Neonatal (4-8 d.p.p) DBA/2J male mice were used for isolation of primary spermatogonia (male germline stem cells, mGSCs) as described previously (27, 28). All animal procedures were in accordance with and approved by the animal ethical committee of the Amsterdam UMC, Academic Medical Center, University of Amsterdam.

Mouse GSCs were cultured as previously reported (25, 27, 29). The cells were cultured on mitotically inactivated mouse embryonic fibroblasts (MEFs; Gibco, A34962), using a medium (medium I) composed of StemPro-34 SFM medium (Thermo Fisher Scientific), StemPro-34 Supplement (Thermo Fisher Scientific), 1% fetal bovine serum (FBS), recombinant human GDNF (10 ng/ml, 450-10, Peprotech), recombinant human bFGF (10 ng/ml, 100-18B, Peprotech), recombinant human EGF (20 ng/ml, AF-100-15, Peprotech), recombinant human LIF (10 ng/ml, CYT-644, Prospec), as well as other components as previously reported (29) and described in Supplementary Table 1 . The cells were refreshed every 2-3 days, and passaged every 5-7 days at a ratio of 1:4-6 on freshly plated mitotically inactivated mouse embryonic fibroblasts. The cells were maintained at 37°C in 5% CO2 in air.

As a feeder cell to support in vitro meiosis of mouse mGSCs, we used an available immortalized Sertoli cell lines SK49 (30). The cells were cultured at 37°C and 5% CO2 in Dulbecco’s Eagle’s medium (DMEM; Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS), penicillin (100 U/mL) and streptomycin (100 U/mL).

SK49 cells, inactivated by mitomycin (10µg/mL, M7949, Sigma), were grown on 12-well plates pre-coated with laminin (20 µg/mL, L2020, Sigma) to a density of 1 x10⁵ per well. Then mGSCs were seeded on these Sertoli cells to maintain mGSCs proliferation for two days at 37°C using medium I, composed of as described previously (17, 29). To induce meiosis, the cells were cultured at 34°C. From day 0 to day 3, medium was changed to medium II, composed of StemPro-34 SFM medium and StemPro-34 Supplement, 10% KnockOut Serum Replacement (KSR), 5% fetal bovine serum (FBS), Retinoic acid (RA) (1µM, R2625, sigma), Recombinant Mouse BMP-4 Protein (20 ng/mL, 5020-BP, R&D Systems), Recombinant Mouse Activin A Protein (100 ng/mL, 338-AC, R&D Systems), and other components described in Supplementary Table 1 . Starting from day 3 after meiosis induction, medium was changed to medium III, composed of StemPro-34 SFM medium and StemPro-34 Supplement, 10% KSR, 5%FBS, RA (1µM), Bovine Pituitary Extract (BPE) (30µg/mL, 13028014, Thermo Fisher Scientific), Follicle-stimulating hormone (FSH) (100 ng/mL, F4021, Sigma), Testosterone (5 µM, 86500, Sigma), and other components described in Supplementary Table 1 , which was refreshed daily. From day 4 to day 6, the same amount of additional 12-well plates was prepared and pre-coated with laminin. Then undifferentiated mGSCs were grown on SK49 Sertoli cells in medium I. At day 6, the in vitro-induced meiotic cells were dissociated by accutase (Thermo Fisher Scientific), and were subsequently seeded on the additional 12-well plates containing undifferentiated mGSCs and SK49 cells. All cells were subsequently cultured in medium III. For the control group without undifferentiated mGSCs, the in vitro-induced germ cells grow on SK49 Sertoli cells using medium III. From day 6 to day 15, cells were collected for immunocytochemistry.

Meiotic spread preparations were prepared as previously described (31). Alternatively, the cells were spread on the slides using a Cytospin (CELLSPIN, 521-1990, VWR) (17). Briefly, in vivo spermatocytes were yielded from seminiferous tubules as previously described (31). Then these spermatocytes were washed three times with 1x phosphate buffered saline (PBS) and diluted in 200 μL PBS/1% BSA containing 30,000 to 50,000 cells for each cytospin spot and spun for 7 minutes at 112g. Similarly, in vitro-induced germ cells were detached from the culture dish using 0.25% trypsin, and transferred to microscope slides by Cytospin. The slides were air dried for 10min, fixed in 4% PFA and stored at 4°C in PBS or stored at -80°C after air drying.

Immunocytochemistry was performed as described previously (17). Omission of the primary antibodies and replacement with mouse, rabbit and sheep isotype IgGs were used as negative control. Primary antibodies and secondary antibodies are described in Supplementary Table 2 .

Fluorescence microscopy images were acquired using a Plan Fluotar 100×/1.30 oil objective on a Leica DM5000B microscope equipped with a Leica DFC365 FX CCD camera. Images were analyzed using Leica Application Suite X and Image J version Java 1.8.0_77. The figures and graphs were constructed using Graphpad prism, Adobe Photoshop CS5 version 13.0.1 and Adobe illustrator version CS6.

For imaging and quantification of meiotic cell types at all different time points, 3 microscope slides for each time point from three independent experiments were assessed. For quantification of RAD51, 13 testicular (in vivo) pachytene spermatocytes and 13 in vitro-derived pachytene spermatocytes were assessed. For MLH1, 12 testicular (in vivo) pachytene spermatocytes and 13 in vitro-derived pachytene spermatocytes were assessed. Statistical significances between the number of foci were determined by applying the Student’s t-test.

To optimize chromosome synapsis, XY-body formation and meiotic recombination in our in vitro spermatogenesis system, we provided a “second wave of in vitro spermatogenesis” by re-plating in vitro-generated early meiotic cells to an undifferentiated layer of GSCs growing on freshly plated proliferatively inactivated SK49 Sertoli cells. The culture system now consists of three time periods with three different culture media that represent: (1) spermatogonial self-renewal, (2) spermatogonial differentiation and initiation of early meiosis, (3) co-culture of meiotic cells with GSCs ( Figure 1A ). Spermatogonial self-renewal, differentiation and initiation of meiosis were induced by using medium I and medium II, respectively, exactly as reported previously (17). Early meiosis was further supported by a third medium (medium III) that, besides follicle-stimulating hormone (FSH), testosterone, and bovine pituitary extract (BPE), also contained retinoic acid (RA). After an induction period of 6 days, the in vitro-generated spermatocytes were collected and seeded on plates containing GSCs and SK49 Sertoli cells that were pre-cultured for two days using medium I to maintain self-renewal. Medium III was subsequently used to support further meiotic progression of the already in vitro-generated spermatocytes and simultaneously initiate the differentiation of the fresh layer of GSCs. To assess meiotic progression, we collected the cell samples at days 6, 8, 10, 12, 14 and 15 ( Figure 1A ). To monitor synapsis of the homologous chromosomes, we stained the in vitro-generated spermatocytes using antibodies against SYCP3, HORMAD1 and CREST serum to mark the synaptonemal complex, unsynapsed chromosomes and centromeres, respectively. In line with the in vivo situation, in which HORMAD1 specifically accumulated on unsynapsed chromosome axes ( Supplementary Figure S1A ) (32), we observed that, in vitro, HORMAD1 co-localized with SYCP3 during the leptotene to zygotene stages and disappeared from the axial elements as chromosome synapsis proceeded, ultimately only staining the unsynapsed X and Y chromosomes at full pachynema ( Figure 1B ). We also stained for SYCP1, which is a central element protein of the synaptonemal complex and thus specifically marks synapsed areas of meiotic homologous chromosomes (33). Indeed, we observed no SYCP1 staining during leptonema, and appearance of several short SYCP1 fibers during zygonema. While at the early pachytene stage, some axial elements marked by SYCP3 still lacked SYCP1, all SYCP3, except on the sex chromosomes, fully co-localized with SYCP1 in in vitro-generated pachytene spermatocytes ( Figure 1C ). Such pachytene spermatocytes were mostly observed from day 8 to day 12 and were only occasionally present at day 14 ( Figure 1A ). In addition, early pachytene spermatocytes were observed from day 8 to day 14, and leptotene/zygotene spermatocytes from day 3 to day 15 ( Figure 1A ). Hence, in the current system for in vitro meiosis, full synapsis of the homologous chromosomes was achieved from day 8 to day 12 after induction of meiosis.

When meiotic synapsis between the autosomes is being completed, the X and Y chromosomes are transcriptionally silenced in a condensed chromatin area called the XY body (34). To investigate whether in vitro- generated pachytene spermatocytes also form an XY body, we used antibodies against the DNA damage response proteins γH2AX, MDC1 and ATR that are known to be restricted to the XY body at the pachytene stage in vivo (35). γH2AX marks DNA double-strand breaks (DSBs) that are continuously being generated on unsynapsed meiotic chromosomes (36). As a binding partner of γH2AX, mediator of DNA damage checkpoint 1 (MDC1) initiates meiotic sex chromosome inactivation (MSCI) and mediates XY body formation (37). As described by us (17), in vitro-generated pachytene-like cells in our pervious culture system did not display full chromosome synapsis, and thus γH2AX staining remained present on many autosomes while no clear XY body could be discerned. However, in the current culture system, 3 to 4 clear pachytene cells per microscope slide, from day 8 to day 14 ( Figure 2A ), were observed to reach full synapsis of the autosomes, marked by SYCP1 ( Figure 2B ), while clear XY bodies were marked by MDC1 ( Figure 2B ) or γH2AX and MDC1 ( Figure 2C ), which is consistent with pachytene spermatocytes in vivo ( Figures 2B, C ). Because the undifferentiated GSCs present in the second culture dish will also initiate meiosis after the change to medium III upon addition of the first generation of meiotic cells, the pachytene spermatocytes (derived from the first generation of meiotic cells) are massively outnumbered by early meiotic cells (derived from these GCSs). We therefore quantified the total numbers of these cell types, and meiotic M-phase cells to record the dynamics of spermatocyte numbers during the culture period, as presented in Supplementary Figure S2A .

Similar to γH2AX and MDC1, the ataxia telangiectasia and Rad3-related protein (ATR) also localizes to unsynapsed chromosomes and the XY body in vivo ( Figure 2D ), where it is involved in MSCI (38). In our culture system, while still being present on some unsynapsed chromosomes during early pachynema, ATR-staining now, together with MDC1, appeared restricted to the sex chromosomes in in vitro-generated pachytene cells ( Figure 2D ). Hence, the current in vitro meiosis system supports full synapsis of the homologous chromosomes and formation of the XY body.

In vivo, meiotic DSBs are repaired via meiotic recombination, which eventually leads to formation of meiotic crossovers between homologous chromosome pairs (39, 40). The number of initial DSB repair sites, marked by DNA repair protein RAD51, has a peak at leptonema, after which it starts to decline from zygotene until about 1-2 meiotic crossovers, marked by the MutL homolog 1 protein MLH1, remain at the late pachytene stage (40). In contrast to in vivo pachytene spermatocytes, in which RAD51-foci were restricted to the X and Y chromosomes, we observed that RAD51-foci remained present on almost all autosomes in in vitro-generated pachytene spermatocytes, even when the XY body, marked by MDC1, was already formed ( Figure 3A ).

Finally, we used antibodies against MLH1 to investigated whether meiotic crossovers were formed in vitro. In addition, we used antibodies against SYCP1 to mark synapsed homologous chromosomes and MDC1 to mark the XY body. In comparison to the in vivo control, which nicely showed 1-2 MLH1 foci on every synapsed chromosome pair, only few meiotic crossovers were visible in in vitro-generated pachytene spermatocytes ( Figure 3B ). For quantification, the number of RAD51 and MHL1 foci in pachytene spermatocytes was counted, showing an increase in RAD51 and a decrease in MLH1 foci in vitro, indicating deficient meiotic recombination in vitro ( Figures 3C, D ).

Meiotic crossovers are required for the formation of chiasmata, physical links between the homologous chromosome pairs that ensure proper segregation of the homologous chromosomes during the first meiotic M-phase (15). In the current culture system, 1-2 M-phase cells per microscope slides were observed between day 8 to day 14 after meiotic induction ( Supplementary Figure S2A ). In accordance with the very inefficient formation of meiotic crossovers, these M-phase cells only displayed pairs of sister chromatids (univalents) with CREST-stained centromeres located at the ends of the paired sister chromatids ( Figure 4A ). In addition, 1-2 flower-shaped cells per microscope slide, identified as a type of premature M-phase cells in our recent study (17), could still be observed between day 8 to day 15 in this current culture system ( Figure 4B ). In all, we conclude that, despite the completion of synapsis and XY body formation, meiotic crossover formation was still not completed in vitro, which prevents formation of bivalents (pairs of homologous chromosomes) during the first meiotic M-phase.

To investigate whether addition of mGSCs is required for synapsis and XY-body formation, we next compared the addition of undifferentiated mGSCs on a freshly plated layer of proliferatively inactivated SK49 Sertoli cells (group A) to a fresh layer of SK49 Sertoli cells alone (group B). Again, we used antibodies against SYCP3, HORMAD1, SYCP1 and MDC1 to mark the synaptonemal complex, unsynapsed chromosomes, synapsed chromosomes and XY body formation respectively. From day 8 to day 12 we observed comparable numbers of pachytene spermatocytes that displayed fully synapsed homologous chromosomes and XY-body formation in both groups ( Figures 5A, B and Supplementary Figure S2B ). Moreover, to assess meiotic crossover formation between the both groups, we used antibodies against SYCP1, MLH1 and MDC1 to mark synapsed chromosomes, meiotic crossover and XY body respectively. Again, and in both groups, only very few meiotic crossovers could be observed ( Figure 5C ). In all, omission of fresh mGSCs did not affect meiotic synapsis and XY-body formation in this culture system.