Data from murine embryos indicate, that the murine homologs of BLIMP1 and PRMT5, are expressed in PGCs from specification on up to their arrival in the genital ridge 1617. Short thereafter, these cells differentiate to become gonocytes and the Blimp1/Prmt5 complex is translocated in the cytoplasm and subsequently, Blimp1 is downregulated. In order to test whether human BLIMP1 and PRMT5 are detected in human fetal PGCs/gonocytes, immunohistochemical analyses were performed on human fetal material. On the 12^th week of pregnancy migrating gonocytes coexpressing PRMT5 and BLIMP1 were detected, (Fig. 1, compare A to B, merged in C, arrows). Next, testes from the 19^th week of pregnancy were analyzed. By this time gonocytes gradually differentiate into prespermatogonia and migrate towards the periphery of the emerging seminiferous tubules to settle down in their niche 18. Both BLIMP1 (Fig. 2A) and PRMT5 (Fig. 2B) were detected at this stage in gonocytes. PRMT5, in contrast to BLIMP1, was detected both in the nucleus and in the cytoplasm. Since the murine Blimp1/Prmt5 complex has been described to mediate symmetrical dimethylation of arginine 3 on histone H2A and/or H4 tails (H2AR3me2s/H4R3me2s) 17 immunohistochemical analysis to detect this modification was performed (Fig. 2D). Co-staining of PRMT5 revealed that the cells displaying high nuclear levels of PRMT5 are in fact positive for the H2AR3me2s/H4R3me2s histone mark (Fig. 2E and 2F, merged). To further analyze the population of cells expressing BLIMP1 we performed double labeling experiments using BLIMP1 (Figure 2G) and the gonocytal markers M2A19 (Figure 2H). BLIMP1/M2A double positive signals were detected in most gonocytes (Figure 2I, arrows). Double labeling for H2AR3me2s/H4R3me2s (Fig 2K) combined with M2A (Fig. 2L) showed, that the M2A positive gonocytes displayed H2AR3me2s/H4R3me2s modifications (Fig 2M). Again, these findings were in accordance with the situation in mice, where the Blimp1 protein is downregulated and the H2AR3me2s/H4R3me2s methylation is gradually lost when germ cells proceed to prespermatogonia 17.

Since the murine Blimp1/Prmt5 complex was specifically detected in early germ cells but not in prespermatogonia16 we next asked whether PRMT5, BLIMP1 and dimethylated histone H2A and H4 could be detected in adult human testes. BLIMP1 was detected in the cytoplasm of round spermatids (Fig. 3A, arrow) and PRMT5 was found in the nuclei and in the cytoplasm of spermatocytes (Fig. 3B arrow) and round spermatids (Fig. 3B black arrowhead). H2AR3me2s/H4R3me2s modification is detected in type A spermatogonia (Fig. 3C red arrowhead), as well as round and elongated spermatids, (Fig. 3C arrow). The cytoplasmatic localization of BLIMP1 in adult testes excludes the functional interaction with PRMT5 and the resulting epigenetic modification. This implicates an alternative mechanism of H2AR3me2s/H4R3me2s modification in adult testes.

We next examined various TGCTs for the presence of BLIMP1/PRMT5 and H2AR3me2s/H4R3me2s. As shown in Figure 4, IGCNU show nuclear BLIMP1 staining (Fig. 4A), cytoplasmatic PRMT5 staining (Fig. 4B) and dimethylation of H2A/H4 (Fig. 4C). Seminomas show predominant nuclear BLIMP1 signal (Fig. 4D) sparse nuclear PRMT5 signal (Fig. 4E) as well as a strong and homogenous signal for H2AR3me2s/H4R3me2s (Fig 4F). In embryonal carcinoma, expression of BLIMP1 (Fig. 4G) and PRMT5 (Fig. 4H) was weak and cytoplasmatic. As expected, histone H2AR3me2s/H4R3me2s methylation (Fig. 4I) was barely detectable and heterogeneous. Yolk sac tumors teratomas and choriocarcinomas stained focally and cytoplasmatic for BLIMP1 and PRMT5 (not shown). Focal cytoplasmatic expression of BLIMP1 and PRMT5 was also observed in differentiated parts of teratoma, while chorioncarcinomas were negative for both proteins. A summary of the results of the immunohistochemical studies is given in Table 1.

In order to quantify the expression of BLIMP1 and PRMT5 we performed RT-PCR analyses on normal testicular tissue as well as on various TGCTs. The RNA levels measured were first normalized to βActin and then calculated as relative expression with normal testicular tissue (N) set at 1. Expression of BLIMP1 was significantly higher in IGCNU (p = 0.029) containing testicular parenchyma and seminoma (Fig. 4K), but not in embryonal carcinoma (EC) (p = 0.16), which was comparable to normal testicular tissue. In contrast, PRMT5 was moderately higher in IGCNU (p = 0.033), while embryonal carcinoma (p = 0,091) and seminoma (p = 0,091) express a similar level of PRMT5 compared to normal testicular tissue (Fig. 4L). These data could be confirmed, using a whole genome expression DNA-Array as reported before 20. Here, the same pattern was observed (see Fig. 4M and 4N).

Finally, we asked whether BLIMP1/PRMT5 and modification of histone H2A and H4 could be detected in TCam-2, a cell line derived from a seminoma patient 2122. Here, we were able to detect BLIMP1 in the nucleus, PRMT5 in the nucleus and the cytoplasm (Fig. 5A–C). RT-PCR analyses showed that BLIMP1 and PRMT5 are expressed in TCam-2 cells (Fig. 5E) and absent JKT1 cells, in agreement with Affymetrix data (Fig. 4M and 4N). Of note, the findings on the JKT-1 cell line are in concordance with the conclusion that it is not a seminoma cell line 2223. Western blot analysis confirmed these results, showing that BLIMP1 and PRMT5, as well as the modified Histones H2A and H4 (Fig. 5F) can be detected. Next, we performed a CoIP on extracts from TCam-2 cells and were able to detect a signal for Blimp1 in material immunoprecipitated with PRMT5 antibody (Fig. 5G). This result demonstrates for the first time that PRMT-5 and BLIMP-1 interact biochemically.

We had shown, that nuclear BLIMP1 and methylated H2A and H4 are expressed in IGCNU and seminoma, yet these cells express either little or cytoplasmic PRMT5 (Fig. 4A–F). We speculated that another methytransferase cooperating with BLIMP1 might be able to compensate PRMT5 function and help in establishing this methylation pattern. PRMT7 which is like PRMT5 a type II methyltransferase seemed a potential candidate since both PRMT5 and PRMT7 have been demonstrated to mediate symmetric arginine dimethylation of sm Proteins required for the spliceosome 24. The CoIP experiment (Fig. 4H), demonstrates that BLIMP1 and PRMT7 interact biochemically. In addition PRMT7 shows a strong nuclear signal in TCam-2 cells (Fig. 4I–M). These results indicate that in germ cell tumors, both PRMT5 and PRMT7 might cooperate with BLIMP1 to establish dimethylation of H2A and H4.

Formalin fixed, paraffin embedded testicular tissues from 46 patients with GCTs (20 seminomas, 15 embryonic carcinomas, 5 Teratomas, 3 yolk sac tumors and 3 choriocarcinomas were collected for this study from archives of Departments of Pathology of University Medical Centers Bonn. Adjacent testicular parenchyma containing IGCNU were studied in 15 cases32. All tumors were classified according to the WHO classification of tumors based on their histology by two independent pathologists. Fresh frozen samples of each of normal testicular tissues (n = 3), seminoma (n = 3), mixed germ cell tumors (n = 3), IGCNU (n = 5) and embryonal carcinomas (EC) (n = 3), as well as RNA extracts of TCam2 33 and JKT-134 cell lines, of which TCam2 resembles a seminoma-like cell-line 212223, were additionally available for this study. Use of the tissue for scientific purposes was approved by the Institutional Regional Committee for Ethics.

Total RNA from at least three samples per tumor entity was extracted with TRIzol (Invitrogen, Karlsruhe, Germany) according to manufacturer's instruction. cDNA-syntesis was performed using SuperScript III reverse transcriptase (Invitrogen, Karlsruhe, Germany) and Oligo d(T)_12–18(Invitrogen, Karlsruhe, Germany) and 100 ng of total RNA according to manufacturers instructions. PCRs were carried out in triplicates with following Primers: BLIMP1 F: 5'-GGGTGCAGCCTTTATGAGTC-3'; BLIMP1 R: 5'-CCTTGTTCA TGCCCTGAGAT-3'; PRMT5 F: 5'TTGCCGGC TACTTTGAGACT-3'; PRMT5 R: 5'-AAGGCAGGA AAGCAGATTGA-3'; GAPDH-F: 5'-TGGTATCGTGGAA GGACTCATG AC-3; GAPDH R: 5'-ATGCC AGTGAGCTTCCCGTTCAGC-3'. (β-Act: 25 cycles BLIMP1 and PRMT5: 30 cycles). After agarose gel electrophoresis of the PCR-products band intensity was measured after RT-PCR with the image analysis software ImageJ 1.37 v (National Institutes of Health, USA, http://rsb.info.nih.gov/ij/) in triplicates and normalized to the according GAPDH band.

Co-IP was performed with DYNABEADS^® (Invitrogen, Carlsbad, USA) following manufacturers instructions. Immunopreciptation was performed with 1,5 μg anti-PRMT5 antibody (Chemicon, Temecula, USA) or anti-PRMT7 (Abcam, Cambrigde UK, 1:250). Western Blot with anti-BLIMP1 antibody followed (provided by H. M. Jäck).

For protein analysis Mini-PROTEAN Electrophoresis Cell and Mini Trans-Blot system was used (BioRad, Munich, Germany). Proteins were isolated using RIPA-buffer and prepared using standard protocol and finaly electrophoresed at 30 mA for 90 min. The gel was blotted onto a PVDF membrane in a BioRad blotting chamber overnight at 30 V at 4°C according to published protocols. After blocking in PBSTM (PBS, 0.1% v/v Tween 20, 5% low fat milk powder) primary antibodies (anti-BLIMP1 1:400 (kind gift from H. Jäck), anti-PRMT5 1:200, Chemicon International, USA) were incubated in PBSTM for 3 h at RT. The secondary antibodies (anti-rabbit-HRP, anti-mouse-HRP: DAKO, Hamburg, Germany) were diluted 1:2000. Finally the membrane was incubated in 2 ml PierceSuper Signal West Pico chemiluminescent substrate (Perbio, Bonn, Germany) and the signal was detected using Kodak X-Ray film (Kodak, Stuttgart, Germany).

DNA Array Dataset used to analyze BLIMP1/PRMT5 expression in Seminoma, embryonal carcinoma, TCam2 and JKT1 were generated as described 32.

For immunohistochemistry on paraffin-embedded tissue, dewaxed, 4-μm thick tissue sections were microwave-pretreated in citrate-buffer. Primary antibodies to PRMT5 (Upstate, Charlottesville, VA, 1:500), PRMT7 (Abcam, Cambrigde UK, 1:250) BLIMP1 (provided by H-M. Jäck, University of Erlangen, Germany 1:500) and H2AR3me2s/H4R3me2s (Abcam, Cambridge, UK, 1:2000) were used for detection. Immunohistochemistry was performed using the DAKO EnVision-AEC Kit and manufacturers protocol (DAKO, Hamburg, Germany) as previously described 7. Briefly, endogenous peroxidase was blocked for 5 min in 0.03% H₂O₂ (diluted in distilled water). Sections were washed in Tris-buffered saline (TBS; 0.05 M Tris and 0.85% NaCl, pH 7.6) and incubated with primary antibodies overnight at 4°C. Thereafter, a HRP-labeled polymer conjugated with a secondary antibody was applied (DAKO EnVision-AEC KIT). Pictures were taken using a Leica microscope fitted with a JVC digital camera (Leica, Bensheim, Germany). Figures were assembled using Adobe CS3 software package. Merge of pictures was performed using ImageJ (NIH, US).