DLG2 over expressed cells.The phosphorylation of T68 CHK2 occurs primarily by ATM.Here we could show that ATM was highly expressed when DLG2 was silenced in the 11q normal NB69 cells, indicating that ATM is operational when DLG2 is inactive indicating that they function within different pathways [50].
We subsequently investigated the effects of DLG2 on initiating apoptosis.We could show that elevated expression of DLG2 resulted in increased total p53, the effect of which was seen in the expression of the transcriptional targets BAX and BCL2 (Fig. 3).The increased ratio of BAX/BCL2 increases the permeability of the mitochondrial membrane as BCL2 cannot inhibit BAX pore formation and leads to apoptosis [51].We could also detect an increase in pS46 phosphorylated p53 level, which is generally associated with severe genotoxic stress and leads to apoptosis [52].Genotoxic stress is present in MYCN amplified NB tumors [43] as well as 11q deleted tumors [7].We confirmed that there was a transient increase of the executioner caspases 3/7 post DLG2 over expression, showing that the pS46 p53 phosphorylation result in the initiation of apoptosis in NB by increasing BAX expression and allow for pS46 p53 mediated binding, a key step for transcription independent apoptosis [53].However, the subsequent events of late apoptosis can be further assayed to confirm that apoptosis continues to completion and ultimately consolidate the results.ATM is one of the essential kinases that phosphorylates S46 p53 directly or indirectly [23,40].In the data that we present DLG2 and ATM appear to be tentatively negatively correlated with low DLG2 increasing ATM protein expression in NB69, When combined with previous results showing that DLG2 is negatively associated with ATR [27], evidence is building that DLG2 prevents NHEJ DNA repair and results in cell death.Further studies into the coalescence of candidate 11q tumor suppressor genes are required to fully understand the causes of increased relapse and poor survival in 11q deleted NB patients.

Conclusions
We have shown that low DLG2 expression with additional DNA breakage stimulus results in high level of genome instability and increased NHEJ repair.DLG2 overexpression mediates the increase of pS46 p53 phosphorylation in response to genotoxicity in both 11q deleted and MYCN amplified NB cells, indicating that DLG2 can sustain the activation of apoptosis.This adds to the growing evidence that DLG2 is a NB tumor suppressor gene.

Fig. 1 .
Fig. 1.NB patients show differential NHEJ gene expression based on clinical classifiers.Microarray-based gene expression profiling of NHEJ genes in primary NB samples stratified by a Risk group (GSE49710), b survival (GSE16476), and 11q status c (GSE3960) and d (GSE73517).e qPCR 48 h post DLG2 over expression (DLG2 -DYK) or silencing (siDLG2) in NB69 cells (n = 9) with GAPDH and GUSB as endogenous controls, pooled average of three experiments.f Immunoblot in NB69 (MYCN and 11q normal), SKNBE (MYCN amplification) and SKNAS (11q loss) cells for DLG2, PARP1, FEN1, PCNA and GAPDH; -= DLG2 silenced, Mock = control, and + = DLG2 over expressed.g qPCR quantification of Drosophila NHEJ genes in fruit flies with silenced dmDLG expression (siDLG) (n = 9) with RPL32 as an internal control.The expression data are presented as median centered log2 fold change and plotted as Tukeys box and whisker plots showing IQR, line at the median, + at the mean with whiskers ± 1.5-fold of interquartile range.Data outside the whiskers are shown as outliers.Statistical significance was determined by one-way ANOVA with Tukey´s multiple comparison test.*P < 0.05, **P < 0.01, ***P < 0.001, ns= not significant.

Fig. 2 .
Fig. 2. DLG2 alters the effects of UVC irradiation on dsDNA strand breaks in NB cells.For DLG2 silenced (siDLG2), mock transfected (control) or DLG2 transfected NB69 (11q normal) and SKNAS (11q deleted) NB cells, alkaline comet assay was used to assess DNA repair by measuring a head diameter and b tail length in nonirradiated controls (black bars) and cells 4 h post UVC irradiation (grey bars).c The activation of the NHEJ pathway determined by the SH-SY5Y NHEJ reporter cell line after DLG2 silencing (siDLG2), mock transfected (control) or DLG2 over expression (DLG2).Effect of UVC irradiation and subsequent recovery on DNA integrity in fruit flies determined by nanopore DNA sequencing mapped to a reference genome, for untreated control (black bars), untreated dmDLG silenced (light grey bars), UVC treated control (grey bars) and UVC treated dmDLG silenced (dark grey bars) flies.Samples stratified based on multiple mapping, uniquely mapping or nonmapping for d percentage of total reads.e The percentage of fusion sequences from the multiple mapped sequences with at least 10 kb between the mapping locations, f median read length with SD stratified by multiple mapping, uniquely mapping or non-mapping classification.Black bars represent control flies and grey bars represent dmDLG silenced.The data shown is representative of three experimental replicates.The data are presented as an interleaved bar chart showing the mean and SD.Significance was determined by two-way ANOVA with Šídák's multiple comparison test *P < 0.05, **P < 0.01, ***P < 0.001, ns= not significant.

Fig. 3 .
Fig. 3. DLG2 induces apoptosis in NB in a p53 dependent manner.a BAX, b BCL2, and c BAX/BCL2 gene expression analysis by qPCR 48 h post DLG2 over expression (DLG2) or silencing (siDLG2) in 11q deleted NB SKNAS cells.The log2 fold change was determined using the Livak method normalized to the mean of GAPDH and GUSB.d Apoptosis induction by DLG2 over expression compared to control measured by Caspase 3/7-level over time in SKNAS cells.e DLG2, BAX, BCL2, p53, and pSer46 p53 protein expression measured by immunoblot in SKNBE (MYCN amp.) and SKNAS (11q Deleted) cells; + = DLG2 over expression.Total p53 expressed as log2 Fold Change and S46 p53/ Total p53 protein levels quantified by optical density in SKNBE (f-g) and SKNAS (h-i) cells.All data is the pooled average of three experiments.The data in c and d are shown as the mean ± SD.The data in a-c are presented as Tukeys box and whisker plots showing IQR, line at the median, + at the mean with whiskers ± 1.5-fold of interquartile range.Significance was determined by (a-c) one-way ANOVA with Dunnett ´s multiple comparison test and (f-i) unpaired two-tailed Student's t-test.*P < 0.05, **P < 0.01, ***P < 0.001, ns= not significant.

Fig. 4 .
Fig. 4. NB cell lines with low DLG2 expression maintain high CHK1 phosphorylation in response Etoposide treatment.For DLG2 silenced (siDLG2), mock transfected (control) or DLG2 transfected NB69 (11q normal) and SKNAS (11q Deleted) NB cells, alkaline comet assay was used to assess DNA repair by measuring a head diameter and b tail length in control (black bars) and etoposide treated (grey bars).c Protein expression patterns for CHK1-S317, CHK1, CHK2-T68, CHK2, H2AX-S139, H2AX and the housekeeping gene GAPDH in SKNBE (MYCN amp.) and SKNAS (11q Deleted) were determined by immunoblot.Total p53 expressed as log2 Fold Change with circles representing control treatment and square symbols etoposide treatment and S46 p53/ Total p53 in SKNBE (d-e) and SKNAS (h-i) cells.The percentage of CHK1 S317 phosphorylation relative to total CHK1 and CHK2 T68 phosphorylation levels relative to CHK2 in SKNBE (f-g) and SKNAS (J-k) were quantified by optical density.All data is the pooled average of three experiments.The data are shown as the mean ± SD.The data shown are representative of three experimental replicates.*P < 0.05, **P < 0.01, ***P < 0.001, ns= not significant.

Fig. A1 .
Fig. A1.Kaplan-Meier plots showing the overall survival probability of selected DNA repair genes in neuroblastoma.The expression groups are determined by median expression with the blue line representing high expression and the red line representing low expression in the GSE16476 dataset.(a) PARP1, (b) MRE11a, (c) FEN1, (d) XRCC1, (e) LIG3, (f) NBS1 and (g) PCNA.Kaplan-Meier plots are determined using the product limit estimator followed by Bonferroni correction.

Table 1
List of genes and probe ID used for qPCR Investigation.