The bromodomain inhibitor JQ1 triggers growth arrest and apoptosis in testicular germ cell tumours in vitro and in vivo

Abstract Type II testicular germ cell cancers (TGCT) are the most frequently diagnosed tumours in young men (20–40 years) and are classified as seminoma or non‐seminoma. TGCTs are commonly treated by orchiectomy and chemo‐ or radiotherapy. However, a subset of metastatic non‐seminomas (embryonal carcinomas) displays only incomplete remission or relapse and requires novel treatment options. Recent studies have shown effective application of the small‐molecule inhibitor JQ1 in tumour therapy, which interferes with the function of ‘bromodomain and extraterminal (BET)’ proteins. JQ1‐treated TGCT cell lines display up‐regulation of genes indicative for DNA damage and cellular stress response and induce cell cycle arrest. Embryonal carcinoma (EC) cell lines, which presented as JQ1 sensitive, display down‐regulation of pluripotency factors and induction of mesodermal differentiation. In contrast, seminoma‐like TCam‐2 cells tolerated higher JQ1 concentrations and were resistant to differentiation. ECs xenografted in vivo showed a reduction in tumour size, proliferation rate and angiogenesis in response to JQ1. Finally, the combination of JQ1 and the histone deacetylase inhibitor romidepsin allowed for lower doses and less frequent application, compared with monotherapy. Thus, we propose that JQ1 in combination with romidepsin may serve as a novel therapeutic option for (mixed) TGCTs.


Introduction
Approximately 1 of 250 men (age 30-35 years) is diagnosed with TGCT each year [1,2]. These tumours presumably result from a block in primordial germ cell differentiation, giving rise to the precursor lesion termed germ cell neoplasia in situ (GCNIS) [3]. GCNIS are found in the spermatogonial niche of the testes near the nursing Sertoli cells and remain dormant until late puberty [2,4]. Between 20 and 45 years of age, they develop into invasive TGCTs, either classified as seminoma or non-seminoma/EC [5].
Interestingly, JQ1 displayed antitumourigenic effects in malignancies such as leukaemia, glioblastoma or prostate cancer models by inhibition of BRD4 [11][12][13][14]. The BET member BRD4 is a mediator of transcriptional elongation in M/G1 cell cycle transition [15]. It recruits the positive transcription elongation factor complex (pTEFb), which activates expression of growth-promoting and cancer-associated genes [16]. JQ1 was described as the first compound efficiently downregulating MYC-dependent gene expression by abrogating pTEFbdependent promoter-proximal pause release of Polymerase II [16,17].
In this study, we analysed the molecular and cytotoxic effects of JQ1 application on TGCT cells, Sertoli cells and fibroblasts. JQ1mediated bromodomain inhibition resulted in a dose-dependent increase in apoptosis and G1/G0 arrest in TGCTs. Further, an additive effect on apoptosis was seen when using JQ1 and the histone deacetylase inhibitor romidepsin in combination. We propose that JQ1 alone or in combination with romidepsin may serve as novel therapeutic option for TGCTs.

JQ1 and romidepsin treatment in vitro
The cells were treated with 100, 250 and 500 nM JQ1 inhibitor (kindly provided by Jay Bradner, Dana-Farber Cancer Institute, Boston, MA, USA) or 2 and 5 nM romidepsin (Celgene, Signal Pharmaceuticals, LLC, San Diego, CA, USA; MTA ID #CC0488464), respectively. As a negative control, the cells were treated with equal amounts of reconstitution solution (DMSO) only.

AnnexinV/7AAD and PI-FACS analysis
AnnexinV/7AAD and PI-FACS analyses were performed as described before [19]. In brief, cells were plated out at a density of 2 9 10 5 cells/well in a six-well plate for EC cell lines and 1 9 10 5 cells/well for TCam-2, FS1 and MPAF. After indicated time-points, apoptosis was analysed via AnnexinV/7AAD FACS staining, using the PE Annexin V Apoptosis Detection Kit I (BD BioSciences, Heidelberg, Germany). For cell cycle analysis, cells were trypsinized, washed in 1 9 PBS and fixed in 100% ice-cold methanol at À80°C for 2 hrs. After fixation, cells were centrifuged and resuspended in 1 ml PI staining solution (PBS + 2 ll PI (1 mg/ml), +20 ll RNAseA (10 mg/ml)). The cells were analysed (50,000 cells/tube) in a FACS Canto (BD BioSciences).

XTT assay
For XTT assay, cells were plated out at a density of 3000 cells/well in a 96-well plate. JQ1/romidepsin was supplemented after 24 hrs. Cells were stained for their viability by XTT after 24/48/72/96 hrs of initial treatment. The XTT assay was performed as described previously [19].

RNA and protein isolation
For RNA and protein isolation, cells were seeded out at a density of 1 9 10 5 cells/well in a 6-well plate prior to initial JQ1/romidepsin treatment. Proteins were isolated using ELISA Lysis buffer (Cell Signaling, Leiden, the Netherlands). The cell lysate was incubated for 10 min. on ice, followed by a 5-min. centrifugation step at 15,300 9 g and 4°C. Protein concentrations were determined using the BCA Protein Assay Kit (Thermo Scientific, Rockford, IL, USA). Total RNA was extracted using RNeasy Mini Kit (Qiagen, Hilden, Germany). RNA quality was assessed by photometric measurement of ratios 260/280 nm and 260/ 230 nm using a NanoDrop photometer (PeqLab, Erlangen, Germany).

Western blot
Western blot analysis was performed as described elsewhere [19]. For detection, the membrane was incubated for 5 min. in 2 ml PierceSuper Signal West Pico chemiluminescent substrate (Thermo Scientific), and the signal was recorded using the Bio-Rad ChemiDoc TM MP Imaging System (Bio-Rad, M€ unchen, Germany). For antibody details, see Table 1. Densitometric quantification of Western blot protein bands was performed with IMAGEJ Software (Wayne Rasband, National Institute of Health, Bethesda, USA). Density values were calculated relative to the loading control (=1).

Quantitative real-time RT-PCR
Quantitative real-time RT-PCR was performed as described previously [20]. In brief, cDNA was synthesized using Maxima First Strand cDNA synthesis Kit (Thermo Scientific).  Table 2 (Table 2). qRT-PCR was performed using the ViiA TM 7 Real-Time PCR System (Life Technologies). Quantitative values were obtained from the Ct. GAPDH was used as housekeeping gene and for data normalization.

Illumina HumanHT-12 v4 expression array
Illumina expression microarray analysis was performed as described elsewhere [20]. In brief, the cell lines were analysed after 24 and 72 hrs of JQ1 treatment (100 nM) and DMSO as solvent control. Total RNA was extracted and RNA quality was assessed by gel electrophoresis in a BioAnalyser 2100 (Agilent Technologies, Santa Clara, CA, USA). Samples were processed on an Illumina Human HT-12 v4 Bead Chip (Illumina, San Diego, CA, USA), which was performed at the Institute for Human Genetics, Bonn, Germany. Bioinformatic analysis and data normalization were performed by Andrea Hofman (Institute for Human Genetics, Bonn, Germany). Microarray data sets are publically available via GEO: GSE87477 (ncbi.nlm.nih.gov/geo/). Venn diagrams were calculated using Venny (http://bioinfogp.cnb.csic.es/tools/venny/) [21]. Protein interaction networks were predicted using STRING analyses (http://string-db. org/) [22].

In vivo xenograft studies
Xenotransplantation was performed as described previously [23]. In brief, 1 9 10 7 cells were resuspended in 500 ll of 4°C cold Matrigel (BD) and injected into the flank of CD1 nu/nu mice. Tumours were then grown for 2 weeks and subsequently treated with JQ1 or romidepsin.

JQ1 and romidepsin treatment in vivo
Ten per cent (2-Hydroxypropyl)-b-cyclodextrin (HP-b-CD) solution was used as a vehicle, which improves drug solubility (Sigma-Aldrich, Munich, Germany) [24]. The concentrated stock of JQ1 in DMSO was diluted 1/10 in 10% HP-b-CD to obtain a final concentration of 50 mg/ml. Romidepsin was added 1/100 to achieve a final concentration of 0.5 mg/ml. For JQ1 treatment, the animals were dosed 50 mg/kg, 5 days/week, i.p. For combination treatment, the animals were dosed 50 mg/kg JQ1 + 0.5 mg/kg romidepsin 3 days/week, i.p. Tumour growth was measured everyday using a calliper.

Immunohistochemistry
Immunohistochemistry was performed as described previously [20,25]. Tumours were dissected, fixed in 4% formalin at 4°C overnight and processed in paraffin wax. Staining was performed semiautomatically in the Autostainer 480 S (Medac, Hamburg, Germany). For antibody details, see Table 1. Stainings were quantified from three individual tumours (n = 3). Significance was calculated using two-tailed Student's t-test.

TGCT cell lines express the BET members BRD2, BRD3 and BRD4
JQ1 is a small-molecule inhibitor that blocks the binding pocket of bromodomain proteins for acetylated lysine residues of histones, thus interfering with the cellular transcriptional machinery [16]. Different studies have demonstrated its potential use in cancer therapy; however, nothing is known so far about the efficacy of JQ1 treatment in TGCTs. First, we determined expression levels of BRD2, BRD3, BRD4 and BRDT by qRT-PCR. We utilized three different EC cell lines (NCCIT, NT2/D1, 2102EP) and their cisplatinresistant subclones (NCCIT-R, NT2/D1-R, 2102EP-R) as well as the seminoma cell line TCam-2. As proxies for somatic cells and cells of the testis microenvironment, we included a human fibroblast (MPAF) and Sertoli cell line (FS1) in our analysis. All cell lines showed highest mRNA expression levels of BRD2 (Fig. 1A). Levels of BRD3 and BRD4 were lower, while expression of BRDT was very weak to absent (Fig. 1A). The cisplatin-resistant subclones display comparable levels of BRD2, BRD3 and BRDT as the parental cell lines (Fig. 1A). BRD4 mRNA levels appeared slightly lower in cisplatin-resistant cell lines (Fig. 1A). By meta-analysing previously published microarray data, we confirmed expression levels of BRD2, BRD3, BRD4 and BRDT in TCam-2, NCCIT, 2102EP, FS1 and MPAF (Fig. S1A) [19]. Additionally, we performed Western blot analysis to compare BRD2, BRD3 and BRD4 protein levels to RNA expression. We detected BRD2, BRD3 and BRD4 in the nuclei of TCam-2, NCCIT, NT2/D1, 2102EP and FS1 Sertoli cells (Fig. 1B).  Interestingly, quantitation of protein bands relative to HDAC1 levels revealed that there is no correlation between RNA expression and protein levels in all cell lines analysed (Fig. 1B). The differences between RNA and protein levels might point at differences in regulation of transcription, RNA / protein half-life or RNA-protein-turnover between the BRD members. Nevertheless, in all TGCT cell lines and Sertoli cells, BRD2, BRD3 and BRD4 protein levels were detected. Next, we used previously published microarray data to analyse expression of BRD2, BRD3, BRD4 and BRDT in TGCT tissues [26]. Expression levels were measured in normal testis tissues, GCNIS lesions, seminomas and ECs and presented as highly similar to the in vitro expression data (Fig. 1C versus A). In normal testis tissue, high BRDT expression levels were detected, while TGCT tissues showed significantly lower levels. We hypothesize that the expression of BRDT in TGCT tissues reflects the presence of BRDT positive germ cells (Fig. 1C) [27,28]. We conclude that in TGCTs, BRD2, BRD3 and BRD4 are expressed, and thus, application of JQ1, which blocks BET function, may affect growth and survival of TGCTs as well as normal gonadal cells.

JQ1 induces apoptosis and cell cycle arrest in TGCT cell lines
Next, we treated TGCT and control cell lines with 100-500 nM JQ1 and used AnnexinV/7AAD FACS analysis to measure induction of apoptosis. In all TGCT cell lines, JQ1 markedly increased the fraction of apoptotic cells in a dose-dependent manner ( Fig. 2A, Fig. S2A). After 16 hrs of treatment, the lowest JQ1 dose (100 nM) initiated apoptosis in the EC cell lines NCCIT(-R) and NT2/D1(-R). The nullipotent EC cell lines 2102EP(-R) displayed apoptosis later after 20 hrs. Interestingly, two (2102EP-R, NT2/D1-R) of three cisplatin-resistant subclones displayed a higher sensitivity to JQ1-induced apoptosis compared with the parental EC cell lines ( Fig. 2A). TCam-2 cells did not display apoptosis before 72 hrs at all three concentrations (100, 250 and 500 nM), suggesting a reduced sensitivity towards JQ1 ( Fig. 2A, Fig. S2B). Indeed, 100 nM JQ1 was tolerated by TCam-2 cells over a time period of 16 weeks (Fig. S2C), but 500 and 250 nM JQ1 induced apoptosis in TCam-2 cells after 8 and 15 days, respectively (Fig. S2B). Notably, the Sertoli cell line FS1 showed strong induction of apoptosis after 72 hrs of JQ1 treatment, indicating adverse effects of JQ1 treatment on the microenvironment of the testes, that is Sertoli cells (Fig. 2A, Fig. S3A). In contrast, cell viability of fibroblasts remained unchanged ( Fig. 2A, Fig. S3A). To confirm the induction of apoptosis, cleavage of PARP was analysed by immunoblotting in JQ1-treated TGCT and control cell lines. Treatment with 500 nM JQ1 resulted in PARP cleavage in TGCT cell lines and Sertoli cells after 24 and 72 hrs, respectively ( Fig. 2B and C). This effect was not detectable in fibroblasts, further stressing the point that fibroblasts are more resistant to JQ1 (Fig. 2C).
In glioblastoma, leukaemia and medulloblastoma cells, JQ1 treatment induced G1 cell cycle arrest [12,17,29]. Here, we used PI-FACS to analyse cell cycle distribution in JQ1-treated TGCT cell lines. As expected, all TGCT cell lines displayed a pronounced decrease of cells in S-phase and accumulation of cells in G1/G0 phase (Fig. 2D). Of note, even TCam-2 cells treated for 16 weeks with 100 nM JQ1 enriched in G1/G0-phase (Fig. S3B). G1/G0 arrest was also seen in fibroblasts, while Sertoli cells display accumulation of cells in G2/M-phase, indicative for a G2/M cell cycle arrest (Fig. 2D). We conclude that in ECs JQ1 treatment leads to induction of apoptosis at concentrations ≥100 nM, while in seminomas apoptosis is induced at concentrations ≥250 nM. Also, JQ1 treatment leads to accumulation of TGCT cells in G1/G0 phase at concentrations ≥100 nM. However, JQ1 treatment is also reducing cell viability and interfering with cell cycle progression of Sertoli cells, indicating an adverse effect on the testis microenvironment. In contrast, human fibroblasts are resistant to the pro-apoptotic effects of JQ1 and display cell cycle arrest only.  fibroblasts (MPAF) were treated with 100 nM JQ1 and subjected to expression microarray analysis (Data S1A-D). Venn diagrams summarize the numbers of deregulated genes in all cell lines analysed (Data S1A-D), indicating a total number of 329 deregulated genes in TCam-2 and 601 genes in NCCIT (Data S1A and B). JQ1-resistant fibroblasts displayed differential expression of 180 genes, while JQ1-sensitive FS1 cells showed deregulation of 48 genes only (Data S1C and D).
Comparing the group of tumour cell lines (TCam-2, NCCIT) to somatic cells (MPAF, FS1), revealed no commonly deregulated genes (Data S1E), indicating a different response towards JQ1 treatment. However, all four cell lines showed up-regulation of genes of the HIST1 and HIST2 clusters, which encode for building blocks of histone molecules [30] (Fig. 3A and B; green labels, Data S1A-D; green labels), an effect that was already demonstrated in previous publications [16,17,31]. We then sorted for commonly deregulated genes in TCam-2 and NCCIT only and found several genes up-regulated after 24 hrs, classified as 'early effects', that is strong up-regulation of the cell cycle arrest-and apoptosis-associated genes CDKN1C, DDIT4, TSC22D1 and BTG1 (Fig. 3A, blue labels) and induction of the mesodermal differentiation marker HAND1 (Fig. 3A, yellow labels) [32][33][34][35][36]. We utilized STRING algorithm to predict protein-protein interactions among the commonly early up-regulated genes. There, interaction of the HIST1 and HIST2 cluster members was predicted (Fig. S4A).
Among the down-regulated genes, we found THY1, a marker for undifferentiated spermatogonia, which is suppressed after 24 and 72 hrs of JQ1 treatment ( Fig. 3A and B, purple labels) [40,41]. By qRT-PCR analysis, down-regulation of THY1 was confirmed in TCam-2, NCCIT, NT2/D1 and 2102EP (only at concentrations ≥750 nM), but was absent in FS1 and MPAF ( Fig. S5A and B). Similarly, treatment of TGCT cell lines with JQ1 for 72 hrs resulted in down-regulation of three members of 'Interferon-induced Transmembrane Protein' family (IFITM1 -3), which were predicted to interact by STRING ( Fig. 3A and B, red labels; Fig. 3D). In mice, Ifitm1-3 encode for cell surface proteins that may confer distinct guidance cues for migrating primordial germ cells [42,43]. Thus, down-regulation of THY1 and IFITM1-3 may suggest a loss of germ cell-like characteristics and the induction of differentiation.
In summary, we were able to show that JQ1 treatment of TGCT cell lines in vitro results in up-regulation of DNA damage and stress sensors in TCam-2, NCCIT, NT2/D1 and 2102EP cells. Further, in EC cell lines, we demonstrated down-regulation of pluripotency-associated genes and strong induction of germ-layer differentiation markers. These effects were not observed in human fibroblasts or Sertoli cells.

JQ1 treatment reduces tumour burden of mice with xenografted EC cells
To evaluate the therapeutic efficacy of JQ1 application in vivo, we xenografted two different EC cell lines (NCCIT, NT2/D1) into the flank of nude mice. After 2 weeks of initial tumour growth, we treated these mice 5 days/week with 50 mg/kg JQ1 i.p. for 2 weeks. After 1 week, the xenografted animals showed a significant reduction in tumour burden after JQ1 treatment compared with animals treated with solvent only (Fig. 4A-D). Additionally, JQ1-treated tumours displayed a pale and opaque texture and showed a reduced infiltration of blood vessels, suggesting reduced angiogenesis ( Fig. 4A and B). In line, previous studies demonstrated suppressed tumour angiogenesis after JQ1 treatment in in vivo models of childhood sarcoma [54]. Immunohistochemistry verified diminished angiogenesis in JQ1-treated NCCIT tumours by detecting significantly reduced CD31 staining, a marker for endothelial cells [55] (Fig. 4E). Additionally, immunohistochemical staining demonstrated a significant reduction of Ki67-positive cells in JQ1-treated tumours, confirming reduced cellular proliferation (Fig. 4F). So, EC xenografts are affected by JQ1 in two ways, they are restricted in tumour growth and neo-angiogenesis.
Combination treatment with JQ1 and romidepsin effectively reduces cell viability of seminoma and EC components JQ1 treatment induced CDKN1C, RHOB, ID2 and ATF3 expression in TCam-2, NCCIT and NT2/D1, as determined by microarray (Data S1A and B) or qRT-PCR analysis (Fig. 5A). Interestingly, we published recently that a similar up-regulation of RHOB, ATF3, CDKN1C and ID2 is detected in TCam-2, NCCIT and NT2/D1 upon treatment with the HDAC inhibitor romidepsin (Fig. 5A) [19]. Thus, we asked, whether further romidepsin effector genes were up-regulated in JQ1-treated TGCT cell lines. Indeed, similar to romidepsin-treated TGCT cell lines, we demonstrated up-regulation of DUSP1 and GADD45B after JQ1 (Fig. 5A). Of note, expression of these markers was not further increased after combinatorial treatment with both epigenetic drugs (Fig. 5A). So, romidepsin and JQ1 treatment induce a similar stress response in TGCT cell lines.
In a murine lymphoma model, combined HDAC and BET inhibitor treatment led to enhanced apoptosis, improving the overall survival of lymphoma-bearing mice compared to mice treated with the respective monotherapies [56]. Further, in human breast cancer cell lines, JQ1 and the HDAC inhibitor mocetinostats were shown to synergistically reduce cell viability and cell cycle progression [57]. Thus, we asked whether a combination therapy would increase the pro-apoptotic effects of JQ1 or romidepsin monotherapy in TGCT in vitro. Therefore, we treated TGCT cell lines once with either JQ1 (100 and 250 nM) or romidepsin (2 and 5 nM) or both drugs in combination prior to a 96hrs incubation period. Cell viability was assessed by XTT assay every 24 hrs relative to solvent-treated control cells (Fig. 5B). We found that TCam-2 were sensitive to low-dose romidepsin treatment (2 nM), but resistant to a single application of JQ1 (100 or 250 nM) (Fig. 5B). In contrast, the cytotoxicity of low-dose romidepsin application (2 nM) in TCam-2 was enhanced by additional application of 100 or 250 nM JQ1 (Fig. 5B). NCCIT cells were sensitive to JQ1 treatment (100 or 250 nM), while a low dose of romidepsin (2 nM) had only mild effects on viability (Fig. 5B). Only application of 5 nM romidepsin plus 100 or 250 nM JQ1 strongly reduced viability of NCCIT cells (Fig. 5B). Hence, we found that compared to EC cell lines seminoma-like TCam-2 are more sensitive to romidepsin, while being more resistant to JQ1. Vice versa, compared to TCam-2, EC cell lines are sensitive to JQ1, while being more resistant to romidepsin. Thus, a subject diagnosed with a mixed TGCT, including seminoma and EC populations, would benefit from a combination of HDAC and BET inhibitors.
We then asked whether this combination of BET and HDAC inhibitors also kills TGCTs in vivo. Therefore, NCCIT and NT2/D1 cells were xenografted into the flank of 6-week-old nude mice and after 2 weeks of initial tumour growth, mice were treated 3 days/week with 50 mg/kg JQ1 + 0.5 mg/kg romidepsin i.p. for 10 days. The animals show a significant reduction in tumour burden after 2 days ( Fig. 5C and D). Importantly, a similar reduction of tumour burden was achieved using a less frequent application of JQ1 + romidepsin (3 times/week) as compared to JQ1-only treated animals (5 times/week) (Figs 5D versus 4C and D). Additionally, romidepsin concentrations used in this in vivo experiment (0.5 mg/kg) were considerably lower than in a study published recently (2 mg/kg) [19]. This suggests that a combination of both drugs may be administered less frequently and at lower doses, still resulting in a comparable reduction of tumour burden as under monotherapy. Similar to JQ1-treated NCCIT-and NT2/D1derived tumours, the JQ1 + romidepsin-treated tumours displayed reduced infiltration of blood vessels and a pale, opaque morphology (Fig. 5C). Further, immunohistochemical staining showed a reduction in Ki67-positive cells, suggestive of reduced cell proliferation (Fig. 5D). In conclusion, additive effects of both drugs allow for a less frequent application scheme compared to monotherapy.

Discussion
The BET protein family has been described as a promising target for epigenetic therapy and led to the development of the putative cancer therapeutic JQ1 [16,58]. A short elimination half-life and a good pharmacokinetic profile provide JQ1 with ideal drug-like properties [59]. JQ1 inhibits the BET member BRD4, which is involved in regulation of genes associated with growth and cell cycle progression; thus, JQ1 was considered as a cancer therapeutic [8].
Here, we analysed the molecular effects of JQ1 on TGCT cell lines, while utilizing human fibroblasts and a Sertoli cell line as controls. We found that JQ1 induces DNA damage and stress response genes, putatively contributing to the induction of cell death and cell cycle arrest in the GCNIS/seminoma-like cell line TCam-2 and the EC cell lines NCCIT, NT2/D1 and 2102EP (Fig. 6). We further demonstrated that tumours from xenografted EC cells were significantly reduced in size and proliferation rate after JQ1 application. Also, we found a reduced infiltration of blood vessels in JQ1-treated tumours (Fig. 6).
Interestingly, EC cell lines show robust down-regulation of pluripotency genes and up-regulation of differentiation markers, suggesting induction of mesoderm (Fig. 6). There are therapeutic strategies aiming to induce terminal differentiation of tumour cells, rendering them less aggressive [60]. In this respect, JQ1 not only induces apoptosis in ECs, but might lead to acquisition of a less aggressive state. In TCam-2, although down-regulation of some pluripotency genes was observed, no indications for a differentiation process were found (except up-regulation of HAND1). This is in line with previous findings, demonstrating the ability of TCam-2 cells to counteract differentiation-inducing stimuli [61,62].
Notably, JQ1 concentrations affecting survival of EC cell lines (≥100 nM) are comparable to doses killing VCaP prostate cancer or MCF7 breast cancer cells [11,63]. In contrast, TCam-2 cells were able to tolerate doses ≤100 nM up to 4 months and showed first signs of apoptosis at doses ≥250 nM. Puissant and Lockwood et al. defined an IC50 >1000 nM as JQ1: resistance [31,64]. Thus, both EC and seminoma cells can be described as JQ1 sensitive. However, EC cells presented as highly sensitive to JQ1 treatment in comparison with the more resistant seminoma-like cell line TCam-2. It is tempting to speculate whether drug sensitivity of TGCT cell lines goes in hand with the ability to differentiate. In line, 2102EP cells, which are nullipotent and thus not prone to differentiate, are also more resistant to JQ1 than the pluripotent EC cell lines NCCIT and NT2/D1.
In fibroblasts and Sertoli cells, we also observed induction of cell cycle arrest, but apoptosis was restricted to Sertoli cells. Thus, JQ1 affects the cell cycle of non-transformed cells and the cytotoxicity on Sertoli cells indicates side effects on the testis microenvironment. Although previous studies in mice did not report any adverse effects of JQ1 on cells of the testis epithelium, further in vivo data need to explore the effects on human testes cells [59,65,66].
Finally, we demonstrate that in TGCT cell lines, JQ1 elicits a similar stress response as the HDAC inhibitor romidepsin, although both drugs act by a different molecular mechanism (romidepsin: inhibits HDACs, leads to histone hyperacetylation; JQ1: inhibits BRD proteins, interferes with reading the histone code) (Fig. 6). Furthermore, the pro-apoptotic effects of JQ1 can be increased in combination therapy [19] (this study). We showed that in mice, a less frequent application of JQ1 + romidepsin reduces tumour burden with same efficacy as JQ1 alone.
How do these results translate to clinical application? Based on different sensitivities of seminomas and ECs to JQ1, our results suggest that JQ1 is a potent treatment option for ECs and at higher doses also seminomas. While Nettersheim et al. previously demonstrated that romidepsin serves as a suitable treatment option for TGCTs in general [19], a combination of JQ1 + romidepsin allows for lower dosing and a less frequent application regimen. Clinical trials will reveal, whether the combination of JQ1 + romidepsin is suitable for treatment of (mixed) TGCTs and is well tolerated by the subject. Fig. 6 Model summarizing the molecular effects of JQ1 on TGCT cell lines. JQ1mediated inhibition of BRD proteins results in a loss of pluripotency-associated markers in EC cell lines, leading to a teratoma-like differentiation. Also, JQ1 inhibits tumour angiogenesis in EC cell lines. In seminoma and EC cell lines, JQ1 leads to up-regulation of DNA damageand stress response genes, which are in parts also induced by the HDAC inhibitor romidepsin (rdps.) (red labelled genes). These genes putatively contribute to G1 cell cycle arrest and apoptosis.

Conflicts of interest
The authors declare no conflict of interest.

Authors' contributions
SJ performed experiments, analysed the data and wrote the manuscript. DN and HS interpreted the data, designed and financed the research study and wrote the manuscript. MF, SS and FH performed experiments. VS and MG contributed essential reagents and tools. GK contributed to writing and interpretation of data.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: