PIMREG expression level predicts glioblastoma patient survival and affects temozolomide resistance and DNA damage response

PIMREG expression strongly correlates with cellular proliferation in both malignant and normal cells. Throughout embryo development, PIMREG expression is prominent at the central nervous system. Recent studies have described high levels of PIMREG transcripts in different types of tumors and correlated with patient survival and tumor aggressiveness. Given the emerging signicance of PIMREG in carcinogenesis and its putative role in the context of the nervous system, we investigated the expression and function of PIMREG in gliomas, the most common primary brain tumors. We performed an extensive analysis of PIMREG expression in tumors samples of glioma patients, assessed the effects of PIMREG silencing and overexpression on the sensitivity of glioblastoma cell lines treated with genotoxic agents commonly used for treating patients and assessed for treatment response, proliferation and migration. We show that glioblastoma exhibits the highest levels of PIMREG expression among all cancers analyzed and that elevated PIMREG expression is a biomarker for glioma progression and patient outcome. Moreover, PIMREG is induced by genotoxic agents and its silencing renders glioblastoma cells sensitive to temozolomide treatment and affects ATR- and ATM-dependent signaling. Our data demonstrate that PIMREG plays a role in DNA damage response and temozolomide resistance of glioblastoma cells and further support the PIMREG role in tumorigenesis.


Introduction
Gliomas are the most common primary brain tumors 1 and are classi ed according the World Health Organization (WHO) histological and grade (I-IV) criteria. The most common gliomas include astrocytomas (grade I-IV), oligodendroglioma (grade II-III) and oligoastrocytoma (grades II-III) 2 . Survival varies signi cantly by grade across glioma subtypes 3 . Glioblastoma (GBM) is the most aggressive and malignant subtype (astrocytomas of grade IV) and corresponds to more than 50% of all glioma cases.
Patients with GBM have the poorest outcome, with an average survival of 15 months 4 . The standard care for primary GBMs comprises surgical resection and concurrent temozolomide (TMZ) with postoperative radiation. However, tumors invariably become resistant and recur 5 .
In normal adult tissue PIMREG expression is restricted to the thymus, spleen, and colon and, to a lesser extent, the small intestine, ovary, and brain 9,11 . During embryogenesis expression is widely distributed in the embryo, with marked expression in the central nervous system (CNS). Notably, the striking expression of this gene's activity in the neural tube in early stages of development decays abruptly at a speci c stage of development 9 . The particular high expression of Pimreg in the developing CNS and the fact that it is precisely regulated in a temporal-spatial manner suggest that Pimreg could be particularly important for the development of this tissue.
First functional evidence that PIMREG plays a role in cellular processes involved in tumorigenesis was provided when silencing the gene resulted in reduced proliferation, altered cell cycle progression, and decreased expression of cell cycle regulators, as well as impaired migratory and clonogenic abilities of leukemia cells 12 .
Also, a variety of studies have described high levels of PIMREG transcripts in different types of tumors. In an integrated analysis of gene expression pro les of tumors samples from breast cancer patients, PIMREG was identi ed together with BIRC5 and CENPA, as the three-gene signature speci cally upregulated in triple-negative breast cancer. In a broader clinical analysis, high expression of PIMREG correlates with poor survival, regardless the molecular type of breast cancer, drawing attention to its potential value as a prognostic biomarker 13 . In another study, a Pan-cancer analysis based on gene expression pro les of paired tumor and adjacent normal tissues, was carried out on 633 samples across 13 different types of cancer. The study identi ed high expression of PIMREG in a variety of cancer types and its expression correlated with poor patient survival 14 . Furthermore, a possible role of PIMREG as a marker of tumor progression and prognosis has been proposed for breast cancer [13][14][15][16][17] , renal cell carcinoma 18,19 and pancreatic cancer 20 . Moreover, PIMREG was described as one of a four-gene signature predictive as biomarker for paclitaxel response in HER-2 negative breast cancers 21 . Further, parallel studies conducted by different groups showed that PIMREG is involved the migration, metastasis, epithelial-mesenchymal transition and aggressiveness of breast cancer [15][16][17] .
More recently, an important function of PIMREG in regulating in ammation and in ammation-dependent carcinogenesis was demonstrated. Xu and colleagues reported that Pimreg deleted mice exhibited impaired differentiation of Th17 lymphocytes, altered gene expression signature and pro-in ammatory activity characteristic of these cells. As a consequence, Pimreg appears to play a key role in the development of Th17-dependent in ammatory diseases in animal models, such as autoimmune encephalomyelitis, dextran sulfate sodium-induced colitis and colitis-associated carcinogenesis 22 . Pimreg regulates Th17 cell differentiation through interaction with the activated STAT3, retaining this transcription factor in the nucleus and thus contributing to the regulation of its target genes 22 .
The interaction of PIMREG with transcription regulators that shuttle between the nucleus and cytoplasm, and consequent retention of these factors in the nuclear compartment and modulation of its transcriptional activities, seem to be a recurrent mode of action of PIMREG. We have previously shown that through PICALM interaction PIMREG increases the nuclear localization of the leukemic fusion protein PICALM/MTT10 (CALM/F10) and interferes with its ability to activate transcription activate transcription in a dose-dependent manner 11,23 .
A similar mechanism was also described for a PIMREG-mediated regulation of NFkB signaling pathway in breast cancer cells 15 . In that report the authors showed that PIMREG promotes the aggressiveness of breast cancer by constitutively activating the NF-kB pathway. PIMREG interacts with the p65 protein at the same binding site as its inhibitor IkB, thus competing for the interaction site. In cells that express high levels of PIMREG, such as breast cancer cells, interaction with p65 occurs in the nucleus, which leads to stabilization of this transcription factor in this compartment, binding to its responsive elements and consequent activation of its target genes 15 .
Given the emerging signi cance of PIMREG in carcinogenesis and its putative role in the context of nervous system, we sought to investigate the expression and function of PIMREG in gliomas. Here we show that patients with most aggressive and malignant glioma subtype, namely GBM, exhibit the highest levels of PIMREG expression among all cancers analyzed. PIMREG expression correlates with glioma progression and impacts patient overall survival. We performed an extensive expression and function analyzes of PIMREG in GBM cell lines treated with genotoxic agents commonly used for treating GBM patients and assessed for treatment response, proliferation and migration. We show that PIMREG is induced by genotoxic agents and its silencing renders GBM cells more sensitive to TMZ treatment and affects DNA damage signaling. Our data suggest that PIMREG plays a role in DNA damage response and TMZ resistance of GBM cells.

PIMREG is overexpressed in GBM and its expression levels correlates with tumor progression and patient survival
We performed a comparative analysis of PIMREG mRNA levels in paired tumor (n=9475) and adjacent non-tumor tissues (n=773) across 25 different types of cancer from the TCGA dataset. The analysis con rmed that PIMREG is highly expressed across all tumor types in comparison to the normal tissue. Among all tumors analyzed, GBM showed the highest levels of PIMREG expression (Fig. 1A). Interestingly, the median expression of PIMREG in GBM patient samples was 8.932 RSEM (log2) while in the combined cohort that included lower-grade gliomas (GBMLGG), it was 4.99 RSEM (log2), indicating that lower-grade gliomas (LGG) express lower levels of PIMREG in comparison to the more aggressive subtype, GBM. To test this hypothesis, we evaluated the PIMREG expression in the GBMLGG group (n=660) classi ed according to WHO grades (II, III and IV), which in turn re ect the stages of malignancy.
PIMREG lowest expression was observed in patients with grade II gliomas, which is comprised by astrocytoma, oligodendroglioma and oligoastrocytoma samples and exhibit the least malignant phenotype among the three groups examined. Intermediate levels of PIMREG expression were observed in patients with grade III gliomas, which consists of anaplastic oligoastrocytoma and anaplastic astrocytoma and present a more aggressive phenotype compared to grade II gliomas. Finally, highest levels of PIMREG were observed in patients with grade IV gliomas or GBM, which is the group with the most aggressive phenotype (Fig. 1B). Thus, PIMREG expression positively correlates with the degree of malignancy in glioma with the highest expression observed in the subgroup with the most malignant phenotype.
It is well accepted that low grade gliomas carrying mutations in the Isocitrate dehydrogenase 1 (IDH1) gene have a better prognosis in comparison to gliomas with the wild type gene 24 . Thus, we evaluated the PIMREG expression in the LGG patient tumors classi ed according IDH1 status. PIMREG expression was signi cantly lower in tumors carrying IDH1 mutation (IDH1mut) than tumors carrying the wild-type gene (IDH1wt) (Fig. 1C). Similarly, lowers levels of PIMREG were observed in LGG tumors carrying mutations in the homolog of Drosophila capicua gene (CIC) (Fig. S1). CIC mutations are characteristic of oligodendrogliomas that harbor co-deletion of chromosome 1p and 19p 25 , which is also a predictor of improved survival 26 .
In order to analyze the impact of PIMREG expression on patient outcome, we analyzed the survival curve of the patients from GBMLGG-TCGA dataset (n = 660) and three additional cohorts from published studies [27][28][29] . Kaplan-Meier curves showed that high levels of PIMREG negatively impact the survival of glioma patients (Fig. 1D). In the GBMLGG -TCGA cohort (n = 660), the median survival for the group of patients with high PIMREG expression was 742 days, while for the group with low PIMREG expression was 2907 days. Similar pattern was observed for the validation cohorts. In the study by Freije et al.
(GSE4412) 27 , the median survival for glioma patients with high PIMREG expression was 223 days whereas for the patients with low expression it was 726 days, in the study by Gravendeel et al., (GSE16011) 28 the median survival for the group of patients with high PIMREG expression was 11.28 months, and with low expression was 26.52 months. Finally, in the study by Phillips et al., (GSE4271) 29 the median survival for the group of patients with high PIMREG expression was 17.5 months, and for the group with low expression it was 28 months. The data presented demonstrates that PIMREG expression correlates with glioma progression, and patients with high expression of this gene in their tumors have worse prognosis than those with low expression.
PIMREG expression is co-regulated with genes enriched for cell cycle, DNA replication and repair pathways In order to identify cellular processes in which PIMREG may play a role in malignant glioma, we performed a pathway enrichment analysis based on the genes whose expression co-regulated with PIMREG in GBM patient samples (GBM -provisional TCGA n = 604). The genes which presented a positive or negative correlation with PIMREG (Sperman's correlation coe cient ³ 0.51 and ³ -0.51) were selected and submitted to pathway enrichment analysis through the DAVID database (https://david.ncifcrf.gov/home.jsp). The rst 10 key pathways with statistical power were enriched for cell cycle function, DNA replication, pyrimidine metabolism and DNA repair pathways, such as Fanconi anemia pathway, mismatch repair (MMR), homologous recombination (HR) and nucleotide excision repair (NER) (Fig. 2A). The list of genes in which expression covaries with that of PIMREG in each pathway is provided in Table S1.
PIMREG is highly expressed in GBM cell lines Next, we analyzed PIMREG gene and protein expression in GBM, LGG and the non-tumor astrocytes (ACBRI371) cell lines. With the exception of U87MG cells, PIMREG mRNA levels were as high as 2 to 3fold in GBM cell lines (SF188, T98G, U138MG, U251MG and U343MG) when compared to ACBRI371 and LGG cells (Fig. 2B). Similarly, protein levels were higher in the GBM cell lines SF188, T98G, U251MG and U343MG, although lower levels were observed in U87MG and U138MG cells (Fig. 2C).
PIMREG is upregulated in GBM cells treated with genotoxic agents PIMREG protein expression was evaluated in GBM U87MG and T98G cells treated with genotoxic agents (TMZ, cisplatin and hydroxyurea) and ionizing radiation (IR), commonly used to treat cancer. PIMREG is localized to the nucleoplasm and nuclear expression levels increased progressively in U87MG and T98G cells treated with increasing concentrations of TMZ, reaching up to 3-fold expression at the concentration of 400µM of the drug. The dose-dependent increase in PIMREG expression was accompanied by phosphorylation of the histone H2AX at serine 139 (gH2AX), used as a marker for DNA damage (Fig. 3A).
Similar results were observed when U87MG cells were treated with increasing amounts of hydroxyurea and cisplatin (Fig.S2A). PIMREG is known to be a marker for proliferation enriched at G2/M phases of the cell cycle 9 . Thus, we performed a cell cycle analysis in order to investigate whether PIMREG upregulation in TMZ-treated cells was a consequence of altered cell cycle progression. The cell distribution along the different phases of the cell cycle were similar for TMZ-treated and non-treated cells, demonstrating that cell cycle was not affected by the TMZ treatment (Fig. 3B). Hence, PIMREG upregulation was not due to increased proliferation of the GBM treated cells nor its arrest at G2 phase.
Exposure of U87MG and T98G to increasing doses of IR also induced accumulation of PIMREG protein, which reached up to 1.7-fold at higher exposure doses (3-7 Gy) in both cell lines (Fig. 3C). Cell cycle analysis demonstrated that IR treatment resulted G2/M phase arrest in both cell lines (Fig. 3D).
In addition, we evaluated the uorescence intensity of PIMREG protein in U87MG cells before and after TMZ treatment, together with the expression of Cyclin A, a cell cycle regulator whose expression during S/G2 phases of the cell cycle mirrors that of PIMREG 9 . The immuno uorescence analysis con rmed PIMREG upregulation in TMZ-treated cells. Moreover, PIMREG protein increase was restricted to Cyclin A positive cells, whereas cells that do not have Cyclin A labeling showed no upregulation of PIMREG. Of note, TMZ treatment did not alter the levels of Cyclin A protein (Fig. 3E). These results indicate that PIMREG is upregulated in TMZ-treated cells restricted to S/G2 phases of the cell cycle. Furthermore, marked upregulation of PIMREG was observed in U87MG cells treated with cisplatin and hydroxyurea. In these cells, the enhanced PIMREG expression was observed throughout the nucleoplasm and did not assemble at DNA damage foci, as assessed by γH2AX staining (Fig. S2B).

PIMREG knockdown sensitizes GBM cells to TMZ treatment
To better understand the PIMREG role in GBM cells, we generated U87MG and T98G cells carrying constructs for PIMREG knockdown and overexpression. The sequence named as miPIMREG#1 showed 74% and 91% reduction of the PIMREG protein in U87MG and T98G cells, respectively. The miPIMREG#2 sequence was not effective to deplete PIMREG in either cell lines and was therefore used as negative control in the experiments to follow. The overexpression vector (PIMREG) led to an increase of PIMREG protein levels by 634-and 317-fold in U87MG and T98G cells, respectively ( Fig. 4A and Fig. S3A). Sensitivity towards TMZ treatment was assessed in PIMREG depleted and overexpressing U87MG and T98G cells.
Clonogenic survival assay demonstrated that GBM cells depleted of PIMREG (miPIMREG#1) formed less colonies than control cells at different concentrations of TMZ treatment (Fig. 4B). The PIMREG overexpression did not impact TMZ sensitivity in either cell lines at the indicated concentrations of treatment (Fig. S4). These results indicate that PIMREG silencing renders GBM cells more responsive to TMZ treatment.

PIMREG knockdown impairs ATM activation and ATR signaling in TMZ-treated GBM cells
In order to investigate a possible relationship between PIMREG and DNA damage response, we evaluated the initial DNA damage signaling cascade in PIMREG depleted and overexpressing U87MG and T98G cells treated with TMZ.
DNA damage response was assessed by Western blotting analysis using antibodies against the activated ATM (phospho-ATM-S1981) and ATR (phospho-ATR-S428) apical sensor kinases of DNA damage [reviewed in 42] and the replication protein A subunit 32 (phospho-RPA32-S33). Phosphorylation of ATM (S1981) was markedly impaired in PIMREG depleted U87MG and T98G cells subjected to TMZ treatment. The reduction of phosphorylation levels was 73% and 53% in the U87MG and T98G cells, respectively, when compared to miLUC control cells. PIMREG overexpression did not impact ATM phosphorylation at the presented condition (Fig. 5A). Additionally, phosphorylation of ATR was not altered by PIMREG depletion or overexpression (Fig. 5B). However, phosphorylation of ATR target RPA32(S33), was reduced in PIMREG depleted U87MG cells subjected to TMZ treatment (Fig. 5C). These results demonstrate that PIMREG knockdown impairs ATM activation as well as ATR-dependent downstream signaling in response to TMZ-induced DNA damage.

TMZ-mediated upregulation of PIMREG is dependent on ATR activity
Next, we analyzed the ability of TMZ treatment to induce PIMREG expression in U87MG cells impaired of DNA damage response pathways. As previously demonstrated, U87MG cells pro cient for both ATM and ATR pathways exhibited increased PIMREG expression upon TMZ treatment in comparison to non-treated parental cells (Fig. 3A). In contrast, TMZ treatment was unable to upregulate PIMREG expression in U87MG cells inhibited for ATR pathway, where PIMREG levels remained comparable to non-treated cells (Fig. 5D). Also, ATM inhibition slightly enhanced the PIMREG upregulation elicited by TMZ treatment. Although this difference was not statistically signi cant, it was a consistently observed. These results demonstrated that PIMREG is a downstream target of ATR regulation and further con rmed a PIMREG/ATM regulatory axis.

PIMREG knockdown reduces proliferation of GBM cells
We further investigated PIMREG function on proliferation of GMB cells. PIMREG silenced or overexpressing GBM cells were assessed for cell growth on day 0, 1, 3, 5, 7 and 9 of culture with FBSdepleted medium. Cell number was markedly reduced in PIMREG silenced T98G cells as detected after 5 days of culture, reaching 40% the number of the miLUC control cells after 9 days of growth (Fig. 6A). Hampered cell cycle progression was further con rmed by the decreased expression of Cyclin A and Cyclin B in PIMREG silenced T98G cells (Fig. 6B). Proliferation rate was not affected in T98G cells ectopically expressing PIMREG. Likewise, U87MG cell proliferation was not affected upon modulation of PIMREG expression (Fig. S5).

PIMREG knockdown reduces migration of GBM cells
We next investigated the impact of PIMREG expression on migration of GBM cells. Cell migration was assessed in PIMREG silenced and overexpressing cell by wound healing and transwell migration assays.
The wound healing assay demonstrated that PIMREG silenced T98G cells exhibited decreased rate of wound closure of 25% and 36% at 72 and 144 hours after the scratch, respectively. In control cells the wound was fully closed at 144 hours after scratching, whereas PIMREG silenced cells exhibited 50% of the wound unrepaired at this time point (Fig. 7A). Annexin V / PI analysis of PIMREG silenced and control T98G cells at 144 hours after scratch showed no differences in viability or cell death (Fig. S6A). Thus, the decreased rate of wound healing by PIMREG silenced T98G cells was not due to cell death but rather impaired migration of these cells.
Next, we analyzed the impact of PIMREG expression on migration towards a chemotaxis stimulus. In transwell migration assays PIMREG depleted U87MG and T98G cells migrated 26% and 29% less than miLUC control cells (Fig. 7B).
PIMREG ectopic expression did not impact migration rate of U87MG and T98G cells in the presented condition in both assays ( Fig. S6B and S6C).

Discussion
In order to explore PIMREG function in primary brain tumors, such as gliomas, we performed an extensive analysis of PIMREG expression in tumors samples of glioma patients from the TCGA dataset and three additional independent cohorts, assessed the effects of PIMREG silencing and overexpression on the sensitivity of GBM cells to TMZ treatment and explored possible functions of PIMREG in gliomagenesis and treatment response.
PIMREG is a marker for proliferation 9 , and in keeping with this activity, it was found highly expressed in tumor samples compared to their respective normal tissue counterpart. GBM exhibited the highest levels of PIMREG expression among the 25 cancer types examined. It was previously shown that Pimreg is highly expressed in the CNS of developing embryos 9 , suggesting a particular role of this protein in the development of normal CNS. In line with this idea is the observation that the glioma-associated oncogene homolog 1 (GLI1) expression was down-regulated upon PIMREG silencing 31 . GLI1 is the positive effector of the Sonic-hedgehog signaling pathway, which is a key pathway in developing CNS 32 . Moreover, Gli1 plays an essential role in GBM 33 .
Expression analysis on patient samples showed that PIMREG mRNA levels correlate with glioma progression. Lower PIMREG transcript levels were consistently observed in well-de ned subgroups characterized by favorable outcome, such as the mutated IDH and CIC subgroups. Moreover, survival curves dichotomized the group of patients with low versus high levels of PIMREG expression, the second exhibiting the shorter overall survival in all cohorts examined. These results suggest that elevated PIMREG expression could be a potential biomarker for tumor progression and predictive of patient outcome.
Pathway enrichment analysis in GBM patient samples revealed a strong association between PIMREG expression and that of genes enriched for cell cycle function, DNA replication and DNA repair pathways.
The association of PIMREG and genes enriched for cell cycle function was foreseeable since PIMREG functions in proliferation 9,34,35 , cell cycle control 31 and metaphase-to-anaphase transition 10 were already reported. However, its putative involvement in a variety of DNA repair pathways was unexpected, and has not been previously described.
DNA repair pathways represent essential mechanisms of GBM cell resistance towards treatment and directly impact clinical outcome 36 . Thus, we explored the putative role of PIMREG in these processes and the possibility of PIMREG contributing to boost malignant phenotype in gliomas.
Treatment of GBM cells with various DNA damage-inducing agents triggered immediate increase of PIMREG protein, as cells accumulated DNA lesions. For TMZ-treated cells, the PIMREG protein accumulated in those cells undergoing S and G2 phases of the cell cycle. The accumulation was not a consequence of cell arrest in the phases where PIMREG is known to be more abundant, such as the S, G2/M phases of the cell cycle 9,10 , but most likely because PIMREG was required to help the cells to cope with the DNA insults elicited by the treatment. In contrast, IR treatment elicited cell arrest and thus, PIMREG accumulation was more likely to be due to the increased number of cells in the G2 phase at the indicated dose.
GBM cells depleted of PIMREG were more sensitive to TMZ treatment than control cells, demonstrating that higher levels of PIMREG are necessary for the cells to better cope with the insults elicited by the TMZ treatment. Intriguingly, PIMREG overexpression was not able to increase the resistance of the GBM cells against TMZ. Since the GBM cell lines express a markedly high amount of PIMREG, it is likely that overexpression of exogenous PIMREG would not make a physiological appreciable difference as observed in PIMREG depleted cells at the indicated TMZ concentration.
Importantly, TMZ-treated GBM cells depleted of PIMREG presented impaired activation of the apical DNA damage response kinase ATM but not ATR. Although activation of ATR was not affected, the phosphorylation of its downstream target RPA32(S33) was reduced. RPA32-S33 is exclusively phosphorylated by ATR in response to a variety of replicative stress insults, necessary to activate S-phase checkpoint 37-39 , Thus, PIMREG may act downstream of ATR as an ATR-dependent signaling response to TMZ-induced DNA damage. Supporting this idea, GBM cells inhibited for ATR pathway were unable to upregulate PIMREG following TMZ treatment, whereas ATM inhibition slightly enhanced the PIMREG upregulation. These ndings, not only demonstrate that PIMREG participate in the cellular response to DNA damage but also suggest the protein is a novel player in the signaling orchestrated by the sensor kinases ATM and ATR. We propose that PIMREG may act upstream of ATM and downstream of ATR, where it may facilitate ATR-dependent RPA32 (S33) phosphorylation.
Classically, ATM is master regulator of cellular responses to DSBs whereas ATR is responsible for regulating DNA replication stress response. Also, both ATM and ATR contribute to establishment and maintenance of the intra-S and G2/M checkpoints and there is considerable crosstalk between their pathways 30 . An example of functional interaction between these enzymes, is ATM-dependent DNA-end resection provides the RPA-ssDNA signal for ATR recruitment and activation 40 . Also, ATR function upstream of ATM in the response to UV or replication fork stall 41 . Whether PIMREG participates in separate ATM-and ATR-mediated signaling pathways or if it takes part in the crosstalk between these pathways remains to be further investigated. Of note, the serine residue at position 137 of PIMREG (S137) lies within a consensus site for ATM/ATR phosphorylation 30 . Indeed, the S137 of PIMREG was identi ed as a phosphosite in 3 different proteomic studies enriched for phospho-ATM/ATR substrates (PhosphoSitePlus® database (http://www.phosphosite.org). These ndings further support PIMREG involvement in the ATM-and ATR-mediated signaling pathways.
We did not assess the precise nature of the damage caused by the treatments used in this study. Whether it caused replication stress with stalled replication forks, ssDNA, or dsDNA damage, and which of these mechanisms were responsible for causing PIMREG upregulation is not known at the present moment.
It is well known that TMZ methylates DNA at different positions and produces adducts to DNA bases. The most frequent lesions (N3-meG and N3-meA) are reversed by base excision repair (BER), whereas the less frequent and more cytotoxic O6-meG lesions are reversed by the direct repair mechanism. When removal of the O6-meG lesion is not effective, the presence of these alkylated bases causes mispairing during replication and consequent activation of the MMR 42 . This mechanism, in turn, does not restore O6-meG lesions, which can lead to the induction of single-strand breaks (SSBs) that, during DNA replication, are converted into double-strand breaks (DSBs) 36 . ATR is activated upon the presence of DNA lesions and replicative stress caused by TMZ treatment. In the proposed model, unsolved lesions and persistent stress would in turn lead to secondary lesions, such as ssDNA breaks and DSBs and consequently ATM activation 43 . It is conceivable to think that activated ATR regulates PIMREG, which in turn enhances ATR-downstream signaling, such as RPA phosphorylation and ATM activation.
We demonstrated that shRNA-mediated PIMEG silencing reduced cell proliferation up to 40% in the T98G GBM cells used in this study. Correspondingly, the expression of the cell cycle regulators cyclin A and cyclin B1 were also diminished in these cells. These results are consistent with our previous report in leukemia cells 31 , although in that study the reduction of proliferation upon PIMREG silencing was much milder compared to the observation using GBM cells in this study. Interestingly, depletion of PIMREG did not alter the proliferation of the other GBM cells analyzed (U87MG), indicating complex regulation must operate to control proliferative PIMREG function in other cellular contexts.
Given the PIMREG function on proliferation and cell-cycle control 9,10,31,34,35 , together with the ndings presented in this study that PIMREG participates in the DNA damage response with ATM and ATR, it is tempting to speculate that PIMREG may play a role in the ATR-and ATM-dependent checkpoint pathways to block cell cycle progression.
Finally, we demonstrated by two independent methods that PIMREG silencing reduces the migratory ability of GBM cells, con rming what was previous described for leukemia cells 31 . However, it is intriguing that PIMREG overexpression did not impact proliferation, ATM activation or cellular migration. One plausible explanation could be that the overexpression of the ectopic protein (634-and 317-fold over the endogenous expression in T98G and U87MG cells, respectively), would augment the levels of the protein beyond a threshold in which experimental variation would no longer be detected by the cell nor elicit a response. Further experiments assessing these questions in a more physiological context should be addressed.
Collectively, our data demonstrate for the rst time that PIMREG plays a role in DNA damage response and TMZ resistance of GBM cells and further support the PIMREG role in the cellular processes involved in tumorigenesis.

Materials And Methods
Public data analysis PIMREG mRNA expression data in primary solid tumors (TP) and adjacent normal tissues (NT) were downloaded from The Cancer Genome Atlas (TCGA) database using the Firebrowse portal 44 . Glioma patient clinical information, RNAseq-based mRNA expression and mutation data were downloaded from TCGA using cBioPortal for Cancer Genomics 45 . The datasets used for the analysis were Glioblastoma Multiform (TCGA, Firehose Legacy, previously known as provisional 604 samples) and Brain Lower Grade Glioma (TCGA, Firehose Legacy, previously known as provisional 530 samples) downloaded as October 2017 and TCGA-GBMLGG merged cohort 6 . PIMREG expression was analyzed in glioma patients strati ed according to WHO grades (II, III and IV) and mutational status of recurrently affected genes. For survival analysis, expression data was retrieved through SurvExpress platform 46 50 . A fragment with the miRNA-embedded siRNA sequences (miPIMREG#1: 5'tgctgaatagaaggctcaaggtcaaggttttggccactgactgaccttgacctagccttctatt-3' and miPIMREG#2: 5'tgctgataaataggtacccgtgagccgttttggccactgactgacggctcacgtacctatttat-3') and the eGFP ORF were excised from the pMIRTOP plasmids with Eco47II and BglII restriction enzymes and the insert was subcloned into the BglII and SwaI digested pRTS-1 (svp) vector. pRTS constructs expressing luciferase ORF (pRTS-LUC) and the miRNA-embedded siRNA targeting luciferase (pRTS-miLUC), used as controls were previously described 50 .
The generated pRTS-PIMREG construct contains the hygromycin B resistance gene (hyg), whereas the pRTS-miPIMREG#1 and -miPIMREG#2 contains the puromycin gene for selection in mammalian cells.
For a detailed description of the constructs contact the authors.

Apoptosis assays
For apoptosis evaluation, cells were seeded in 12-well plates at density of 5x10 4 cells/well. After 24 hours of incubation under normal culture condition, FBS was removed and cells were incubated further for 144 hours. After that, cells were washed in ice-cold PBS and stained with annexin V for 15 minutes at RT. PI was added (2 ng/µL) prior to the FACS analysis, performed using FACSCanto and FACSDIVa software (BD Biosciences). Statistical analyzes were performed with the One-way ANOVA test, using the GraphPad Prism 4.0 software.

Clonogenic survival assay
Stably transfected U87MG and T98G cells expressing of the ectopic gene (PIMREG and LUC) or miRNAembedded siRNA sequences (miPIMREG#1, miPIMREG#2 and miLUC) were seeded in 6-well plates at density of 500-2500 cells/well and treated with TMZ. The drug was removed and cells were further incubated for 9 days under normal culture conditions. After that period, cells were xed with 4% PFA, stained with 0.5% crystal violet and the number of colonies formed were counted. Statistical analyzes were performed with the One-way ANOVA test, using the GraphPad Prism 4.0 software.

Migration assays
Migration of U87MG and T98G cells expressing of the ectopic gene (PIMREG and LUC) or miRNAembedded siRNA sequences (miPIMREG#1, miPIMREG#2 and miLUC) was assessed by two independent methods.
For the wound healing assay, cells were grown to con uence and wounded by making a scratch with a sterile pipette tip. Images were acquired at 0, 72 and 144 hours following the scratch using an inverted bright eld microscope. ImageJ v1.38 software was used to calculate the percentage of wound closure determined by reduction of the distance between the wound edges from time 0 set as 100%.
Statistical analyzes were performed with the One-way ANOVA test, using the GraphPad Prism 4.0 software.
Cells were seeded above the lter at a density of 3x10 4 cells/well in 200 μl of serum-free medium and the lower compartment was lled with and 500 μL of 10% FBS containing medium. After 24 hours, the cells which migrated through the lter and reached the lower compartment were xed with 4% PFA and stained with crystal violet 0.1%. The migrated cells were counted using an inverted microscope. Statistical analyzes were performed with the One-way ANOVA test, using the GraphPad Prism 4.0 software.   The TMZ concentration and IR dose used are indicated. Membranes were blotted with PIMREG, phospho-H2AX(S139) and β-Actin antibodies. Representative blots are shown (left panel). The bar graphs represent densitometric measurement of the bands normalized by β-Actin and relative to untreated control cells set as 1 (right panel). Results are shown as mean ± SD of three independent experiment.

Declarations
One-way ANOVA test. (B;D) Cell cycle analysis of U87MG and T98G cells treated for 24h with TMZ (400 μM) or 24h after exposure to ionizing radiation (4 Gy). DNA content was analyzed by ow cytometer.
Results are shown as mean ± SD of three independent experiments. Student's t-test. (e) Nuclear expression of PIMREG and Cyclin A proteins in U87MG cells treated with TMZ (400 μM) for 24h and analyzed by immuno uorescence microscopy. Fluorescence intensity was quanti ed by ImageJ v1. 38.
Values plotted on the graphs (right panels) represent the PIMREG (upper panel) and Cyclin A (lower panel) uorescence intensity of TMZ non-treated and treated cells. The uorescence intensity of PIMREG was compared between cells with Cyclin A positive and negative staining (upper graph). Horizontal lines represent the median. 50 cells were evaluated in each condition. Pictures were acquired with 63X magni cation. Student's t-test. TMZ: temozolomide. IR: ionizing radiation. NT: non-treated. AU: Arbitrary Units. (*) p-value < 0.05, (**) p-value < 0.001, and (***) p-value < 0.0001.