GDNF-expressing STO feeder layer supports the long-term propagation of undifferentiated mouse spermatogonia with stem cell properties

The development of a stem cell culture system would expedite our understanding of the biology of tissue regeneration. Spermatogonial stem cell (SSC) is the foundation for lifelong male spermatogenesis and the SSC culture has been optimized continuously in recent years. However, there have been many inconveniences to reconstruct SSC self-renewal and proliferation in vitro, such as the frequent refreshment of recombinant cytokines, including GDNF, the essential growth factor for SSC maintenance. In the present study, we observed that both STO and MEF cells, which were previously used as feeders for SSC growth, did not express GDNF, but a GDNF-expressing STO feeder could support undifferentiated mouse spermatogonia propagation in vitro for three months without the refreshment of recombinant growth factor GDNF. The cell morphology, growth rate and SSC-associated gene expression remained identical to the SSCs cultured using previous methods. The transplantation of SSCs growing on these GDNF-expressing STO feeders could generate extensive colonies of spermatogenesis in recipient testes, functionally validating the stemness of these cells. Collectively, our data indicated that the further modification of feeder cells might facilitate the self-renewal and propagation of SSCs in vitro.

shown that more than a dozen trophic factors, including GDNF, could influence SSC self-renewal or differentiation 9 , it remains unclear whether feeder cells can express and/or secrete enough GDNF or other factors to regulate the SSC fate decision in vitro.
In the present study, we aimed to investigate whether the expression of the growth factor GDNF in feeder cells is sufficient to support SSC propagation. By profiling the cytokine gene expression in STO and MEF cells, we found that none of the feeders expresses GDNF. We therefore generated a GDNF-expressing STO cell line by constructing lentivirus particles with mouse GDNF cDNA. Using this cell line as feeders, SSCs have grown and proliferated in vitro for over three months without apparent phenotype change or function loss. Our data suggest that the modification of feeder layers could be an alternative scheme to optimise the self-renewal and growth of mouse spermatogonial stem cells in vitro.

Expression profile of trophic factor genes in STO and MEF cells.
Unlike SSCs, which grow with a typical clump-forming and grape shaping morphology, both the STO cells (distributed from Dr Alan Bradely) and MEF cells (used for ESC growth) in culture were shown as monolayer (Fig. 1a). Many trophic factors contributed to the self-renewal or fate decision of SSCs in testis niche or in vitro, therefore we first investigated whether the STO and MEF cells as well as the SSCs could express these genes by cDNA amplification. The identity of SSCs in culture was first verified by Zbtb16, Lin28a, Dazl, Ddx4, Pou5f1 at mRNA levels and ZBTB16 and LIN28A at protein levels ( Supplementary Fig. 1). Both STO and MEF cells have detectable expression of Fgf2, Lif, Csf1, Ngr1, Notch1, Egf, Wnt5a and Vegfa, while Igf-1 expression was only detected in STO cells and Bmp4 expression in MEF cells. Meanwhile, there was detectable expression of Notch1, Egf, Fgf8 and Vegfa in SSCs, indicating the possibility of autocrine in SSCs. However, we are unable to detect the expression of neither Gdnf, nor Nodal or Wnt3a at any of the somatic feeder layer or stem cells themselves (Fig. 1b).
Establishment of a GDNF-expressing STO line. All developed cultures for the SSCs derived from several species was supported by the NS0 cell (mouse myeloma cell)-derived human recombinant GDNF 6,7,10,11 . GDNF is a conserved protein among mice, rats, dogs, cows, chimpanzees and humans (Fig. 2a). To generate GDNF-expressing STO cells, we packaged the mouse GDNF cDNA into a pCDH-EF1-MCS-T2A-Puro vector plasmid and integrated a 7818bp sized lentivirus construct (Fig. 2b). We transplanted the lentivirus into six-week old C57BL/6 mouse testes. One month after transplantation, in the seminiferous tubules of the mouse testes, we found a significant proliferation and accumulation of LIN28A positive germ cells, which was used as a marker for Scientific RepoRts | 6:36779 | DOI: 10.1038/srep36779 spermatogonial progenitors 12 , demonstrating the biological function of over-expressing GDNF in the SSC niche by virus transplantation (Fig. 2c, HE and IHC staining). Using these lentivirus particles, we generated a STO cell line with a high level of GDNF expression, which are validated by a quantitative RT-PCR and Western Blot at the mRNA ( Fig. 2d) and protein levels, respectively (Fig. 2e).
Next, we conducted an ELISA assay to measure GDNF concentration secreted from STO cells. As a result, we found that the average of 4.13 ± 1.56 ng/ml (approximately from 2.53 to 5.92 ng/ml) mouse GDNF was detected in the culture medium, which was generated by 200,000 culturing STO cells within 24 hrs ( Fig. 2f and Supplementary datasheet 1).

Long-term propagation of germ cells on a GDNF-secreting feeder layer. SSCs enriched germ
cells were isolated from prepuberty mice of C57BL/6 or 129 × C57BL/6 ROSA mT / mG tdTOMATO transgenic mice by CD90.2 conjugated magnetic bead sorting, and the cells were plated into culture wells with a pre-coated GDNF-expressing STO feeder layer, complying with previous methods 13 . In addition, 1ng/ml FGF2 was supplied into the medium for SSC culture, because we found the relative expression of FGF2 was much lower in STO when compared to Sertoli cells ( Supplementary Fig. 2). We found that the initial clumps of SSCs could be easily established in culture (Fig. 3a, at day 1 and 7). A continued proliferation of SSCs was able to be achieved by a weekly subculture for three months in vitro (Fig. 3a, at day 90).
In total, we have established one line of SSCs from the mice with C57BL/6 background (Fig. 3a), and two lines of SSCs from ROSA mT / mG transgenic mice with 129 × C57BL/6 background (Fig. 4a). In order to make a growth comparison of above SSC line (medium only supplied with 1ng/ml FGF2) with other SSC line supported by normal STO but medium supplied with recombinant GDNF, we then harvested SSCs from established lines and started with 100,000 above cells for each line to generate a 25-day proliferation comparison. As a result, we found no obvious alterations on the cell growth, the size and morphology of SSCs when compared with previous studies 6,7 (Fig. 3b). The proliferation of these SSCs supported by GDNF-expressing STO feeder appeared even faster than cells supported by normal STO; however, adding additional recombinant GDNF to culture did not significantly improve the proliferation of SSCs on GDNF-expressing feeders (Fig. 3b).

Functional validation of stemness for cultured germ cells.
SSCs are capable of colonizing in recipient testes and maintaining long-term spermatogenesis, therefore spermatogonial cell transplantation provides a definitive method to define SSC presence and function in a given cell population. Using GDNF-expressing STO feeder layer, we have established a line of germ cells from ROSA mT / mG tdTOMATO transgenic mice and these cell clumps are RFP fluorescence positive (Fig. 4a, DIC, RFP and Merged images). We transplanted the cultured germ cells into busulfan-treated mouse recipient testes. Three months later, we found many fluorescent-positive regeneration colonies of spermatogenesis in the recipient testes (Fig. 4b). Next, we dissected the testes and found, inside of seminiferous tubules, normal spermatogenic wave and elongated sperms which were not present in busulfan-treated tubules (Fig. 4c). These data, taken together, functionally validated the presence of stem cells in cultured germ cells.
Phenotype analysis of cultured germ cells. Furthermore, we did a karyotype analysis on cultured SSCs supported by GDNF-expressing STO feeders. Compared to previous culture methods with the frequent refreshment of exogenous GDNF, we observed no apparent loss on chromosome numbers when SSCs were cultured on GDNF-expressing feeders (Fig. 5a). We then investigated the expression of those genes that are functionally related to stem cell renewal (Pou5f1, Zbtb16, Etv5, Bcl6b), differentiation (Stra8, C-kit) (Fig. 5b), and cell cycling (Ccna2, Ccnb1, Ccnd2, Ccne2) (Fig. 5c). We did not detect any significant difference in gene expression related to cell renewal and cycling. Although the C-kit gene was statistically higher (p = 0.0443, t-test) for SSCs cultured on GDNF-expressing STO feeders than for the control culture (STO with GDNF, FGF2,GFRA1, Fig. 5b), the expression of both Stra8 and C-kit remained at fairly low levels in the former culture, indicating the majority of germ cells kept undifferentiated in the GDNF-secreting STO culture. We also found no changes for other SSC-associated genes, including Gfra1, Sall4, Vasa, Id4, Pax7 ( Supplementary Fig. 3). These data indicated that SSC culture supported by GDNF-expressing feeders was likely identical to previously reported SSC culture.
In vitro proliferation of SSCs was dependent on GDNF. In order to determine the response to GDNF depletion for these SSCs who grown on the GDNF-expressing STO, we then collected these cells and plated them on the normal STO feeder without GDNF expression. One week proliferation was compared by omitting recombinant GDNF or FGF2 from the serum-free medium (Fig. 5d). Unsurprisingly, we found these SSCs experienced a rapid loss in the culture by GDNF withdrawal; however, the daily refreshment of FGF2 was beneficial to SSC survival and maintenance (Fig. 5d, p < 0.05).

Discussion
Spermatogonial stem cells (SSCs) are the foundation of spermatogenesis throughout adult life. The development of techniques to maintain SSCs in vitro essentially allows for improving our understanding of the processes and regulation of spermatogenesis and aids in solving problems associated with male infertility 14,15 . Several mammalian germ cell lines have been previously constructed through the over-expression of genes, such as the simian virus 40 large tumor antigen, temperature-sensitive mutant of p53, and mouse telomerase 16,17 ; however, it is only by supplementing the growth factor GDNF, a member of the TGFβ family into the testis cell culture, so that a successful establishment of stem cell lines for(from) germline lineages with the capability of long-term self-renewal and complete differentiation through functional transplantations 6,7 , is observed.
Although recent data showed that feeder-free culture could support germ cell proliferation with a lower SSC frequency after transplantation 18 , both STO and MEF cells derived from mouse embryo are beneficial to mouse SSC growth, which is similar to the cultures of ESCs and/or PGCs (primordial germ cells) [19][20][21] . It was speculated that growth factors and cytokines secreted by the feeders and interaction between stem cells and feeder cells can be attributed to beneficial effects to stem cells 9 . Many growth factors and cytokines have been reported to regulate SSC survival and proliferation 9 and our data indicated that both STO and MEF cells have detectable expression of growth factor genes including Fgf2, Lif, Csf1 and Egf. Of those factors, endogenously produced FGF2 has also been shown to be beneficial for mouse SSCs, which is likely through MAP2K1 kinase activation to regulate downstream genes 22,23 and a recent study showed that FGF2 even independently supports in vitro SSC expansion without GDNF 24 . Furthermore, the growth factors LIF and BMP4 could balance self-renewal and differentiation in mouse SSCs and BMP4 could induce differentiation in Kit-negative spermatogonia through its Alk3 and Smad5 receptor signaling pathways 25 . We found that Bmp4 was only expressed by MEF feeder cells, but a direct culture comparison of SSC culture on STO and MEF cells and an evaluation of the potential effects of BMP4 secreted by feeders have not been reported. Nevertheless, both STO and MEF cells did not express GDNF. It was shown that the GDNF released from testis Sertoli cells and the SSCs population in the cryptorchid testes expanded by approximately 20 fold 26   growth has not been validated. Recently testosterone-regulated GDNF secretion by peritubular myoid cells has also been observed, which ensured the optimal GDNF level for repopulating the undifferentiated spermatogonia including SSCs 30 . However, testis somatic cell feeders were likely to inhibit SSCs survival and replication in vitro 9 , and it is unknown whether the overexpression of GDNF and/or other growth factors in STO or MEF could support SSC survival or the long-term proliferation of a SSC line.
Taking this information into consideration, in the present study we generated a GDNF-secreting STO cell line and applied it as a feeder layer to support SSCs in vitro. The concentration of mouse GDNF is about 4.13ng/ ml generated by 200,000 STO cells within one day, which is about 25% of the dose applied in previous culture by human recombinant GDNF (20 ng/ml), but every other day refreshment is required for previous culture. Even though, we found no obvious defects on the size and morphology of SSC clumps. Significantly, we found a better growth rate than culture used in previous studies (Fig. 3b). The results strongly suggested that a direct and durative secret of GDNF from feeder layers might be more beneficial to SSC growth than every other day refreshment of human recombinant factors into medium.
Collectively, we found that a successful SSC propagation for over three months was achieved on GDNF-expressing STO feeders. Our results strongly indicated that the modification of a feeder layer could be useful to optimise the culture system for SSCs and avoid the tedious work and high costs that accompany the addition of recombinant growth factors to culture.

Materials and Methods
Cell culture and donor mice. The STO cells (Passage 4 to 6) were gifted by Dr. Ralph Brinster from the University of Pennsylvania, USA with the permission of Dr. Alan Bradley from the Wellcome Trust Sanger Institute, Cambridge, UK. The STO cells were cultured in DMEM medium (Life Technology) supplemented with 7% foetal bovine serum (FBS, Hyclone). Before applying for the SSC culture, the cells were inactivated with 10µg Mitomycin-C (MITC) ( Sigma-Aldrich) following the previous protocol 13 . An initial purification of SSC-enriched germ cells was achieved through cell sorting according to the previous protocol 13 . Briefly, testes were harvested from of 6-8 days old C57BL/6 background mice or 129 × C57BL/6 background ROSA mT / mG tandem dimer 13 transgenic mice (The Jackson Laboratory, Maine, USA). After the digestion with 1 mg/ml Collagenase IV (Sigma-Aldrich), 0.25% Trypsin-EDTA (Life Technology) and 7 mg/ml DNase (Sigma-Aldrich), single cell suspension was incubated with CD90.2 conjugated microbeads, and applied with a magnetic-activated cell sorting process. Approximately 100,000 CD90.2 positive cells were seeded on MITC inactivated STO cell feeders. Both normal STO and GDNF-expressing STO cells were plated at 100,000 cell per well on 24-well plates. A serum-free medium containing 1 ng/ml FGF2 (BD Biosciences) was used as the culture as described in the previous protocol 13 . A synchronous operation for medium changes was maintained for either cells cultured on normal or on GDNFexpressing STO feeders, while the former also required 20 ng /ml of the recombinant growth factor GDNF(R&D Systems) and 150 ng/ml of GFRA1(R&D Systems). The SSCs were subcultured to new STO feeders every five days at a ratio of 1:2 or 1:3.
All the procedures for animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) of Nanjing Medical University (ID: 2011082112). In addition, all experiments were performed in accordance with the institutional guidelines of Nanjing Medical University.
Plasmid construct and transfection. The mouse CDS sequence of GDNF was taken from the GeneBank database. We amplified the GDNF sequence from the C57BL/6 mouse strain testis cDNA and constructed GDNF cDNA into the pCDH-EF1-MCS-T2A-Puro vector by performing a restriction enzyme cutting with XbaI and BamHI. A CaCl 2 -mediated plasmid transfection was administered using the protocol provided by System Biosciences (pCDH cDNA cloning and Expression lentivectors). Human HEK293T cells were used to generate the lentivirus suspension. The virus was collected from a 10 cm culture dish when HEK293T cell density was approximately 80%. The virus suspension was diluted (1:1) with a STO culture medium, and 100,000 STO cells were placed in each well of a six-well plate and incubated with the virus for 24 hours; then, the fresh STO medium was supplied and the cells were eventually harvested for verification within 3 days.
Recipient mice and SSC transplantation. 129/SvCP × C57BL/6 heterozygous male mice (Jackson Laboratory) were prepared as the recipient mice. 55 mg/kg Busulfan (Sigma-Aldrich) were administered to mice at 2 months old, and the recipients were then ready for transplantation five weeks after. For the lentiviral transduction of testis tissue, the GDNF lentivirus particles were concentrated by 100-fold by ultracentrifugation at 50,000 g for 90 min at 4 °C (Optima L-100 XP Ultracentrifuge), and a 10 μ l concentrated lentivirus medium was injected into the rete of each testis of the 6-week-old C57BL/6 (B6) mice. For cell transplantation, 100,000 SSCs (ROSA mT / mG tdTomato) cultured on the GDNF-expressing STO layer were transplanted into the rete of each testis of the Busulfan-treated recipient mice. Three months after transplantation, the testes were harvested and the donor-cell derived colonies were visualised by a fluorescence dissecting microscope (Zeiss Lumar V12).

RNA isolation and real-time PCR.
The total RNA was isolated by Trizol Reagent (Life Technology), followed by a chloroform and isopropanol separation. First-strand cDNA was synthesised with a reverse transcription kit (Takara Bio). A quantatitive PCR was performed with the SYBR Premix Ex Taq II mixture (Takara Bio) in the 20μ l reaction system by StepOne Plus system (Applied Biosystems). The primer information is listed in the supplementary information. The expression level of each gene was calculated by a 2 −ΔCT value. ELISA Assay. Starting with approximately 200,000 cells per well in a 12-well culture plate, normal STO and GDNF-expressing STO cells were cultured in 1 ml STO medium for 24 hrs. Medium were immediately collected and applied for the ELISA assay. The procedure was followed by an instruction in a commercial GDNF mouse ELISA Kit (Cat#ab171178, mouse GDNF ELISA kit, Abcam). OD (optical density) values were generated through the Microplate Reader (BioTek Synergy2, USA). 20 ng/ml human recombinant GDNF in STO medium served as positive control.
Karyotyping analysis. The Cultured SSCs were incubated with 1 mg/ml of Colchicine (Dahui BioCompany, China) before the trypsin digestion. The Cells were then incubated with 75 mM potassium chloride for 15 min and fixed with methanol/acetic acid (3:1) for 1 h. A metaphase spread was performed and the sample slides were stained with DAPI (Sigma-Aldrich) and analysed by confocal microscopy (LSM 700, Zeiss).