Testicular specimens were biopsied from obstructive azoospermia (OA) patients and immediately placed aseptically into Dulbecco’s modified Eagle’s medium (DMEM, Gibco) containing 1,000 U/ml penicillin and streptomycin (PS, Gibco) for further processing. The diagnosis of OA patients was confirmed by pathological examination. This study was approved by the Institutional Ethical Review Committee of Shanghai First People’s Hospital (license number 2016KY196), Shanghai Jiao Tong University School of Medicine. All participants had provided written consents. In total, eight testis samples were collected from OA patients from age 25 to 40. Collected testis tissues were cut into small pieces and placed in the cryopreservation medium consisting of DMEM, 20% fetal bovine serum (FBS, Hyclone), 10% dimethyl sulfoxide (DMSO, Sigma-Aldrich), and stored in liquid nitrogen. For experiments, the testis tissues were rapidly thawed at 37°C and washed in DMEM. Recovered testicular samples were processed with a two-step enzymatic digestion, as described previously (Hermann et al., 2007; Hermann et al., 2009). Briefly, testicular tissues were first digested with collagenase type IV, followed by digestion with 0.25% trypsin, 0.38 g/L of EDTA, and 1.4 mg/ml DNase I for 5 min. The digestion was stopped by adding DMEM with 10% FBS, and dissociated single cells were collected by centrifugation. Cell suspensions were seeded into gelatin-coated culture plates in DMEM/F-12 (Gibco) supplemented with 10% FBS for 24–36 h, according to a procedure previously described (Wu et al., 2009). More than 95% of somatic cells (including Sertoli cells, Leydig cells, and peritubular cells) were attached to culture plates, while non-adherent germ cells were collected by centrifuge. The germ cells were then seeded into laminin-coated culture plates in DMEM/F12 containing 1x PS, 6 mM l-Glutamine (Gibco), 100 μM β-mercaptoethanol (Sigma-Aldrich), 1xB27 (Gibco), 20 ng/ml human GDNF (Sino Biological), and 20 ng/ml human basic fibroblast growth factor (b-FGF, Sino Biological) for 2 days at 34°C in a humidified 5% CO₂ incubator. These freshly isolated laminin-binding cells were enriched with SSCs and spermatogonia. In total, ∼5×10⁵ human SSCs and spermatogonia were obtained from 8 cryopreserved biopsied samples. After 2-days culture, about 30,000 human germ cells per testis were injected for xenotransplantation.

Male nude mice (BALB/c-nu/nu) and ICR mice at 6 weeks were injected intraperitoneally with 40 mg/kg busulfan (Sigma-Aldrich) and were subsequently used as recipients 6 weeks later. Male Kit ^w/w-v mice at 12-week old were used as recipients. All animals used for this study were housed at an SPF-graded facility with healthy conditions. Germ cells to be transplanted were suspended in ∼10 μl PBS with 10% trypan blue (v:v; Invitrogen) and injected into the seminiferous tubules of recipient testes via the efferent ducts. The contralateral testis in the same mouse with mock injection with PBS and trypan blue was used as a control. PBS was used to minimize any potential effects of proteins, nutrients, or small molecules in the culture media on germ cell proliferation and development. Six weeks after transplantation, animals were euthanized, and their testes were removed for further analyses. All animal experimental procedures were conducted in accordance with the local Animal Welfare Act and Public Health Service Policy with approval from the Committee of Animal Experimental Ethics at East China Normal University (Ref #:M20170325).

Histology and IHF were performed as previously described (Zhao et al., 2018). Briefly, mouse testis samples were fixed with 4% PFA solution, paraffin-embedded, and sectioned with 4 μm thickness. Following the antigen retrieval by citrate (pH6.0, boiling for 15–20 min and cooling down for 30 min), testis sections were blocked with 1% goat serum (Abcam, ab7481) in PBS at 4°C for 12–16 h, stained with primary antibodies at 4°C for 12–16 h, washed three times (15–30 min each time) with PBS at room temperature, and then stained with goat anti-rabbit IgG AlexaFluor 568 (Invitrogen) and goat anti-mouse IgG AlexaFluor 488 (Invitrogen) at 4°C for 12–16 h, and washed three times (15–30 min each time) in dark with PBS at room temperature. Primary antibodies used in this study: mouse anti-DDX4 (Abcam, ab27591), rabbit anti-DDX4 (Abcam, ab13840), rabbit anti-NuMA (Novus Biologicals, NB100-74636), rabbit anti-GFRα1 (Abcam, ab8026), mouse anti-PCNA (Abcam, ab29), and mouse anti-PLZF (Santa Cruz, sc-28319). The fluorescein-conjugated secondary antibodies were used at 1:300 dilution. Images were obtained with a Leica confocal microscope.

For IF, cells cultured on gelatin-coated coverslips were washed twice with 1× PBS and fixed in 4% PFA for 20 min at room temperature. Alternatively, cells were dissociated by Trypsin (Gibco), placed onto slides by cytospin preparation and fixed in 4% PFA for 20 min at room temperature. IF was performed as previously described (Zhao et al., 2018). Briefly, after treatment with 0.2% Triton-100 for 15 min, fixed cells were stained with primary antibodies at 4°C for 12–16 h, washed three times with PBS (10 min every time), and then stained with goat anti-rabbit IgG AlexaFluor 568 (Invitrogen) at 1:300 dilution at 4°C for 12–16 h, and washed three times (10 min each time) in the dark with PBS at room temperature. Primary antibodies used in this study: DDX4 (Abcam, ab13840), GPR125 (GeneTex, GTX51219), and PLZF (R&D, MAB2944). Images were obtained with a Leica fluorescent microscope.

All PSC culture and differentiation were performed as previously described (Zhao et al., 2018). Briefly, human embryonic stem cell line H1 (WiCell) was maintained in chemically defined Essential 8 medium (Stem Cell Technologies). Cells were passaged every 5–7 days and dissociated by 1 mg/ml Collagenase IV (Millipore). ESCs were induced when reaching 80–90% confluence in differentiation medium [α-MEM (Gibco) containing 2 mM l-glutamine (Gibco), 1× Insulin-Transferrin-Selenium-X (Gibco), 0.2% KnockOut SR XenoFree CTS (Gibco), 1 ng/ml human b-FGF, 20 ng/ml human GDNF (Sino Biological), 0.2% chemically defined lipid concentrate, and 200 μg/ml vitamin C (Sigma)] (Zhao et al., 2018). Medium was changed every day. No passage of cells was performed during differentiation. By day 12, most cells were PLZF+ SLCs. SLCs were collected for analyses or sorted for transplantation on day 12 of ESC differentiation.

Mouse peripheral blood samples were obtained via tail tip into anticoagulant EDTA-containing tubes. After being treated with red blood cell lysis buffer (Beyotime Biotechnology), blood cells were pelleted at 1,200 rpm for 15 min and re-suspended in DMEM with 10% FBS, followed by incubation with PE-CD3 (Cat. #: 561824), PE-Cy7-CD8 (Cat. #: 561097), FITC-CD4 (Cat. #: 561828) or FITC-CD19 (Cat. #: 561740) antibodies from BD Pharmingen. Subsequently, cells were washed with DMEM and analyzed with a BD Fortessa analyzer (BD Biosciences).

Cells at day 12 of PSC differentiation were prepared for flow cytometry analyses as described previously (Zhao et al., 2018). Briefly, cells were dissociated by TrypLE (Gibco), fixed by 4% PFA for 20 min at room temperature, suspended in FACS buffer (PBS with 5% fetal calf serum, 0.2% Triton-100, and 0.5% Tween 20), and centrifuged at 500×g for 5 min before incubation with antibodies. Antibodies used for examining PSC-derived SLCs: EpCAM (Biolegend, 324203), TNAP (Biolegend, 327305), TRA-1-81 (Santa Cruz, Sc-21706), INTEGRINα6 (Biolegend, 313615) and PLZF (R&D, MAB2944). Subsequently, cells were washed with DMEM and analyzed with a BD Fortessa analyzer (BD Biosciences).

To collect SLCs from in vitro differentiated PSCs by flow cytometry, dissociated cells were stained at 4°C for 45–60 min in DMEM containing antibodies and then sorted by a FACS Aria II (BD Biosciences). Antibodies used for SLC sorting: EpCAM (Biolegend, 324203), TNAP (Biolegend, 327305) and CD90 (Abcam, ab133350).

Total RNAs were extracted with Trizol (Thermo Fisher Scientific) and cDNAs were synthesized using a PrimeScript^® RT reagent Kit (TaKaRa) following manufacturer’s protocols, as previously described (Zhang et al., 2016). Gapdh was used as the house-keeping gene to normalize various sorted cell populations. The primer sequences used in this assay are provided in Supplementary Table S1.

Data were presented as mean ± standard error (SEM). Unpaired Student’s t-test was conducted to examine between-group differences using GraphPad Prism. All experiments were performed independently for more than three times unless otherwise stated. *: p < 0.05; **: p < 0.01.

To obtain in vivo developed human spermatogonia, human testicular samples were biopsied from obstructive azoospermia (OA) patients, and the existence of undifferentiated spermatogonia in these samples was confirmed by immunohistofluorescence (IHF) with an antibody against a SSC marker GFRα1 (Figure 1A). Biopsied testis tissues were processed with two-step enzymatic digestion as described previously (Hermann et al., 2007; Hermann et al., 2009). Non-adherent germ cells were separated from somatic supporting cells (e.g., Sertoli cells) that were attached on gelatin-coated plates. These germ cells were further seeded onto laminin-coated dishes (Figure 1B) and cultured for 2 days to enrich for spermatogonia before immunofluorescence (IF) assay and transplantation. More than 95% (based on 10 fields of view) of the enriched cell population was stained with an antibody against a germ cell-specific protein, DDX4, supporting their germ cell identity (Figure 1C). In addition, these cells contain ∼80% undifferentiated spermatogonia (based on 10 fields of view), as labeled by the high expression of SSC markers, GPR125 (Figure 1D) and PLZF (Figure 1E) (Costoya et al., 2004; Seandel et al., 2007; He et al., 2012). In total, human germ cells were collected from eight OA patients.

To explore whether the immune deficiency enhances the engraftment of human spermatogonia, three mouse strains, including nude mice, ICR, and Kit ^w/w-v male mice, were used as xenotransplantation recipients. Busulfan-treated nude mice (Supplementary Figure S1A) and ICR mice (Supplementary Figure S1B) represent immunocompromised and immune-competent recipients, respectively. The ablation effects of endogenous germ cells in these mice by busulfan were confirmed by histology at week 6 post-treatment (Supplementary Figure S1A, B). Preparation of infertile mouse recipients using busulfan is time-consuming and may potentially introduce confounding factors when residual endogenous germ cells are left or recovered in recipients. Therefore, Kit ^w/w-v genetically infertile mice, which lack endogenous spermatogenesis (Nocka et al., 1990; Reis et al., 2000), were also tested as potential xenotransplantation recipients. Although KIT plays a crucial role in lymphocyte development (Rodewald et al., 1997; Waskow et al., 2002), we observed comparable percentages of CD3⁺ T cells (Figure 2A), CD8⁺ T cells (Figure 2B), and CD4⁺ T cells (Figure 2C), with moderately decreased CD19 ⁺ B cells (Figure 2D) in Kit ^w/w-v mice with those in C57BL/6J wildtype mice (Supplementary Figure S1C), suggesting that Kit ^w/w-v mice contain intact adaptive immunity.

About 30,000 human germ cells from OA patients were injected per testis for xenotransplantation. The contralateral testis in the same mouse with mock injection was used as a control. In total, we injected 5 nude mice and 5 ICR mice, both busulfan-treated, as well as 6 Kit ^w/w-v mice. Six weeks after transplantation, recipient testes were dissected for histology and IHF. We found germ cells in about 3% seminiferous tubules in ∼60% of recipient mice (3 for nude and ICR mice, and 4 for Kit ^w/w-v mice) with human spermatogonial injection, and some of them were colonized at the basal membranes (Supplementary Figure S2). By contrast, no germ cells were detected in the contralateral testes with mock injection (Supplementary Figure S2).

It has reported that endogenous spermatogenesis may recover from busulfan treatment (Morimoto et al., 2021), and occasionally, residue spermatogonia may also exist in Kit ^w/w-v testes (Kubota et al., 2009). To assess the origin of germ cells in these transplanted mice, NuMA (Merdes and Cleveland, 1998; Taimen et al., 2004), an antibody against human cells, was used. We confirmed that NuMA could specifically detect human but not mouse cells by Western blotting, IHF, and IF (Supplementary Figure S3). We performed IHF with this NuMA antibody and an antibody against DDX4, a germ cell-specific protein (Tanaka et al., 2000; Castrillon et al., 2000). No DDX4+/NuMA+ cells were detected in control testes from busulfan-treated nude mice. By contrast, multiple DDX4+/NuMA+ cells were clearly observed at the basal membrane of the contralateral testes from the same mice with transplantation of human germ cells (Figure 3A, Supplementary Figure S4A). Similar results were observed in both busulfan-treated ICR mice (Figure 3B, Supplementary Figure S4B) and Kit ^w/w-v mice (Figure 3C, Supplementary Figure S4C). Because only undifferentiated spermatogonia would survive long-term in mouse testes, these DDX4+/NuMA+ germ cells were likely human spermatogonia. Indeed, all DDX4+ cells also highly expressed the SSC marker, PLZF (Costoya et al., 2004) (Figure 3D). We did not observe significant difference in the number of colonized spermatogonia across these three types of recipients, suggesting that successful engraftments can be achieved with germ cells developed in vivo from human testes using both immune-competent and immune-compromised recipients.

In a previous study, we robustly derived SLCs from human PSCs (Zhao et al., 2018). These SLCs highly express human spermatogonia-specific genes, including PLZF, CD90, GPR125, DRMT1, and DMRT3 (Zhao et al., 2018). We thus further investigated whether these in vitro PSC-derived SLCs behave similarly to in vivo developed spermatogonia from human testes. However, upon injection into testes of busulfan-treated nude mice with whole PSC differentiated population, residual undifferentiated PSCs often formed teratoma (data not shown). To exclude contaminating PSCs in transplanted populations, we need to identify appropriate surface antigens to collect SLCs with flow cytometry.

In our SLCs derived from PSCs, PLZF is the most prominently expressed spermatogonial protein (Zhao et al., 2018). Because PLZF is a transcription factor and located in the nucleus, it cannot be used for live-cell sorting. However, we could use PLZF as a marker for spermatogonia to identify surface antigens that are enriched in SLCs but not in PSCs with flow cytometry on fixed cells. Because TNAP (tissue non-specific alkaline phosphatase) and EpCAM were reported to be highly expressed in primordial germ cells and pro-spermatogonia (Sasaki et al., 2015), we analyzed their co-existence with PLZF in SLCs differentiated from a human ESC line, H1 (Supplementary Figure S5). We found that the majority (92.9%) of PLZF+ SLCs were stained as EpCAM negative (-) and TNAP positive (+) (Figure 4A). In addition, 97.4% of EpCAM-/TNAP+ cell population from differentiated PSCs were positive for PLZF (Figure 4B) and INTEGRINα6 (Figure 4C), a common spermatogonial marker in mouse, monkey and human (Shinohara et al., 1999; Maki et al., 2009; He et al., 2010). Further, TRA-1-81 is a marker for undifferentiated PSCs. We found that 86.2% of TRA-1-81 positive cells were co-stained with EpCAM+/TNAP+ population, while 92.9% of TRA-1-81 negative cells, which lost pluripotency, were EpCAM-/TNAP+ (Figure 4D). These data suggest that we may use EpCAM-/TNAP+ to enrich SLCs and to exclude TRA-1-81 + undifferentiated PSCs (Figure 4E).

We further evaluated the gene expression of cell populations separated by differential EpCAM and TNAP expressions (Figure 4F) with real-time RT-PCR assays. Because CD90 was reported to be spermatogonial marker in mice, primates and humans (Kubota et al., 2003; Maki et al., 2009; He et al., 2010), we also analyzed the cells sorted by EpCAM and CD90 staining (Figure 4G). We confirmed that the transcript level of EpCAM was dramatically lower while TNAP or CD90 was significantly upregulated in both EpCAM-/TNAP+ and EpCAM-/CD90+ populations, compared to those from the whole differentiated PSC populations (Figures 4H,I), supporting reliable sorting results. We also observed that the expression of TKTL1, PLZF, and DMRT3, markers specific for spermatogonia, were significantly elevated in both EpCAM-/TNAP+ and EpCAM-/CD90+ SLCs, with a relatively higher PLZF transcript level in EpCAM-/TNAP+ population (Figures 4H,I). These data support the usage of EpCAM-/TNAP+ population to represent PLZF+ SLCs derived from PSCs.

We next injected ∼100,000 EpCAM-/TNAP+ SLCs per testis into seminiferous tubules of Kit ^w/w-v mice and busulfan-treated nude mice. The contralateral testis from the same mouse was injected with PBS as a control. No teratoma were found in any of 6 Kit ^w/w-v mice and 4 busulfan-treated nude mice we transplanted, indicating that these SLCs lost pluripotency during PSC differentiation. We found no DDX4+/NuMA+ human germ cells in the control testes of transplanted recipients by IHF. By contrast, in 2 nude mice and all 6 Kit ^w/w-v recipients, DDX4+/NuMA+ germ cells were clearly observed at the basement membrane of the other testis from the same mouse 6 weeks post-injection of EpCAM-/TNAP+ SLCs (Figure 5A). Interestingly, compared to engrafted nude mice (with ∼4% of seminiferous tubules containing human germ cells), much more DDX4+/NuMA+ cells were detected in Kit ^w/w-v testes with ∼20% human germ cell-containing seminiferous tubules (Figure 5B, Supplementary Figure S6, 7), suggesting that Kit ^w/w-v mice provided a suitable microenvironment for the survival of human PSC-derived spermatogonia. In addition, these DDX4+ germ cells in Kit ^w/w-v testes displayed high expression levels of the proliferation marker, proliferating cell nuclear antigen (PCNA) (Figure 5C), suggesting that these human SLCs could go through mitotic division in mouse testes. Taken together, our data provide evidence that SLCs derived from human PSCs are indeed undifferentiated spermatogonia, and they maintain their potential to proliferate and repopulate mouse testes upon xenotransplantation.