SETD2 and miR-21 as therapeutic targets for NUT midline carcinoma

Background: Nuclear protein in testis (NUT) midline carcinoma (NMC) is a rare and highly aggressive tumor with the bromodomain containing 4 (BRD4)-NUT (NUTM1) gene fusion. BRD4 is a member of the bromodomain and extra-terminal domain (BET) family of proteins, and BET inhibitors have been investigated in NMC clinical trials. However, few targeted therapies are available for NMC, and novel therapeutic targets remain to be determined. We determined the role of two epigenetic regulators as possible therapeutic targets for NMC. Methods: We performed next-generation sequencing (NGS) in NMC cell lines (HCC2429 and Ty82). H3K36me3 expression was studied using western blotting. The ecacy of AZD1775, a WEE1 inhibitor, was evaluated using the MTS and γH2AX assays. We established an NMC cell line that was resistant to BET inhibitors. The sensitivity of the cells resistant to AZD1775 was analyzed using the MTS assay. RNA sequencing was performed to determine miRNA expression levels. TaqMan miRNA assays were used to analyze miR-21 expression. The ecacy of the miR-21 inhibitor was evaluated using the MTS assay. We established a digital PCR (dPCR) assay to detect NUT gene rearrangements to identify patients with NMC. Using NGS, a patient with NMC was identied with the SETD2 mutation.


Page 3/16
Background Nuclear protein in testis (NUT) midline carcinoma (NMC), also referred to as NUT carcinoma, is a rare and highly aggressive tumor with a predilection for the midline structures, affecting both children and adults.
NMC is genetically identi ed by the presence of the NUT gene, also known as the NUT midline carcinoma family member 1 (NUTM1) gene rearrangement [1,2]. The most frequent translocation is observed between the NUT gene on the 15q14 chromosome and the bromodomain containing 4 (BRD4) gene on the 19q13.1 chromosome, which accounts for more than 70% of the rearrangements [3][4][5]; while the other NUT fusion partners include BRD3, nuclear receptor binding SET domain protein 3 (NSD3), zinc nger protein 532 (ZNF532), and other genes [6][7][8][9]. NMC lacks speci c pathological features and can occur in any organ, such as the thorax, head, neck, or other midline organs [10]. Patients with NMC can be misdiagnosed with other malignant tumors due to poor differentiation, unawareness of the disease, and lack of diagnostic tests. The prognosis of patients with NMC is signi cantly poor, with a median survival time of 6-9 months [3,10].
The development of novel treatment strategies for NMC is challenging. Pediatric tumors harboring t (15;19) were rst reported during 1990s [11,12]. In 2003, French and colleagues identi ed the BRD4-NUT fusion gene in poorly differentiated midline carcinoma [2]. Studies published from 2000 onwards revealed that the fusion gene plays an important role in NMC development, promoting increased expression of MYC and other oncogenes [13]. BRD4 is a member of the bromodomain and extra-terminal domain (BET) family of proteins and binds to acetylated lysine in histones [14]. However, NUT shuttles between the nucleus and the cytoplasm and recruits p300 histone acetyltransferase (HAT) to enhance histone acetylation [6,8,15]. Hence, it has been postulated that BRD4-NUT has a predilection for the nucleus where the NUT portion recruits p300 and increases histone acetylation. Then, the fusion protein increases hyperacetylated regions of chromatin up to 2 Mb in size, "megadomains" that drive transcription of the underlying DNA, such as MYC and TP63 [8].
Therefore, targeted therapy studies have been conducted based on the molecular mechanisms underlying aberrant signaling through fusion proteins. BET inhibitors have been investigated, and the rst-in-class BET inhibitor JQ1, which competitively binds to bromodomains, has shown prodifferentiative and antiproliferative effects on NMC in preclinical studies [16]. More recently, novel BET inhibitors have been developed, and a BET inhibitor OTX015 (Birabresib) has shown tumor regression and symptomatic relief in two patients with advanced NMC, achieving an overall survival (OS) of 18 months and 19 months; in a phase Ib trial, it showed a 30% response rate (RR) in patients with NMC [17,18]. In contrast, a phase I/II study demonstrated that another pan-BET inhibitor, GSK525762, showed a 22% RR in patients with NMC [19]. Both inhibitors showed good tolerability in clinical trials; however, the e cacy of BET inhibitors is limited, and NMC can often develop resistance to them [18,19].
Accordingly, the development of novel treatment strategies, such as combination chemotherapy with BET inhibitors, is required. To date, no other therapeutic targets have been identi ed in NMC. Studies that focus on simultaneous gene mutations and microRNAs (miRNAs) will help identify novel therapeutic targets. In this study, we aimed to investigate the role of two epigenetic regulators, SET domain containing 2 (SETD2) and miR-21, as possible therapeutic targets for NMC. instructions. Brie y, cells were seeded into 96-well plates (3,000-5,000 cells/well) overnight in triplicates; cells were then treated with the indicated concentrations of drugs for 3 days, and the assays were performed.

Methods
Stealth RNAi siRNA (Thermo Fischer Scienti c, USA) was used for NUTM1 knockdown, NUTM1-siRNAs (HSS138007, HSS138008, HSS138009) and Stealth RNAi siRNA Negative Control (#12935300, Thermo Fischer Scienti c). Lipofectamine RNAiMAX transfection reagent and Opti-MEM I reduced serum medium (Thermo Fischer Scienti c) were used for siRNA transfection. The cells were treated with 50 pmol of siRNA for the indicated times, and the following assays were performed. miRNA function analysis was performed using miRCURY LNA miRNA mimics and inhibitors (QIAGEN, Japan), hsa-miR-21-5p inhibitor (YI04100689), has-miR-21-5p mimic (YM00473093), miRNA inhibitor controls, and miRNA mimic controls. Lipofectamine 3000 (Thermo Fischer Scienti c) and Opti-MEM I reduced serum medium were used for miRNA transfection. In 96-well plates, the cells were treated with 0.66 pmol of miRNA mimics or 5 pmol of miRNA inhibitors for 72 h, and the assays were performed.

Next-generation sequencing
Genomic DNA was extracted from the cell lines using the Blood & Cell Culture DNA mini kit (QIAGEN, Japan) and from patient-derived tumors using the QIAamp DNA FFPE Tissue kit (QIAGEN, Japan). DNA was quanti ed using the Qubit 2.0 uorometer with the Qubit dsDNA HS assay kit (Thermo Fisher Scienti c). We used 50 ng of DNA from cell lines and 10 ng of DNA from tissue samples for PCR ampli cation using the Ion AmpliSeq Library kit 2.0, and Ion AmpliSeq Comprehensive Cancer Panel (Thermo Fischer Scienti c). The Ion Express Barcode Adapters (Thermo Fischer Scienti c) were ligated to the PCR products for barcoding the tissue samples. AMPure XP beads (Beckman) were used for PCR product puri cation. The libraries were sequenced using an Ion PGM System (Thermo Fischer Scienti c). DNA sequencing data were obtained using the Torrent Suite ver. 4.0 software (Thermo Fischer Scienti c). The variant caller ver. 4.0. is called variants. The reads were aligned with the GRCh38 reference genome.
We used the Ion Reporter ver. 5.0, and CLC Genomics Workbench ver. 9.5.1 (QIAGEN) for additional analysis.
Small RNA was extracted from the NMC cell lines using a PureLink miRNA isolation kit (Thermo Fischer Scienti c). Small RNA was quanti ed using the Agilent 2100 Bioanalyzer with an Agilent RNA small RNA kit (Agilent Technologies). A small RNA library was prepared using the Ion Total RNA-Seq Kit v2 (Thermo Fischer Scienti c), and the libraries were sequenced on the Ion PGM System. The reads were aligned with miRBase miRNAs. miRNA expression was analyzed using the CLC Genomics Workbench ver. 9.5.1.

MicroRNA assay
Small RNA was extracted from NMC cells using the PureLink miRNA Isolation kit. Small RNA was quanti ed using a NanoDrop (Thermo Fischer Scienti c). miRNA expression was analyzed using TaqMan MicroRNA assays (Thermo Fischer Scienti c), mir-21-5p (#000397), and RNU48 (#001006). In this study, RNU48 was used as a housekeeping miRNA. Reverse transcription (RT) was performed using the TaqMan MicroRNA Reverse Transcription Kit (Thermo Fischer Scienti c) and the indicated RT primers in the assay kits. PCR was performed using the QuantStudio 3D Digital PCR Master Mix, QuantStudio 3D Digital PCR 20 K Chip Kit v2, and ProFlex 2x Flat PCR System (Thermo Fisher Scienti c). PCR was performed according to the manufacturer's instructions [21]. Absolute quanti cation was performed using the QuantStudio 3D Digital PCR System (Thermo Fischer Scienti c). We analyzed the data using the QuantStudio 3D AnalysisSuite Cloud Software (Thermo Fischer Scienti c).

Statistical analysis
The measurements are presented as mean ± SE. Statistical analysis was performed using the GraphPad Prism Software 7.0 (GraphPad Software, USA), and the results were analyzed using Student t-test. Twosided p values <0.05 were considered to be statistically signi cant.

Results
A SETD2 loss-of-function mutation was detected in NMC To date, concomitant mutations other than fusion genes have not been fully elucidated in NMC. Additionally, the fusion protein regulates cell growth and epigenesis of NMC, and the role of epigenetic regulators, such as miRNA, remains to be studied in NMC. Therefore, we investigated gene mutations and miRNA expression in NMC using NGS to identify novel therapeutic targets for NMC.
First, we performed gene mutational analysis using the NMC cell lines HCC2429 and Ty82. Using a comprehensive cancer panel for NGS, 10 non-synonymous mutations were identi ed in HCC2429 cells, while 19 were identi ed in Ty82 cells (Fig. 1a). Among these mutations, we focused on the novel SETD2 mutation, SETD2-p.Ser2382fs, because SETD2 is an epigenetic regulator. Further, this mutation was common in both cell lines.
SETD2 is a tumor suppressor gene [22][23][24]. SETD2 is a non-redundant trimethyltransferase responsible for the trimethylation of lysine 36 on histone H3 (H3K36me3). We observed that the frameshift mutation was located on exon 17 before the WW and Set2 Rpb1 interacting (SRI) domains of the SETD2 gene (Fig. 1b). Hence, it was speculated that the mutation would cause the loss of SETD2 function, i.e., decreased H3K36me3 expression in NMC. Indeed, western blot analysis revealed decreased levels of H3K36me3 in NMC cells, particularly in HCC2429 cells (Fig. 1c).
There is synthetic lethality between H3K36me3 de ciency and WEE1 inhibition: H3K36me3-de cient cancers are signi cantly sensitive to WEE1 inhibitors, which induce DNA damage in tumors [25]. We then evaluated the anti-tumor e cacy of the WEE1 inhibitor, AZD1775 (MK1775, Adavosertib) in NMC cells.
The cell viability assay showed that NMC cells were sensitive to AZD1775 compared with A549 as a control lung cancer cell line harboring wild-type SETD2 (Fig. 2a). Additionally, the expression levels of γH2AX were increased in NMC cells treated with AZD1775 (Fig. 2b). Furthermore, we investigated the growth suppression effects of the combination of AZD1775 and JQ-1 (Fig. 2c). As a result, additive effects of the combination were observed, suggesting that the e cacy of WEE1 inhibition is independent of BET inhibition. NMC often acquires resistance to BET inhibitors. We then established HCC2429 cells resistant to BET inhibitors (HCC2429-JQR) by sustained treatment with JQ-1, the HCC2429-JQR cells were resistant to both JQ-1 and another BET inhibitor, OTX015 (MK8628, Birabresib) (Fig. 3a). As expected, the cell growth assay showed that even HCC2429-JQR cells were sensitive to AZD1775 (Fig. 3b).
miR-21 regulated the growth of NMC Next, we performed miRNA expression analyses. miRNAs are key regulators of epigenetics and pathogenesis in many cancers [26]. However, the relationship between miRNAs and BRD4-NUT has not been fully elucidated. To determine changes in miRNA expression by BRD4 or NUT inhibition, we performed RNA-seq on HCC2429 cells treated with NUT siRNA or JQ-1. Using WB, HCC2429 cells treated with NUT siRNA showed decreased NUT expression (Fig. 4a). The MTS assay showed that NUT siRNA had negligible effects on the growth of HCC2429 cells (Fig. 4a). However, RNA-seq analysis revealed that the expression levels of miRNA were quite different between the siRNA-treated and non-treated cells; the most common miRNA was let-7a-1/let-7a-2/let-7a-3 in untreated cells, whereas miR-21-5p (hereafter miR-21) was the most abundant one in siRNA-treated cells (Fig. 4b). In contrast, JQ-1 did not signi cantly change miR-21 expression in HC2429 cells (Fig. 4b). miR-21 is an oncomiR, and we speculated that miR-21 could be associated with NMC growth [27]. The cell growth effects of miR-21 mimics or inhibitors on NMC were studied. As expected, miR-21 mimics increased the growth of HCC2429 cells; however, miR-21 inhibitors decreased their growth (Fig. 4c).
Short-term treatment with JQ-1 did not alter the expression levels of miR-21 in NMCs (Fig. 4b). However, we suspected that miR-21 is related to BET inhibitor resistance because studies have shown that tumors with acquired resistance to BET inhibitors do not have gatekeeper mutations and drug pump activation [28]. Next, we investigated miR-21 expression in HCC2429-JQR, and the miR-21 expression levels between HCC2429 and HCC2429-JQR were compared using the TaqMan miRNA assay. We found that miR-21 expression was increased in the HCC2429-JQR cells compared to that in the parent cells (Fig.  5a). The MTS assay demonstrated that the miR-21 inhibitor suppressed the growth of HCC2429-JQR cells (Fig. 5b).
Since AZD1775 caused cell growth suppression in the HCC2429-JQR cells as described previously, we determined the e cacy of the combination with the miR-21 inhibitor and AZD1775. Notably, the combination of AZD1775 and the miR-21 inhibitor had additive effects on HCC2429-JQR cells (Fig. 5c).
The SETD2 loss-of-function mutation was detected in a patient with NMC Finally, we investigated gene mutations in clinical NMC samples. Because the thorax is the most frequent location of NMC, we analyzed the expression of the NUT fusion genes in patients with malignant thoracic tumors, whose tumors were located in the midline of their bodies [29]. Thirty-two tumor samples collected via transbronchial biopsy were retrospectively analyzed. Screening for NMC was performed using immunohistochemistry (IHC), which detected NUT, expressed in the nuclei of NMC cells [30,31]. However, in some cases, IHC analysis could not be performed because of the lack of tissue, especially in small biopsy samples. In this study, most of the biopsy samples were very small, and it was assumed that these samples included a small number of cancer cells. We then established a highly sensitive dPCR assay to detect the BRD4-NUT fusion gene. In our preclinical study, the dPCR assay successfully detected the BRD4-NUT fusion gene in 2429 cells; the dPCR assay detected a 1000-fold diluted fusion gene (Fig.  6a). As a result, we identi ed one NMC tumor among the 32 tumors using the assay (Fig. 6c). Because the recurrent SETD2 mutation was observed in the NMC cell lines, gene mutational analysis was performed to detect the SETD2 mutation in the NMC tumor, using the cancer comprehensive panel for NGS. As expected, we identi ed the SETD2-p.Ser2382fs in the tumor.

Discussion
In this study, we describe two epigenetic regulators, SETD2 and miR-21, as therapeutic targets for NMC, regardless of resistance to BET inhibitors. To the best of our knowledge, this is the rst report showing the e cacy of targeted therapy with a WEE1 and miR-21 inhibitor in NMC.
First, we identi ed a novel SETD2 mutation in NMC. SETD2 has three conserved domains, AWS-SET-PostSET, WW, and Set2 Rpb1 interacting (SRI) domains. The SET domain (1550-1673), which mediates trimethylation of H3K36; the WW domain (2391-2420) is associated with protein-protein interactions, preferentially binding to the proline-rich region of proteins, and the SRI domain (2469-2548) binds to the phosphorylated C-terminal domain of RNA polymerase II, allowing SETD2 to move to transcription elongation complexes [32][33][34]. Taken together, SETD2 plays a key role in homologous recombination repair and genome stability by catalyzing trimethylation at H3K36 [35].
SETD2 is ubiquitously expressed in human tissues, and somatic mutations of the SETD2 gene have been reported in several types of cancer, in which loss of SETD2 function leads to decreased H3K36me3 levels [22]. For instance, in clear cell renal cell carcinoma, widespread DNA hypomethylation associated with SETD2 loss-of-function mutations was observed, and the SETD2 mutation seemed to be related to genomic alterations leading to tumorigenesis [36,37]. We detected the SETD2-p.Ser2382fs mutation in NMC. This frameshift mutation is located just before the WW domain, suggesting that the WW and SRI domains in the mutant SETD2 have no additional normal functions. It has been reported that SRI domain de ciency abolishes trimethylation of H3K36me3 [38,39]. Indeed, H3K36me3 expression was decreased in NMC cells. On the other hand, H3K36me3 expression levels in Ty82 cells were different from those in HCC2429 cells. Our NGS analysis showed that the frequency of the frameshift mutation was 93.9% in HCC2429 cells and 89% in Ty82 cells; the residual SETD2 gene was mutant SETD2-p.Pro2381Leu (P2381L) in both cell lines (data not shown). Considering the mutation site, the SETD2-P2381L mutation may be a passenger mutation. The mutant SETD2-P2381L could compensate for SETD2 de ciency caused by the frameshift mutation in Ty82 cells, although the bona de activity of the SETD2-P2381L enzyme remains unknown. If there is a monoallelic de ciency of SETD2 in Ty82 cells, they may retain the trimethyltransferase activity. The mechanistic details remain to be elucidated as to how SETD2-p.Ser2382fs and SETD2-p.Ser2382fs functions in NMC. In addition, there might be possible mechanisms that compensate for H3K36me3 in Ty82 cells.
Identi cation of the SETD2 mutation can lead to the development of targeted therapies for NMC. In H3K36me3-de cient tumors, WEE1 inhibition has a synthetic lethal interaction with H3K36me3 loss; the WEE1 inhibitor AZD1775 selectively kills SETD2-de cient cancer cells through dNTP starvation because of RRM2 depletion [25]. In our study, the NMC cells were more sensitive to AZD1775 than the BET inhibitors, and AZD1775 induced DNA damage in NMC, which was concordant with the results of the present study.
The enzymes regulating epigenesis in histones are categorized as writers, erasers, readers, or others [40,41]. Among these enzymes, SETD2 belongs to the writers that add post-translational modi cations, whereas BRD4 is one of the readers that recognize acetyl groups on histone lysines. Therefore, our ndings suggest that aberrant gene expression and tumorigenesis in NMC might occur through hyperacetylation by BRD4-NUT and hypomethylation by SETD2 loss. We found that the combination of the BET inhibitor and WEE1 inhibitor had additive effects on NMC. Moreover, a recent report showed that the combination therapy targeting BET and p300, which belongs to writers and acetylates histone lysine residues, was more effective than BET inhibitor alone [42]. Given the epigenetic categories in histone modi cation, the combination of BET inhibitors and target inhibitors in another epigenetic category might be useful for the treatment of NMC.
Next, we found that miR-21 regulated the growth of NMC. Emerging studies have reported the functions of miRNAs in NMC. A study showed a set of 48 dysregulated miRNAs in NMC, in which the miRNAs targeting critical genes other than BRD4 and NUT were analyzed; however, miR-21 was not included in the 48 miRNAs [43]. Another study screened an miRNA mimic library and identi ed miR-3140 that targets and suppresses BRD4 by binding to its coding sequence [44]. Another report analyzed miRNA expression in NMC using clinical samples that identi ed three cases of sinonasal NMC, and two out of three NMCs showed upregulation of miR-21, miR-143, and miR-484 [45]. miRNA expression is regulated by DNA methylation and histone modi cations [46]. Therefore, histone modi cation changes caused by SETD2 de ciency or BRD4-NUT might be associated with miR-21 expression. Indeed, altered promoter methylation of miR-21 has been reported in SETD2-de cient cancers [36]. Additionally, H3K36me3 is required for DNA mismatch repair (MMR), while miR-21 downregulates MMR gene expression [47,48]. Overall, together with SETD2 de ciency, miR-21 might be a key regulator in NMC.
We established HCC2429-JQR cells resistant to BET inhibitors. In our study, miR-21 expression was increased in the resistant NMC, and both the miR-21 inhibitor and AZD1775 were effective in the resistant cells. BET family proteins include BRD2, BRD3, BRD4, and BRDT. The BET proteins have conserved tandem bromodomains BD1 and BD2, which selectively bind to acetylated lysine residues in histones. BET inhibitors competitively bind to the individual or both bromodomains and inhibit BET activity [16,17]. Therefore, tumors resistant to BET inhibitors are expected to have gatekeeper mutations at the bromodomains, as seen in epidermal growth factor receptor mutations in lung cancer [49]. However, it is unlikely that the tumors acquire resistant mutations. In triple-negative breast cancer (TNBC), gatekeeper mutations, new driver gene alterations, and drug pump activation were not observed in BET-resistant TNBC cells [28]. This was true in other malignant tumors such as ovarian cancer, prostate cancer, and leukemia, in which alternative signaling pathways other than BET itself were associated with acquired resistance [50][51][52][53]. Therefore, it is possible that NMC might not acquire gatekeeper mutations, although we did not evaluate the mutations in bromodomains in resistant NMC cells. Recent work has shown that adaptive kinome reprogramming is associated with acquired resistance to targeted therapies, and aberrant kinase activation has occurred in BET-resistant cancer cells without gatekeeper mutations [50,54]. miR-21 potentially targets more than 400 genes, using data from miRDB (http://www.mirdb.org), which include various genes of receptor tyrosine kinases (RTKs). Therefore, an increase in miR-21 might be associated with acquired resistance to BET inhibitors by activating RTKs and downstream pathways in NMC. Therapeutic targets other than BET, NUT, and their associated proteins have not been reported. We demonstrated the e cacy of AZD1775 and a miR-21 inhibitor, which could overcome resistance.

Conclusions
Our comprehensive study found SETD2 de ciency and miR-21 as therapeutic targets for NMC. WEE1 and miR-21 inhibitors are novel therapeutic options for NMC and should be investigated in clinical trials. MS performed a historical examination of thoracic tumors. KO performed the gH2AX assay. RY established a BET-inhibitor-resistant NMC cell line. NH, YM, MK, and YO collected biopsy samples. All authors read and approved the nal manuscript.