c-Myc–miR-29c–REV3L signalling pathway drives the acquisition of temozolomide resistance in glioblastoma

Abstract

See Happold and Weller (doi:10.1093/awv301) for a scientific commentary on this article.

Resistance to temozolomide poses a major clinical challenge in glioblastoma multiforme treatment, and the mechanisms underlying the development of temozolomide resistance remain poorly understood. Enhanced DNA repair and mutagenesis can allow tumour cells to survive, contributing to resistance and tumour recurrence. Here, using recurrent temozolomide-refractory glioblastoma specimens, temozolomide-resistant cells, and resistant-xenograft models, we report that loss of miR-29c via c-Myc drives the acquisition of temozolomide resistance through enhancement of REV3L-mediated DNA repair and mutagenesis in glioblastoma. Importantly, disruption of c-Myc/miR-29c/REV3L signalling may have dual anticancer effects, sensitizing the resistant tumours to therapy as well as preventing the emergence of acquired temozolomide resistance. Our findings suggest a rationale for targeting the c-Myc/miR-29c/REV3L signalling pathway as a promising therapeutic approach for glioblastoma, even in recurrent, treatment-refractory settings.

See Happold and Weller (doi:10.1093/awv301) for a scientific commentary on this article.

Resistance to temozolomide (TMZ) poses a major clinical challenge in glioblastoma multiforme treatment. Using recurrent TMZ-refractory glioblastoma specimens, TMZ-resistant cells, and resistant-xenograft models, Luo et al. report that loss of miR-29c drives the acquisition of TMZ resistance through enhancement of REV3L-mediated DNA repair and mutagenesis.

Open in new tabDownload slide

Introduction

Glioblastoma multiforme (GBM), the most common form of primary brain tumour in adults, has the highest mortality rate among all brain malignancies. This is primarily owing to its persistent tumour growth and highly aggressive invasion, as well as resistance to antitumour drug and ionizing radiation (Stupp et al., 2009). Temozolomide (TMZ) is an alkylating agent currently used as first-line therapy for the treatment of GBM (Plowman et al., 1994). However, a major impediment to effective treatment is acquired resistance to TMZ. Although considerable attention has been focused on several key pathways that have been implicated in TMZ resistance, including elevated expression of the repair protein O-6-methylguanine-DNA methyltransferase (encoded by MGMT), a deficiency in the mismatch repair pathway or an active base excision repair pathway (Trivedi et al., 2005; Happold et al., 2012), there is an urgent need to advance our understanding of TMZ resistance and to identify novel mechanisms, biomarkers and therapeutic targets for predicting treatment response or preventing the emergence of therapy-resistant tumours.

MicroRNAs (miRNAs) are a series of small, highly conserved, non-coding RNA molecules 18–25 nucleotides in length, which usually repress transcription or induce mRNA degradation by binding to complementary sequences in the three prime untranslated regions of their target mRNA molecules. miRNAs are involved in regulating diverse biological processes, including cell development, differentiation, proliferation, and apoptosis (He and Hannon, 2004). Accumulating evidence suggests that miRNAs play central roles in the establishment, progression and recurrence of human cancers through modulating the expression of critical genes and signalling networks involved in tumorigenesis and downstream malignant processes (Lu et al., 2005; Calin and Croce, 2006). miRNA-29c (miR-29c) is a member of the miR-29 family located on chromosome 1q23. miR-29c, attributing predominantly tumour-suppressing properties, has been shown to be downregulated in multiple types of cancers, including gastric cancer (Han et al., 2015), hepatocellular carcinoma (Bae et al., 2014), colorectal cancer (Zhang et al., 2014a), chronic lymphocytic leukaemia (Stamatopoulos et al., 2009), lung cancer (Wang et al., 2013a), breast cancer (Nygren et al., 2014), bladder cancer (Fan et al., 2014), nasopharyngeal carcinoma (Sengupta et al., 2008), oesophageal cancer (Ding et al., 2011), melanoma (Nguyen et al., 2011) and malignant pleural mesothelioma (Pass et al., 2010). Recently, it has been reported that miR-29c is required for normal brain development and modulates neurite growth (Zou et al., 2014), and deregulation of miR-29c expression has been demonstrated to be associated with human gliomas by affecting distinct malignant biological behaviours, including growth, invasion and angiogenesis (Fan et al., 2013; Wang et al., 2013b). Moreover, miR-29c has been verified to be involved in mediating drug and ionizing radiation resistance of cancer cells (Zhang et al., 2013; Yu et al., 2014).

Recent studies have indicated that activation of the DNA strand break repair pathway plays a possible role in the tolerance to TMZ (Yoshimoto et al., 2012). It is now well recognized that translesion synthesis and homologous recombination, two critical post-replication DNA repair pathways, account for cell survival after DNA strand break damage (Lange et al., 2011; Curtin, 2012). Translesion synthesis is a DNA damage tolerance process mediated by specialized DNA polymerases to bypass damaged genomic DNA. REV3L is a major translesion synthesis DNA polymerase that is involved in spontaneous and DNA damage-induced mutations during the process of translesional replication (Gibbs et al., 1998). REV3L has also been implicated in promoting homologous recombination in DNA strand break repair (Sharma et al., 2012). It has been shown that overexpression of REV3L results in strongly enhancing mutagenesis (Xiao et al., 1998; Van Sloun et al., 2002; Sonoda et al., 2003; Okada et al., 2005), which may contribute to malignant transformation, disease progression and drug resistance. Conversely, inactivation of REV3L expression increases the sensitivity of human cancer cells to a variety of DNA-damaging agents and reduces the formation of resistant cells (Knobel et al., 2011). Recently, our group has demonstrated that REV3L is significantly upregulated in human gliomas (Wang et al., 2009). Moreover, Roos et al. (2009) reported that REV3L overexpression may contribute to TMZ resistance in human gliomas. However, it is still unclear how REV3L maintains its oncogenic and chemoresistant properties in gliomas.

c-Myc (encoded by MYC) is a transcriptional regulator that controls various cellular processes, including cell growth and proliferation, differentiation and programmed cell death (Dang et al., 2006). Its abnormal expression was detected in many types of human cancers (Dang, 2012). It has been proposed that c-Myc overexpression coordinates the expression of a multitude of genes that could potentially contribute to its neoplastic properties and silencing of c-Myc has been proved to be an effective therapeutic potential for eradicating malignancies (van Riggelen et al., 2010; Li et al., 2014b). Indeed, recent studies reported that c-Myc is an essential driver of proliferation and correlates with advanced tumour stage and poor prognosis for the patients with GBM (Orian et al., 1992; Herms et al., 1999; Annibali et al., 2014). Moreover, c-Myc was one of the cytogenetic characteristics in the drug-resistant cancer cells, whereas inhibition of c-Myc overcomes resistance to several chemotherapeutic drugs (Abaza et al., 2008; Li et al., 2014a; Nanbakhsh et al., 2014; Pan et al., 2014; Xia et al., 2015; Xie et al., 2014). However, it was unclear whether c-Myc is responsible for TMZ resistance in GBM.

In this study, we identify the c-Myc/miR-29c/REV3L signalling pathway as a master regulator of TMZ resistance, and suggest that disruption of the c-Myc/miR-29c/REV3L signalling pathway may have dual anticancer effects, sensitizing resistant tumours to TMZ as well as preventing the emergence of acquired TMZ resistance.

Materials and methods

See Supplementary material for details on RNA extraction and quantitative reverse transcription polymerase chain reaction (qRT-PCR); microarray hybridization, data processing and vitalization; immunofluorescent staining; fluorescent in situ hybridization; and transfection and stable cell establishment.

The cell viability assay and colony formation assay; and frequency measurement of TMZ-induced mutagenesis at HPRT were performed as described in the previous work (Wang et al., 2009); immunoprecipitation, western blotting, and RNA isolation from RNA-induced silencing complex (RISC) were performed as described in Wang et al. (2014). See Supplementary material for details.

For the homologous recombination DNA repair assay, a synthetic homologous recombination repair substrate DR-GFP system was developed and kindly provided by Dr Maria Jasin from Memorial Sloan-Kettering Cancer Center (New York, NY). See Supplementary material for details.

Cell lines and patient tissue specimens

Cell lines and patient samples information are detailed in the online Supplementary material. This study was approved by the institutional review board and the ethics committee of Nanjing Medical University and Harbin Medical University, and written informed consent was obtained from all participants.

Flow cytometric analysis

Flow cytometry of apoptosis was performed as described previously (Zhang et al., 2014b). For the quantification of γ-histone 2AX (γ-H2AX, encoded by H2AFX), cells were harvested and treated for flow cytometry using Phospho-Histone H2AX (Ser139) (20E3) rabbit monoclonal antibody (Alexa Fluor® 488 Conjugate) (Cell Signaling) as per the manufacturer’s instructions.

Terminal deoxynucleotidyl transferase dUTP nick end labelling analysis

Terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) apoptosis detection kit (Millipore) was used according to the manufacturer’s instructions.

Luciferase reporter assay

The wild-type and mutated putative miR-29c target on REV3L 3’-UTR were cloned into pGL3-control luciferase reporter plasmid (Invitrogen). Firefly and Renilla luciferase signals were determined using a Dual-Luciferase Assay Kit (Promega).

Alkaline comet assay

Alkaline comet assay was performed according to the manufacturer’s instructions (Trevigen). See Supplementary material for details.

Tumour xenograft models and therapeutic regimens

See Supplementary material for details on subcutaneous, patient-specific orthotopic xenograft studies and therapeutic regimens. Investigators were blinded to the treatment groups. All animal experiments were conducted with the approval of the Nanjing Medical University Institutional Committee for Animal Research and in conformity with national guidelines for the care and use of laboratory animals.

MRI of orthotopic mouse tumours

Intracranial tumour growth was monitored under in vivo conditions in isoflurane-anaesthetized mice by MRI after inoculation using a Bruker 7.0 T scanner (Bruker BioSpin GmbH) with a 16 cm bore. See Supplementary material for details.

Immunohistochemistry

The immunohistochemistry assay was conducted on GBM samples or nude mouse xenograft tumour tissues to detect and score REV3L, c-Myc, MGMT and cleaved caspase-3 expression using methods described previously (Wang et al., 2010). REV3L antibody for immunohistochemistry was obtained from Sigma-Aldrich. c-Myc and MGMT antibodies for immunohistochemistry were obtained from LifeSpan Biosciences. Cleaved caspase-3 antibody for immunohistochemistry was obtained from Cell Signaling.

Statistical analysis

All experiments were performed in triplicate with means and standard error of the mean or standard deviation subjected to the Student’s t-test for pairwise comparison or ANOVA for multivariate analysis. Kaplan-Meier survival analysis was performed using Graphpad Prism 5 software. A significance level set at P < 0.05 was considered significant for all the tests.

Results

miR-29c is significantly downregulated in TMZ-resistant glioblastoma multiforme specimens and cell lines

To assess the relevance of deregulated miRNAs in TMZ resistance, we conducted a comprehensive microarray analysis to compare miRNA expression profiles in primary cells obtained from matched pairs of primary GBM and recurrent TMZ-refractory GBM from 10 participants. We identified 23 miRNAs to be differentially expressed (fold change >2.0). Among them, 14 miRNAs were downregulated and nine were upregulated in the recurrent GBM samples as compared with that in the primary tumour samples (Fig. 1A). We observed that the largest change was the downregulation of miR-29c, a predominantly tumour-suppressing miRNA, in recurrent cells. Using quantitative reverse transcription-PCR analysis, we confirmed that miR-29c was dramatically downregulated in the recurrent tumour samples versus their corresponding primary tumour tissues (Fig. 1B). Moreover, using fluorescent in situ hybridization, we found that the recurrent tumours exhibited significantly lower miR-29c staining intensity compared with primary tumours (Fig. 1C). Thus, we next sought to explore whether miR-29c played a role in TMZ resistance. To answer this question, we established two TMZ-resistant cell lines using U251 and U87 cells (U251/TMZ-R and U87/TMZ-R). miR-29c was found to be significantly downregulated in the U251/TMZ-R cell lines (4.47-fold) and U87/TMZ-R cell lines (3.81-fold) compared with their respective parental cell lines (Fig. 1D). Furthermore, we examined endogenous miR-29c levels in four GBM cell lines (U251, U87, LN229 and A172) and normal human astrocytes. The results showed that U251, U87 and LN229 cell lines had significantly lower miR-29c levels than those in the normal human astrocytes, whereas no significant difference was observed between A172 cells and normal human astrocytes (Fig. 1E). Together, these results indicate a possible role of miR-29c in the regulation of TMZ resistance.

miR-29c overexpression sensitizes resistant glioblastoma multiforme cells to TMZ in vitro and in vivo

Recent studies reported that miR-29c could inhibit glioma cell growth (Fan et al., 2013; Wang et al., 2013b). Thus, we examined whether miR-29c could affect proliferation of TMZ-resistant GBM cells. Intriguingly, miR-29c overexpression had little effect on U251/TMZ-R or U87/TMZ-R cell proliferation (Supplementary Fig. 1A–C), indicating that miR-29c might exert its function in a context-dependent manner. The observed significant decrease in miR-29c expression in TMZ-resistant GBM specimens and cell lines prompted us to determine whether re-expression of miR-29c may sensitize the resistant GBM cells to TMZ. To test this possibility, we treated miR-29c-transduced U251/TMZ-R cells with TMZ. Indeed, ectopic expression of miR-29c resensitized U251/TMZ-R cells to TMZ treatment, as indicated by decreased colony formation (Fig. 2A). Moreover, flow cytometry, western blot analysis, and TUNEL assays showed that the miR-29c-overexpressing U251/TMZ-R cells exhibited a significant increase in TMZ-induced apoptosis when compared with cells transduced with scramble controls (Fig. 2B–D). Parallel experiments were performed in U87/TMZ-R cells, and similar results were found (Fig. 2A–C and Supplementary Fig. 2). To provide further support to the notion that miR-29c expression is correlated with TMZ resistance, we assessed the effect of miR-29c inhibition on TMZ resistance in A172 cells (TMZ-sensitive). As shown in Fig. 2E–G, repression of miR-29c in A172 cells led to an increase in cell viability and a concomitant decrease in apoptosis after TMZ exposure, indicating that decreased miR-29c may confer resistance to TMZ in GBM cells.

To evaluate the biological effect of miR-29c on TMZ resistance in vivo, we used a cholesterol-conjugated 2’-O-methyl modified miR-29c (Chol-miR-29c), which has suitable pharmacokinetic properties for tumour xenograft studies. U251/TMZ-R cells or their derivatives (luciferase-expressing) were injected subcutaneously into NOD-SCID mice. Tumour-bearing mice were treated with Chol-miR-29c or Chol-miR-NC in the presence or absence of TMZ. Consistent with the in vitro results, Chol-miR-29c alone had little effect on TMZ-resistant tumour growth (Fig. 3A and B). Of note, mice treated with TMZ plus Chol-miR-29c exhibited a significant reduction in tumour growth when compared with mice treated with TMZ plus Chol-miR-NC (Fig. 3A–C). Moreover, TUNEL staining showed that the apoptotic level was higher in tumours treated with TMZ plus Chol-miR-29c than those treated with TMZ plus Chol-miR-NC (Fig. 3D). Furthermore, the sensitization effect of miR-29c was further confirmed by the survival curves, which TMZ plus Chol-miR-29c-treated xenografts displayed significantly increased survival as compared with control xenografts following TMZ treatment. The median survival for TMZ plus Chol-miR-NC-treated groups was 78 days whereas no mice died in TMZ plus Chol-miR-29c-treated groups after 120 days (Fig. 3E). Parallel experiments were repeated in U87/TMZ-R-derived tumour xenograft models, and similar results were obtained (Supplementary Fig. 3A–C). To determine the potential impact of suppressing miR-29c on TMZ resistance in vivo, we inoculated equal numbers of A172 cells (TMZ-sensitive) into nude mice and treated those mice with Chol-anti-miR-29c or Chol-anti-miR-NC in the presense or absence of TMZ. Consistent with the effect on the A172 cells in culture, the mice treated with TMZ plus Chol-anti-miR-29c showed increased tumour burden (Supplementary Fig. 3D and Supplementary Data) and reduced survival (Supplementary Fig. 3F). In sum, these results demonstrate that overexpression of miR-29c resensitizes TMZ-resistant GBM cells to TMZ in vitro and in vivo.

Error-prone post-replication repair protein REV3L is a direct target of miR-29c

Potent effects of miR-29c downregulation in conferring TMZ resistance prompted us to explore the downstream effectors of miR-29c. To identify downstream targets, we integrated mRNA expression profiling with bioinformatics predictions. Using microarray technology, we produced a list of genes that were downregulated in miR-29c-overexpressing U251/TMZ-R and U87/TMZ-R cells. Overlapping this list with the candidates generated by three prediction algorithms, including miRDB, miRanda and Starbase (Clip-seq), we narrowed the miR-29c target candidates down to 42 genes. We further verified the mRNA expression level of each candidate on this 42-gene list by qRT-PCR in U251/TMZ-R and U87/TMZ-R cells overexpressing miR-29c. This round of screening yielded 16 putative mRNA targets that were significantly and consistently downregulated by miR-29c overexpression in both GBM cell lines, which are listed in Supplementary Table 3. To further narrow down the candidates that are most likely to be direct targets of miR-29c, we used RNA-ChIP (chromatin immunoprecipitation) analysis to identify the mRNAs selectively enriched in the Ago2/RISC complex after miR-29c overexpression (Fig. 4A). As an internal positive control, miR-29c incorporation into RISC was significantly increased in miR-29c-overexpressing cells (Fig. 4B). We then examined those candidate genes that do contain miR-29c seed sequences in their 3’-UTR and identified REV3L, an error-prone post-replication repair protein, which is known to have a role in conferring TMZ resistance (Roos et al., 2009), with significantly elevated enrichment in miR-29c-overexpressing cells compared with the vector control group (Fig. 4C). Moreover, we examined several previously identified targets of miR-29c, including BIRC2, CDK6, DNMT3B, GNA13, IGF1R, ITGB1, LRP6, PTP4A and SIRT1, but discovered little or only slight increases in the levels of these mRNAs incorporated into RISC (Fig. 4C and Supplementary Table 4), consistent with previous notion that miRNAs suppress their targets in a context-dependent manner.

Next, we performed western blot analysis to test whether the protein levels of REV3L could be altered by miR-29c. As shown in Fig. 4E and F, both U251/TMZ-R and U87/TMZ-R cells with miR-29c overexpression showed a significant reduction of REV3L protein levels. Conversely, inhibition of miR-29c in A172 cells led to a significant increase in levels of REV3L. To explore further whether REV3L represented a direct target of miR-29c, we performed luciferase activity assays, in which the 3’UTR with wild-type or mutated miR-29c binding sites were embedded downstream of the firefly luciferase (Fig. 4D). Indeed, the luciferase expression was repressed by miR-29c in a consistent and dose-dependent manner in U251/TMZ-R and U87/TMZ-R cells, whereas expression of the luciferase with mutated 3’UTR was not altered significantly (Fig. 4G). Parallel experiments were performed in HEK293T cells, and similar findings were noted (data not shown). Together, these data support the notion that REV3L serves as a direct target of miR-29c.

miR-29c overexpression circumvents TMZ resistance through downregulation of REV3L

Based on the preceding results, which support a critical role for miR-29c in mediating TMZ resistance and the previous findings that REV3L contributes to TMZ resistance (Roos et al., 2009), it appears that the decreased expression of miR-29c would confer resistance to TMZ through modulating REV3L. To investigate whether this is true in GBM patients, we established primary cultured cell lines from the paired primary and recurrent tumour samples (GBM2 and GBM4), both of which exhibited a significant decrease in the expression of miR-29c after tumour recurrence (Fig. 1B and C). We detected the GBM-associated molecular markers, including the methylation status of the MGMT promoter, the mutation status of the IDH 1 and 2, TP53 and K-Ras for the cell line pairs in comparison with the original tumours. All molecules and their mutations were consistent between original tumour and cell line pairs (data not shown). Moreover, we confirmed that both recurrent GBM2 and GBM4 cells have significantly lower miR-29c expression and higher REV3L levels (Supplementary Fig. 4A and Supplementary Data) compared with their corresponding primary cells, whereas overexpression of miR-29c in recurrent GBM2 (lacking MGMT expression) and GBM4 cells resulted in a significant decrease in the levels of REV3L (Supplementary Fig. 4C). We also measured the expression of MGMT, a critical determinant for TMZ resistance, in primary and recurrent GBM4 cells. However, MGMT levels were not affected in GBM4 cells with miR-29c overexpression or control (Supplementary Fig. 4B and Supplementary Data). Furthermore, we analysed the response of recurrent GBM2 and GBM4 cells following exposure to TMZ and discovered that different doses of TMZ had little effect on the colony-forming ability of recurrent GBM cells (Supplementary Fig. 4D), indicating that the recurrent GBM cells exhibited a TMZ-resistant phenotype.

Next, we used intracranial xenograft mice bearing luciferase-expressing recurrent GBM2 cells (lacking MGMT expression) to assess the effect of miR-29c/REV3L axis on chemosensitization. Tumour progression during TMZ treatment was monitored using in vivo bioluminescence imaging and MRI, respectively. Bioluminescence images revealed that no significant differences in the tumour volume between recurrent GBM2 cells transduced with miR-29c and those cells transduced with miR-NC. By contrast, mice bearing miR-29c-transduced recurrent GBM2 cells displayed a significant reduction in signal upon TMZ treatment compared with xenografts transduced with scramble miRNAs (Fig. 5A). T₂-weighted MRI and haematoxylin and eosin staining confirmed this finding that mice bearing recurrent GBM2 cells transduced with miR-29c showed a drastic decrease in tumour volume as compared with control xenografts (Fig. 5B and C). We also examined the expression of REV3L and MGMT by immunohistochemistry and found that tumours overexpressing miR-29c had a significantly lower level of REV3L protein compared with control tumours (Fig. 5D). MGMT expression could not be detected in mice bearing intracranial recurrent GBM2 xenografts, indicating that the effect of miR-29c/REV3L signalling on TMZ sensitization is not attributable to MGMT expression (Fig. 5D). Moreover, miR-29c-overexpressing tumours treated with TMZ showed a higher apoptotic index as demonstrated by active caspase-3 staining (Fig. 5D) and increased TUNEL staining, and significantly lower levels of Ki-67 and CD31 expression compared with control tumours. (Fig. 5E, F and Supplementary Fig. 4E and F). The median survival of mice bearing intracranial scramble-transduced tumour xenografts was 67.5 days, whereas only one miR-29c-transduced xenograft was dead after 120 days (Fig. 5G). We also performed the parallel experiments in recurrent GBM4 cells (with MGMT expression) (Supplementary Fig. 4A). Similarly, a significant reduction in the fluorescent area of tumour foci and tumour volume (Supplementary Fig. 5A and Supplementary Data) and an increase in survival (Supplementary Fig. 5C) in miR-29c-transduced recurrent GBM4 xenografts were observed following TMZ treatment. Together, these data demonstrated that miR-29c overexpression overcomes TMZ resistance through downregulation of REV3L.

Inhibition of REV3L in recurrent glioblastoma multiforme cells with low miR-29c expression reverses TMZ resistance

To further detect the role of REV3L in miR-29c-modulated chemosensitization, we knocked down the expression of REV3L in the recurrent TMZ-refractory GBM2 and GBM4 cells transduced with anti-miR-29c or scramble antisense oligonucleotides (Fig. 6A), and treated those cells with vehicle or TMZ. Compared with scramble-transduced cells, inhibition of REV3L expression remarkably sensitized GBM2 and GBM4 cells expressing anti-miR-29c to TMZ, as demonstrated by decreased colony formation (Fig. 6B) and enhanced apoptosis (Fig. 6C and D). Further, repression of REV3L in recurrent GBM2 cells transduced with anti-miR-29c rendered the resistant cells sensitive to TMZ in vivo, as shown by a significant reduction in luminescence signal (Fig. 6E) and tumour volume (Fig. 6F and G) as well as increased expression of cleaved caspase-3 (Fig. 6H), all of which could recapitulate the sensitization effect of miR-29c overexpression on TMZ-resistant tumours. Histological analysis confirmed that all mice bearing tumours with downregulation of miR-29c showed a significant upregulation of REV3L (Fig. 6H).

To control for cell line heterogeneity and further evaluate the role of miR-29/REV3L signalling on TMZ resistance, we cotransfected U251/TMZ-R or U87/TMZ-R cells with REV3L (lacking the miR-29c target site in the 3’ UTR) and miR-29c mimics. Overexpression of REV3L in miR-29c-transduced U251/TMZ-R or U87/TMZ-R cells antagonized the effect of miR-29c on clonogenic growth and apoptosis upon TMZ treatment (Fig. 2A–D). Moreover, we demonstrated that transfection of A172 cells expressing anti-miR-29c with short hairpin (sh)RNA targeting REV3L abrogated drug resistance induced by miR-29c downregulation (Fig. 2F and G). Collectively, these results suggest that inhibition of REV3L in GBM cells expressing low-level miR-29c reverses TMZ resistance.

miR-29c inhibition, independent of MGMT or the mismatch repair pathway, promotes TMZ resistance

MGMT is a DNA repair protein that repairs the O⁶-methylguanine lesion created by TMZ (Pegg et al., 1995) and is known as a key factor for cellular resistance to TMZ (Sarkaria et al., 2008). Given that MGMT levels were not altered in miR-29c-overexpressing cell lines (Supplementary Fig. 4C), we asked whether miR-29c-induced hypersensitivity is associated with a lack of bypass of O⁶-methylguanine adducts, miR-29c- or scramble-transduced recurrent GBM4 cells was treated with O⁶-benzylguanine (O⁶-BG, a specific MGMT inhibitor). The cells were then exposed to TMZ, and the frequency of apoptosis was determined by flow cytometry 72 h later. Interestingly, there was no difference in the numbers of apoptotic cells in response to TMZ when cells were pretreated with or without O⁶-BG (Supplementary Fig. 6B). Similar results were found in recurrent GBM2 cells (lacking MGMT expression) (Supplementary Fig. 6A). These data indicate that miR-29c downregulation induces TMZ resistance which is independent of MGMT expression or a lack of formation of O⁶-BG adducts.

It has been shown that loss of the mismatch repair pathway plays an important role in conferring resistance to TMZ (Cahill et al., 2007; Yip et al., 2009). To examine whether miR-29c/REV3L-mediated TMZ resistance is associated with mismatch repair deficiency, we examined the effect of MSH6, a key mismatch repair gene, or REV3L or both knockdown on TMZ-induced toxicity in U251 cells (Supplementary Fig. 6C). Consistent with previous findings, MSH6 inhibition could substantially rescue the sensitivity of U251 cells to TMZ. In contrast, MSH6 deficiency could not abrogate the sensitivity of REV3L-depleted cells to TMZ (Supplementary Fig. 6E). We also knocked down MSH6 in miR-29c-transduced U251 cells (Supplementary Fig. 6D). Our results showed that loss of MSH6 in miR-29c-transduced U251 cells failed to recover TMZ resistance (Supplementary Fig. 6F). Together, these results suggest that mismatch repair deficiency is not involved in miR-29c/REV3L-mediated TMZ resistance.

Disrupting miR-29c/REV3L signalling impedes homologous recombination, delays DNA repair and induces chromosomal aberrations

Homologous recombination is a major DNA repair pathway that mediates DNA strand break repair and recovery of stalled replication forks, which has been shown to contribute to TMZ resistance (Liu et al., 2009). Because REV3L has been suggested to participate in homologous recombination repair (Zhang et al., 2007; Sharma et al., 2012), we asked whether miR-29c/REV3L signalling could potentially have an impact on homologous recombination pathway. We measured homologous recombination activity using a GFP-based DNA strand break repair system in U87 cells, which contains an integrated, single-copy homologous recombination reporter with an I-SceI recognition site (U87 DR-GFP cells). U87 DR-GFP cells were transfected with scrambled and miR-29c mimics or siCtrl and siREV3L. siRNAs against BRCA1 and RAD51 were used as positive controls. We found that ectopic expression of miR-29c dramatically reduced homologous recombination efficiency as indicated by the reduced numbers of GFP-positive U87 cells (Fig. 7A and B). Similar effects were observed after knockdown of REV3L in U87 cells (Fig. 7A and B). These results indicate that disrupting miR-29c/REV3L signalling may repress homologous recombination-mediated repair.

The phosphorylated histone family member X (γ-H2AX), an early cellular response to the induction of DNA double strand breaks, could form discrete nuclear foci and facilitate DNA repair following DNA-damaging agents treatment (Kinner et al., 2008). To investigate the potential role of miR-29c in DNA damage and repair process, we assessed the effects of miR-29c on γ-H2AX foci formation upon TMZ treatment. γ-H2AX foci was increased in the scramble-transduced U87 cells at the early time points and was attenuated at 2 h, suggesting that there was a rapid repair of DNA damage in the control cells. In stark contrast, miR-29c-transduced U87 cells displayed slight higher γ-H2AX foci without TMZ and showed a significant increase in γ-H2AX foci at 2 h and even at 12 h in response to TMZ (Fig. 7C and D). Moreover, quantification of γ-H2AX using flow cytometry was performed to confirm this result. Consistently, the active foci of γ-H2AX were significantly higher in miR-29c-transduced U87 cells relative to the control cells from 4 h to 36 h after the treatments (Fig. 7E). Furthermore, western blot analysis showed that U87 cells with miR-29c overexpression had persistently high levels of γ-H2AX at 4 h and 12 h after TMZ treatment compared with the control cells (Fig. 7F). In addition, we found that γ-H2AX levels were dramatically increased in REV3L-depleted cells in both untreated condition and after TMZ exposure (200 µM) at 4 h, 12 h and 24 h relative to the control cells (Fig. 7G). Using an alkaline comet assay, we measured the persistence of double strand breaks after TMZ treatment, as an indicator of unrepaired damaged DNA. U87 cells overexpressing miR-29c had significantly higher residual DNA damage (Fig. 7H) relative to control cells. A significant higher amount of DNA breaks were also observed in REV3L-depleted cells (Fig. 7H). Chromosome aberration analysis showed that either overexpressing miR-29c or silencing REV3L in U87 cells markedly increased the number of chromosomal breakages per metaphase in the absence or presence of TMZ (Fig. 7I). Immunoblot assay confirmed that U87 cells with miR-29c overexpression had lower levels of REV3L protein relative to control cells, whereas the expression of multiple homologous recombination-related proteins, including BRCA1, BRCA2 and RAD51, was not altered (Supplementary Fig. 7). Taken together, these results suggest that miR-29c targeting of REV3L directly impaired homologous recombination-mediated repair, induced a delay in DNA repair and a high frequency of chromosomal aberrations.

miR-29c overexpression prevents TMZ-induced mutagenesis and acquired chemoresistance

Previous studies from our group and others have suggested that REV3L play a key role in the development of drug-induced mutagenesis, and REV3L-dependent mutagenesis can also actively promote chemotherapeutic resistance (Wu et al., 2004; Lin et al., 2006; Wang et al., 2009; Doles et al., 2010). Therefore, we asked whether miR-29c/REV3L signalling is involved in the development of TMZ-induced mutagenesis and acquired resistance in GBM cells, we initially treated miR-29c-transduced U87 cells with TMZ (50 µM) for 1 h and then cultured for 2 weeks. HPRT mutation assay was then performed to assess the frequency of TMZ-induced 6-thioguanine-resistant mutants. As shown in Fig. 7J, miR-29c-transduced U87 cells displayed significant reduced frequencies of 6-thioguanine-resistant mutants induced by TMZ by 4.35-fold compared with control cells. Parallel experiments were performed in REV3L-depleted U87 cells and similar results were observed (4.01-fold decrease) (Fig. 7J). These results suggest that blocking miR-29c/REV3L signalling could inhibit TMZ-induced mutagenesis. We next sought to determine whether miR-29c could prevent the emergence of TMZ resistance in TMZ-treated GBM cells. miR-29c-transduced or control U87 cells were treated with TMZ (33 µM), assayed for cell survival at 48 h, and allowed the cells to recover for an additional 7 days, at which point we initiated a subsequent round of TMZ treatment. Relative drug resistance was determined by normalizing viability measurements to the initial values collected during the first round of treatment. Compared with the control cells, miR-29c-transduced U87 cells showed a reduced resistance profile after repeated TMZ exposure (Fig. 7K). In sum, these data support the notion that activation of miR-29c/REV3L signaling drives TMZ-induced mutagenesis, which might, at least in part, predispose to the acquisition of resistance to TMZ.

c-Myc upregulates REV3L through downregulation of miR-29c

Recent studies reported that c-Myc may play a possible role in modulating TMZ resistance (Pyko et al., 2013), and activation of c-Myc could lead to widespread miRNA repression (Chang et al., 2008; Cairo et al., 2010; Zhang et al., 2012), including miR-29c. Thus, we asked whether c-Myc is involved in mediating miR-29c/REV3L signalling. We knocked down or overexpressed c-Myc in the recurrent GBM2 and GBM4 cells, U251/TMZ-R and U87/TMZ-R cells. Our results showed that the expression levels of miR-29c were significantly decreased in c-Myc-overexpressing cells and increased in c-Myc-depleted cells (Fig. 8A). Meanwhile, overexpression of c-Myc enhanced, while repression of c-Myc reduced REV3L levels in recurrent GBM2 and GBM4 cells (Fig. 8B and C). Of note, the introduction of miR-29c inhibitor reversed the c-Myc shRNA-mediated attenuation of REV3L (Fig. 8C), whereas the enhancement effect of c-Myc on REV3L expression could be abrogated by transfection with the miR-29c mimics (Fig. 8D), indicating that c-Myc-mediated miR-29c downregulation contributes to REV3L protein suppression.

Next, we determined the role of c-Myc in TMZ-induced DNA damage. Similar to the effect of miR-29c overexpression or REV3L depletion, silencing c-Myc significantly decreased the homologous recombination efficiency and increased the number of TMZ-induced γ-H2AX foci (Fig. 8E and F). Moreover, we showed that c-Myc-depleted recurrent GBM2 cells had significantly higher residual DNA damage and a delay in DNA repair in the presence of TMZ, compared with control cells (Fig. 8G). Noticeably, the effects of c-Myc repression on TMZ-induced DNA damage could be abolished upon transfection with miR-29c inhibitor. More importantly, inhibition of c-Myc sensitized the resistant recurrent GBM2 cells to TMZ in culture (Fig. 8H and I) and in vivo orthotopic xenografts (Fig. 8J and K). Using immunohistochemistry analysis, we confirmed that mice bearing tumours with knockdown of c-Myc had a significant decrease in REV3L expression (Fig. 8L). Taken together, these results indicate that c-Myc is a predominant upstream regulator of miR-29c/REV3L signalling-mediated TMZ resistance.

Clinical relevance of c-Myc, miR-29c and REV3L in patients with glioblastoma multiforme

To address the clinical significance of the c-Myc/miR-29c/REV3L interaction, we determined the expression of miR-29c, c-Myc and REV3L in 397 GBM patients using The Cancer Genome Atlas (TCGA) database. We found that the stronger expression of c-Myc and REV3L was significantly associated with lower miR-29c expression (Supplementary Fig. 8A and Supplementary Data). We also evaluated the endogenous expression pattern of miR-29c, c-Myc and REV3L in 138 patients with GBM who received TMZ-based chemotherapy. As shown in Supplementary Fig. 8C and Supplementary Data, tumours with a low level of miR-29c tended to express high levels of c-Myc and REV3L, whereas tumours with a high level of miR-29c tended to express low levels of c-Myc and REV3L. Finally, we assessed whether miR-29c levels are associated with TMZ efficacy. Kaplan-Meier survival analysis revealed that patients with high miR-29c expression displayed a favourable response to TMZ treatment compared with patients with low miR-29c expression (Supplementary Fig. 8E and Supplementary Data) [overall survival (OS), 19 months versus 11 months, homologous recombination, 1.727, 95% confidence interval (CI): 1.011–2.443, P < 0.001; progression-free survival (PFS), 14.6 months versus 7.8 months, homologous recombination, 1.884, 95% CI: 1.168 to 2.6, P < 0.001]. These findings shed light on the clinical significance of c-Myc/miR-29c/REV3L signalling in the regulation of TMZ resistance.

Discussion

TMZ, an oral alkylating agent, is currently used as the standard care for newly diagnosed GBM (Stupp et al., 2005). As yet, there is no consensus on the optimal approach for patients with recurrent GBM. Recently, multicentre, non-randomized phase II clinical trials were performed to evaluate the efficacy and safety of continuous dose-intense TMZ for recurrent GBM (Perry et al., 2010; Norden et al., 2013). The results showed that continuous dose-intense TMZ may be a reasonable option in patients with GBM who experience recurrence after standard upfront therapy. However, the efficacy of TMZ in the treatment of primary and recurrent GBM remains unsatisfactory, because the tumour is either insensitive to TMZ therapy or rapidly develops resistance. Thus, identifying and blocking critical pathways that mediate resistance could positively impact the successful treatment of this disease. Here, our detailed mechanistic in vitro and in vivo experiments, complemented by analysis of clinical data sets, showed that c-Myc/miR-29c/REV3L signalling may be an important regulator of TMZ resistance in human GBMs.

Recent studies have linked the acquisition of chemoresistance to the altered expression of miRNAs (Migliore and Giordano, 2013). In the present study, we identified a set of significantly differentially expressed miRNAs associated with TMZ resistance between paired primary and recurrent GBMs who were insensitive to TMZ treatment. Of all miRNAs that showed altered expression, miR-29c was decreased to the greatest extent. We next evaluated the effect of miR-29c on the sensitivity of established and primary TMZ-resistant GBM cells to TMZ. We found that ectopic restoration of miR-29c could substantially sensitize the resistant GBM cells to TMZ in vitro, in vivo subcutaneous and orthotopic xenograft mouse models without affecting cell proliferation, MGMT and the mismatch repair pathway. In contrast, inhibition of miR-29c attenuated TMZ-induced cell death and conferred a TMZ-resistant phenotype in TMZ-sensitive GBM cells. Moreover, our findings offer preclinical proof of efficacy for miR-29c in a valid mouse model of GBM, in which TMZ-refractory explants expressing high-level miR-29c displayed better overall survival when treated with TMZ compared with explants expressing controls, suggesting that targeting miR-29c could benefit treatment for patients with GBM receiving TMZ.

Enhanced DNA repair can allow damaged or mutated tumour cells to survive, thereby contributing to resistance and tumour recurrence (Turner et al., 2015). Error-prone translesion synthesis, a major post-replication DNA repair mechanism, is known to underlie the mutagenic effects of multiple chemotherapeutic agents and has been proved to play a key role in the development of acquired drug resistance (Doles et al., 2010; Xie et al., 2010). REV3L, a major error-prone polymerase of translesion synthesis, has been recently reported to be significantly overexpressed in human gliomas (Wang et al., 2009), and also associated with homologous recombination pathway (Sharma et al., 2012) and TMZ resistance (Roos et al., 2009). However, the mechanism by which its expression is driven and sustained remains unclear. Our study identified that REV3L is a direct and functional target of miR-29c. Enforced expression of miR-29c in TMZ-resistant GBM cells could reduce REV3L levels, which impaired the homologous recombination process, induced persistent DNA damage and genomic instability, and rendered resistant cells sensitive to TMZ. Conversely, antagonizing miR-29c enhances activation of REV3L levels, leading to increased DNA repair and subsequent resistance to TMZ. Furthermore, re-expression of REV3L lacking its 3’ UTR was sufficient to rescue miR-29c-mediated sensitization to TMZ, deeming REV3L as a critical regulator of miR-29c-mediated phenotypes. Thus, our findings may partially explain the unexpected overexpression of REV3L protein in GBM via miR-29c downregulation. Another important finding from our study is that we discovered that disrupting miR-29c/REV3L signalling resulted in a significant reduced TMZ-induced mutagenesis and prevented the emergence of acquired TMZ resistance. Thus, we propose that activation of miR-29c/REV3L signalling could drive drug-induced mutagenesis, which might contribute to the mutagenic drug resistance acquisition.

Recent studies have suggested that c-Myc has been associated with widespread miRNA repression (Chang et al., 2008; Cairo et al., 2010; Zhang et al., 2012), including miR-29c. Although several studies have documented that c-Myc is associated with the promoters of DNA strand break repair genes and the expression of mismatch repair genes (Luoto et al., 2010), whether inhibition of c-Myc leads to decreased DNA repair capacity and sensitization of tumour cells to chemotherapeutic agents remains elusive. In the current study, we showed that c-Myc, through miR-29c, regulates the expression of REV3L. On c-Myc overexpression, the downregulation of miR-29c and the corresponding induction of REV3L are causally required for accelerated homologous recombination-mediated repair capability and increased TMZ-induced mutation, which subsequently facilitates the acquisition of resistance to TMZ. By contrast, upregulation of miR-29c through c-Myc abrogation results in the inhibition of REV3L, which is able to reverse TMZ resistance in vitro and in vivo. Therefore, our findings revealed that c-Myc overexpression strongly promotes TMZ resistance and mediates activation of translesion synthesis and/or homologous recombination repair pathways via miR-29c/REV3L signalling.

Given that our findings point to c-Myc/miR-29c/REV3L axis as a key regulator of TMZ resistance both in vitro and in vivo models, we asked whether expression of this axis itself might be useful in predicting the response of GBM to TMZ. Indeed, analysis of miR-29c expression in clinical data set demonstrated that miR-29c might function as a predictive marker for patients with GBM who received TMZ. Patients who had high-level miR-29c expression exhibited better survival upon TMZ treatment, whereas patients with low-level miR-29c expression did not benefit from TMZ therapy. These observations suggest that pretreatment evaluation of miR-29c expression could be used to predict the therapeutic response to TMZ for a particular patient. However, our study had several potential challenges. Problems such as in vivo delivery and potential off-target effects, must be resolved before this axis can be used a therapeutic target. The main obstacle for the systemic delivery of miRNAs in GBM therapy is to determine their uptake in GBM cells. Future work is needed to establish novel delivery systems, which could overcome miRNAs degradation by nucleases, renal clearance, failure to cross the blood–brain barrier, ineffective endocytosis by target cells, and activation of the host immune responses. The potential off-target effects of miRNAs therapy are that the same miRNA can generally have diverse effects by targeting multiple mRNAs and regulate different mechanisms. Thus, off-target effects may cause unexpected consequences even when a specific miRNA is targeted. An important approach to reducing miRNA or siRNA-like off-target effects is chemical modification (Jackson et al., 2010). It has been shown that, 2’-O-methyl modification of a single position in the siRNA guide strand was shown to reduce most siRNA off-target silencing. In our study, we used a cholesterol-conjugated 2’-O-methyl-modified miR-29c (Chol-miR-29c) that has suitable pharmacokinetic properties for tumour xenograft studies and confirmed its effectiveness in treating TMZ-refractory tumours. Several recent studies have also implemented seed-region analysis for identification of potential off-target transcripts using genome-wide enrichment of seed sequence matches (Yilmazel et al., 2014). Further research is needed to transform this knowledge into effective therapeutics against GBM and eventually eliminate false positive results attributable to sequence-specific off-target effects associated with miRNA.

In conclusion, this study highlights the importance of c-Myc/miR-29c/REV3L signalling as a master modulator of TMZ resistance. Dysregulation of the c-Myc/miR-29c/REV3L signalling pathway in GBMs indicates an aggressive subgroup developing early resistance to TMZ. We propose that targeting this novel pathway may serve as a promising strategy for therapeutic intervention of GBM, even in recurrent, treatment-refractory settings.

Acknowledgements

We thank Dr Yoshiki Murakumo (Nagoya University Graduate School of Medicine, Nagoya, Japan) for providing the REV3L plasmid used in this study. We thank Dr Maria Jasin (Memorial Sloan-Kettering Cancer Center) for providing DNA HR repair constructs. We also thank Dr Chengyi Hu and Dr He Wang (Harbin Medical University) for the assistance with glioma sample collection.

Funding

This work was supported by the National Natural Science Foundation of China (No. 81201978, No. 81172389, No. 81072078, No. 81200362, No. 81372709, No. 81272792); the Jiangsu Province’s Natural Science Foundation (BK2012483, BK2010580); the Program for Advanced Talents within Six Industries of Jiangsu Province (2012-WSN-019); the National High Technology Research and Development Program of China (863) (2012AA02A508); International Cooperation Program (2012DFA30470); the Jiangsu Province’s Key Provincial Talents Program (RC2011051); the Jiangsu Province’s Key Discipline of Medicine (XK201117); the Jiangsu Provincial Special Program of Medical Science (BL2012028); the Program for Development of Innovative Research Team in the First Affiliated Hospital of Nanjing Medical University; the Provincial Initiative Program for Excellency Disciplines, Jiangsu Province and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

Supplementary material

Supplementary material is available at Brain online.

Abbreviations

Abbreviations
 
• GBM

  glioblastoma multiforme

 
• γ-H2AX

  γ-histone 2AX

 
• TMZ

  temozolomide

References

Author notes