Introduction

Through spermatogenesis, spermatozoa are generated from spermatogenic cells that are originated from spermatogonial stem cells (SSCs). It is critical for this process to be continuous and successful that SSCs are maintained in a homeostatic balance between self-renewal and differentiation (1). The SSCs, as the least differentiated spermatogonia, belong to a subgroup of undifferentiated spermatogonia (uSPG) that are morphologically categorized into three subtypes, i.e., A_single (As), A_paired (Apr), and A_aligned (A_al) cells (2). So far, three models have been proposed for the mechanism underlying SSCs’ self-renew based on the dynamic transitions among subgroups. In the ‘A_s model’, A_s spermatogonia serve as SSCs that are capable of both self-renew and further transformation into A_pr and A_al that eventually give rise to spermatogonia (3, 4). Later, based on the discovery of Ngn3 and Gfrα1 as SSCs markers, the ‘fragmentation model’ suggests all three subgroups with stem cell potential and the SSCs renewal is achieved through the fragmentation of pairs and chains (5). Further work on SSCs markers such as ID4 and PAX7 inspired the ‘hierarchical A_s model’, in which only specific A_s spermatogonia possess the potential for long-term self-renewal whereas the majority are restricted in their capacity (6, 7). Though their standing points of view differ, each model seems well supported by the respective collection of evidence, which to some extent reflects the nature of heterogeneity and dynamics among SSCs subpopulations.

In recent years, great insights into SSCs behaviors and regulations have been provided by a body of pioneer works, especially with recent advances in single-cell gene-expression profiling, highlighting great heterogeneity of SSCs and focusing on characterizing the nature of SSCs states especially for seeking the primitive subgroup among them. Within the population of uSPG, a number of genes relatively higher expressed in primitive subfractions have been identified and well investigated, i.e., Gfra1, ID4, Ret, Eomes, Pax7, Nanos2, Shisa6, T, Pdx1, Lhx1, Egr2 and Plvap (5–15). Particularly, Gfra1, ID4, Eomes, Pax7, Nanos2, and Plvap are further validated as the SSCs markers through lineage tracing experiment, which is considered to be a reliable method to study the origin and development of stem cells. However, some essential and primitive sub-populations remain undiscovered, and the identification of which is of great significance for elucidating the developmental process of SSCs renewal and its behavior in testis.

Adult stem cells (ASCs), as the undifferentiated primitive cells that can be found in nearly all types of tissues in mammals, are characteristic for a unique quiescent status reflected by both reversible cell cycle arrest and specific metabolic alterations (16). Putative the primitive SSCs subgroups appear to share this characteristic, as revealed in recent single-cell RNA-sequencing (scRNA-seq) analysis in humans and mice, being largely non-proliferative while capable of reciprocating between the quiescent and activated status (17–21). However, rigorous biological validation of these populations is lacking through live imaging or genetic lineage tracing, or other means. On the other hand, cells in a quiescent state are supposed to be more resilient to genotoxic insults, which shall enable the primitive SSCs to sufficiently restore spermatogenesis upon such disturbance.

Here, we identified a subpopulation of adult SSCs specifically marked by forkhead box protein C2 (FOXC2). In adult mice, spermatogenic cells derived from the FOXC2⁺ population were able to complete the whole spermatogenesis. Upon the loss of this specific subpopulation of SSCs, the undifferentiated spermatogonia pool was exhausted, eventually leading to defective spermatogenesis. Specifically, FOXC2 is required for maintaining SSCs quiescence by promoting the expression of negative regulators of cell cycle phase transition, thus symbolizing the primitive state of these adult SSCs. Moreover, the FOXC2⁺ population endured the chemical insult with busulfan and effectively restored spermatogenesis, thereby critical for keeping the reproductive homeostasis in male adult mice. Thus, our results demonstrate that FOXC2 marks the primitive SSCs subpopulation in the adult testis, and is also required for the homeostasis and regeneration of SSCs.

Results

Identification of FOXC2⁺-SSCs as the quiescent and developmental starting point of adult uSPG

We performed single-cell RNA-seq (10x genomics) of the uSPG from adult mice testes marked by THY1, a widely recognized surface marker for uSPG with self-renewing and transplantable state (22, 23), to dissect the heterogeneity and developmental trajectory (Fig.1A, Fig. S1A, B). Among 5 distinct clusters identified, Cluster1 was characterized by the high expression of stemness markers whereas other clusters were featured by progenitor or differentiating spermatogonia (dSPG) markers (Fig.1B, Fig. S1C, D). Primarily mapped to the extreme early point of the developmental trajectory, Cluster1 cells appeared quiescent and likely represented the primitive state of uSPG populations (Fig.1B, Fig. S1E-G). The top10 differentially expressed genes (DEGs) associated with Cluster1 are featured by SSCs markers such as Mcam (24), Gfra1 (5), Tcl1 and Egr2 (12, 18) (Fig.1C, Fig. S2A, Supplemental Table S1) in addition to six others expressed in different stages of germ cells and/or somatic cells, in which only FOXC2 was exclusively localized in the nucleus of a subgroup of ZBTB16⁺ uSPG (25, 26) in mice (Fig. 1D, Fig. S2B). More specifically, in adult mice, FOXC2 displayed differential expressions among various subtypes of uSPG, being more specific in As (59.9%) than other subtypes including A_pr (5.2%), A_pr- 1 (4.1%), A_al4-1 (1.83%), A_al8-1 (1.5%), and A_al16-1 (1.67%) (Fig. 1E). There was only a small fraction (5.1%) was active in proliferation as indicated by MKI67 (Fig. 1F), suggesting that FOXC2⁺ cells are primarily quiescent. Additionally, when examining the SSCs markers validated previously by lineage tracing (27), we found that FOXC2 displays a higher level of co-localization with GFRA1 and EOMES than PAX7 and NEUROG3 (28), indicating the FOXC2⁺ cells contain but differ from the known SSCs subsets (Fig. 1G).

Figure 1.

We next analyzed the expression of FOXC2 in adult human testis using the published scRNA-seq dataset (17) (GSE112013). As expected, FOXC2 was also specifically expressed in the human SSCs, most of which were MKI67^- (Fig. 1H, Fig.S2C). Pseudotime analysis showed that the FOXC2⁺ cells located at the start of the developmental trajectory with a proportion of about 90% that were MKI67^- (Fig. 1l). Immunofluorescence staining confirmed that FOXC2⁺ cells were a subset of ZBTB16⁺ spermatogonia in adult human testis, and most of them were MKI67^- (Fig. 1J), possibly representing the A_dark SSCs also known as the reserve stem cells or ‘true SSC’ in human testis(29–33). These results suggested that FOXC2 was similarly expressed in the SSCs of adult human and mouse testis and may possess a conserved function.

FOXC2⁺-SSCs can sufficiently initiate and sustain spermatogenesis

We generated Foxc2^CRE/+;R26T/G^f/f mice in which FOXC2⁺ cells were specifically labeled with GFP to enable the progeny tracing after tamoxifen treatment (Fig. S3A) (34). Tamoxifen was introduced at 2-month of age, after which the FOXC2-expressing lineage (GFP⁺) was tracked at d3 (day3), w1 (week1), w2, w4, w6, m4 (month4), m7, and m12 respectively (Fig. 2A). At d3, the tracked cells were both GFP⁺ and FOXC2⁺ (Fig. 2B) and constituted 0.027% of the total testicular cells as indicated by the fluorescence-activated cell sorting (FACS) analysis (Fig. 2C). FACS-sorted GFP⁺ cells were then transplanted into testes of recipient mice pre-treated with busulfan, in parallel to THY1⁺ cells derived from eGFP^Tg/+ mice as control. Two months after transplantation, FOXC2⁺ cells generated 5 times greater number of colonies than the THY1⁺ control (Fig. 6D, E), indicating that the FOXC2⁺ cells possess higher stemness as convinced by stronger transplantable viability.

Figure 2.

At w1, all GFP⁺ cells were identified as uSPGs, encompassing A_s, A_pr, and A_al-4 (Fig. 2F_a). Specifically, FOXC2⁺ A_s gave rise to 3 types of A_pr, i.e., FOXC2⁺/FOXC2⁺, FOXC2⁺/FOXC2^-, and FOXC2^-/FOXC2^- (Fig. 2F_{c1, b, c2, d2}), which then either produced FOXC2⁺ or FOXC2^- A_s through symmetric or asymmetric division (Fig. 2F_{c3, d1, f1}), or developed into A_al with no more than one FOXC2⁺ cell in the chains (Fig. 2F_{e, f2}). These results confirmed that FOXC2⁺ cells were capable of self-renewal to sustain the population as well as replenishing the uSPG pool by producing downstream progenies, thereby serving as primitive SSCs. In the following 2-6 weeks, GFP⁺ colonies further expanded and produced GFP⁺ sperms in the epididymis, from which healthy GFP⁺ offspring were given birth by C57 female recipients (Fig. 2G). The GFP⁺ colonies constituted 83.67%, 90.48%, 96.78%, 98.55%, and 99.31% of the total length of the seminiferous tubules at w6, m2, m4, m7, and m12 respectively (Fig. 2H, I). All offspring were GFP⁺ from m4 onwards (Fig. 2J). Additionally, the EOMES⁺, GFRA1⁺ and PAX7⁺ cells were all GFP⁺ at w2, further confirming these progenies were derived from the FOXC2⁺ cells (Fig. 2K). Overall, FOXC2⁺-SSCs can produce all subtypes of uSPG, thus initiating spermatogenesis in adult mice.

Specific ablation of the FOXC2⁺-SSC results in depletion of the uSPG pool

We then prepared Foxc2^Cre/+;R26^DTA/+ mice to investigate the physiological requirement of FOXC2⁺-SSCs in spermatogenesis (34). FOXC2⁺ population in 2-month-old mice was specifically ablated with tamoxifen-induced diphtheria toxin (DTA). The testes of these mice were examined at day3, day7, and day14 post tamoxifen induction (Fig. 3A). Gradual loss of weight in testes coincided with the reduction in the size of testes in all the mice while body weight was maintained (Fig. 3B, C). Specifically, at d3, there were no detectable FOXC2⁺ cells in addition to the decrease in the number of GFRA1⁺, LIN28A⁺ (35) and ZBTB16⁺ uSPG at the basement membrane of seminiferous tubules; at d14, all GFRA1⁺, LIN28A⁺ and ZBTB16⁺ uSPG disappeared while vacuoles formed at the basement membrane with remaining spermatocytes and spermatids in the seminiferous lumen (Fig. 3D-F, Fig. S3B). Meanwhile, the expression of DDX4 (36) and DAZL (37) as germ cell markers was significantly reduced along with nearly undetectable expression of uSPG markers such as ZBTB16, LIN28A, GFRA1, RET, and NEUROG3 (28) (Fig. 3G). These results indicate an uSPG exhaustion as the result of the FOXC2⁺-SSCs ablation, therefore supporting the critical role in spermatogenesis played by FOXC2⁺ population.

Figure 3.

FOXC2⁺-SSCs are resilient to genotoxin and indispensable for germline regeneration

Next, we examined the regenerative viability of FOXC2⁺-SSCs. At d20 post busulfan treatment (20mg/kg), FOXC2⁺ cells constituted the majority of uSPGs (Fig. 4A). Following a sharp decrease in cell number in the first five days, ZBTB16⁺ and GFRA1⁺ cells began to recover from d25 while the number of FOXC2⁺ cells remained stable (Fig. 4B), indicating that this population is insensitive to busulfan. We then checked changes in the proportion of MKI67⁺ cells, active in proliferation, in FOXC2⁺ population after busulfan treatment (Fig. 4C, D). At d30, the MKI67⁺ proportion rose to 15.92%, indicating a higher level of proliferation, albeit the total cell number stayed static (Fig. 4B, D), thereby becoming the driving force in restoring spermatogenesis. Up to d120, the MKI67⁺ proportion had settled gradually back to the pre-treatment level, accompanied by the full recovery of spermatogenesis (Fig. 4D). Further details of this process were revealed during lineage tracing (Fig. 4E). Three days after tamoxifen induction, the 2-month-old Foxc2^CRE/+;R26T/G^f/f mice were treated with busulfan. Consistent with the results above, at d20, the survived uSPG were predominantly GFP⁺ (Fig. 4F). Over 68.5% of the total length of the seminiferous tubules were GFP⁺ at m2, and this proportion rose to 95.43%, 98.41%, and 99.27% at m4, m7, and m12 respectively (Fig. 4G, H), which was comparable to the proportion by tamoxifen induction alone (Fig. 2I). From m4 onwards, nearly all germ cells, spermatids, and their offspring were GFP⁺ (Fig. 4G, I). Together, these results confirmed that FOXC2⁺-SSCs are indispensable for germline regeneration that is central to spermatogenesis recovery from interruptions.

Figure 4.

FOXC2 is essential for SSCs maintenance in adult mice

We then focused on dissecting FOXC2’s role in the SSCs maintenance using Foxc2^f/-;Ddx4-cre mice (38) (Fig. 5A). No significant difference was observed in the expressions of various uSPG markers, including ZBTB16 and LIN28A, between Foxc2^f/-; Ddx4-cre and Foxc2^f/+ mice at the age of 1 week (Fig. S4B). However, adult Foxc2^f/-;Ddx4-cre mice displayed clear testis weight loss without significant body weight loss (Fig. 5B, C). Moreover, in these mice, we observed severe degeneration of seminiferous tubules, reduced number of spermatids in the epididymis, and decreased size of the uSPG population with age (Fig. 5D–G) but without apparent signs of apoptosis (Fig. S5B). The 6-month-old Foxc2^f/-;Ddx4- cre mice were infertile, in which over 95% seminiferous tubules were Sertoli-only with hardly detectable expressions of DAZL, DDX4, LIN28A, and ZBTB16 (Fig. 5D–F, H). Therefore, FOXC2 is essential for maintaining the SSCs homeostasis and normal spermatogenesis in adult mice.

Figure 5.

FOXC2 maintains the SSCs homeostasis via negative regulation of cell cycle

We collected THY1⁺ uSPGs from 4-month-old Foxc2^f/+ and Foxc2^f/-; Ddx4-cre mice and compared their transcriptome signatures revealed from scRNA-seq (Fig. 6A). The pseudotime analysis identified Cluster1, which represented the FOXC2-expressing SSCs in Foxc2^f/+ mice corresponding to the FOXC2-deleting SSCs in the Foxc2^f/-; Ddx4-cre mice, was specifically assigned to the extremely early stage of the development trajectory in respective samples, which was validated by the expression of corresponding markers (Fig. 6B, Fig. S5A, B). Aggregated analysis of the overall uSPG populations showed that cells derived from Foxc2^f/-; Ddx4-cre mice were specifically associated with the late stage of the development trajectory, as opposed to Foxc2^f/+ mice where nearly all the cells derived were concentrated at the early stage of development (Fig. 6C, Fig. S5C). This implies that the loss of Foxc2 prompts the SSCs to progress into a more differentiated stage with defection in maintaining the primitive identity of SSCs. Further analysis of the cells in Cluster1 revealed two distinct subclusters, i.e., Subclusters0 and Subclusters1 (Fig. S6A). Formed primarily by the Cluster1 cells derived from Foxc2^f/+ mice, Subclusters0 was featured by stemness markers, while Subcluster1, representing the majority of Cluster1 cells from Foxc2^f/-; Ddx4-cre mice, was featured by progenitor markers (Fig. S6B, C). Consistently, pseudotime analysis showed that Cluster1 cells from Foxc2^f/+ mice projected a forward stage of the developmental trajectory indicated by stemness markers, whereas Cluster1 cells from Foxc2^f/-; Ddx4-cre mice were associated with a later stage of the developmental trajectory (Fig.6D, Fig. S6D, E). More specifically, less number of cells were found at the starting state1 in Cluster1 from Foxc2^f/-; Ddx4-cre mice than in Foxc2^f/+ mice, with rather more cells in the developmental progression (from state1 to state5), especially at the advanced state5 (Fig. 6E). Thus, FOXC2 deletion caused defective SSCs maintenance and committed the primitive SSCs to a differentiation destiny. Further, there were 932 genes down-regulated in Cluster1 cells derived from Foxc2^f/-; Ddx4-cre mice in comparison to Foxc2^f/+ mice (Fig. 6F, Supplemental Table S2), which were functionally associated with both stem cell population maintenance and mitotic cell cycle (Fig. 6G). Consistently, the GSEA analysis revealed a more progressive cell cycle in Cluster1 upon Foxc2-knockout (Fig. 6H), confirming the role of FOXC2 in regulating the cell cycle of the primitive SSCs.

Figure 6.

We then performed Cleavage Under Targets and Tagmentation (CUT&Tag) sequencing to explore the underlying mechanism (39, 40), for which GFP⁺ SSCs from Foxc2C^RE/+;R26T/G^f/f mice 3 days after tamoxifen induction, representing the FOXC2⁺-SSCs, were isolated for CUT&Tag sequencing (Fig. 7A). Specific peaks enriched in the promoter region of 3629 genes (Fig. 7B, C; Supplemental Table S2) showed functional enrichment in biological processes such as DNA repair and mitotic cell cycle regulation (Fig. 7D). By overlapping with the 932 genes down-regulated in Cluster1 cells from Foxc2^f/-; Ddx4-cre mice, we obtained 306 genes as the candidates subjective to the regulation by FOXC2 (Fig. 7E; Supplemental Table S2). Further, GO enrichment analysis of these genes highlighted a distinctive functional cluster (11 genes) focusing on the negative regulation of cell cycle (Fig. 7F; Supplemental Table S3) (41–50). More specifically, significant peaks enrichment at the promoter region were observed for these candidate genes (Fig. 7G). Meanwhile, as predicted using the JASPAR Scan function (binding potential >0.8), there showed strong binding potential of FOXC2 towards these candidate genes (Fig. 7I) via the binding motif of FOXC2 (Fig. 7H), which was further confirmed by the results from the CUT&Tag qPCR (Fig. 7J). Overall results implied that FOXC2 may function as a gatekeeper that ensures the quiescent state of the primitive SSCs by impeding cell cycle progression.

Figure 7.

Discussion

In this work, a comprehensive analysis of uSPG populations with scRNA-seq and the following lineage tracing study by whole-mount immunofluorescence assay led to the identification of FOXC2-expressing SSCs as an important and primitive SSCs subpopulation in adult mice. Further investigation through functionality analysis confirmed FOXC2 is essential for SSCs self-renewal and stemness, thereby is required for maintaining the SSCs population that is critical for continuous spermatogenesis. Importantly, our data demonstrated that the colonies formed by FOXC2⁺ cells constituted nearly the total length of the seminiferous tubules (99.31 %), implying that the FOXC2⁺- SSCs can support the complete spermatogenesis in adult mice.

GFRA1⁺ A_pr and A_al cells were found to break randomly and a portion of them can return to the stem cell state (5). Interestingly, our findings showed FOXC2 appeared in one of the A_pr or A_al cells at times, therefore raising a possibility that the subset of GFRA1⁺ cells that return to stem cell state after intercellular bridge break, maybe FOXC2⁺ due to different cell cycle state. If so, based on both findings, GFRA1⁺FOXC2⁺ could represent a quiescent state whereas GFRA1⁺FOXC2^- is proliferate active, which certainly requires further validation possibly through multiple lineage tracing and live imaging.

We observed that the FOXC2⁺-SSCs were almost all in a non-proliferative state (~94.9%), and further revealed that FOXC2 functioned in the negative regulation of cell cycle progression, thus confirming that FOXC2-expressing SSCs are quiescent SSCs population in adult mice. The finding that FOXC2 inhibited cell cycle and differentiation of SSCs in testis is consistent with that reported in other tissues (51, 52). In general, the quiescent state is a protective mechanism for stem cell storage and prevents stem cells from damage or depletion under genotoxic stresses (1, 53–55). In our work, after the busulfan treatment, the quantity of FOXC2⁺ cells remained stable and the survived uSPGs were predominantly FOXC2⁺, indicating its insensitivity to cytotoxic agents. However, the proportion of MKI67⁺FOXC2⁺ cells increased by 15.92% after 30 days of the busulfan treatment and decreased back to the pre-treatment level (5.08%) at 120 days, implying that the quiescent FOXC2⁺ cells were able to transform into the proliferative FOXC2⁺ cells to replenish the SSCs pool to maintain the SSCs homeostasis and normal spermatogenesis. We further confirmed by lineage tracing analysis that FOXC2-expressing cells were the only remaining SSCs population and were responsible for germline regeneration after the busulfan treatment, indicating that FOXC2⁺-SSCs represent a functionally important stem cell population with regenerative ability. In the future, more insights into the unique regulation of SSCs can be drawn from studying and comparing the transition between the quiescent and proliferative states in FOXC2⁺ and other SSCs subpopulations.

According to our findings, we proposed a model for the maintenance of the FOXC2+ SSCs subpopulation (Fig. 7K). Under physiological conditions, FOXC2⁺ A_s cells (including FOXC2⁺GFRA1⁺, FOXC2⁺EOMES⁺ cells, etc.) constitute the primitive population of SSCs, of which only a small proportion (~5.1%) cells are proliferative while the majority remains quiescent (Fig. 7Ka). This primitive population can divide symmetrically or asymmetrically into different A_pr and A_al (Fig. 7Kb). Then FOXC2⁺ cells (Fig. 7Kb) may break from the syncytial and return to A_s state (Fig. 7Ka) to maintain the stable number of the primitive SSCs. FOXC2^- progenies, derived from the FOXC2⁺ primitive population, form a transit amplification (TA) SSCs pool (Fig. 7Kc) to support spermatogenesis. However, it requires continuous supply from the FOXC2⁺ population and is subject to exhaustion when the supply is disrupted. In the context of regeneration conditions, the FOXC2⁺MKI67^- cells can survive and set out the recovery process (Fig. 7Kd). At the early stage, increasing proportions of FOXC2⁺MKI67^- cells are committed to transforming into proliferative FOXC2⁺MKI67⁺ cells, strengthening the supply to the TA SSCs pool (Fig. 7Ke). At the late recovery stage, MKI67⁺/MKI67^- ratio returns to the physiological level in FOXC2⁺ population (Fig. 7Ka), leaving the total number of FOXC2⁺ cells stable therefore maintaining the SSCs homeostasis. However, it is necessary to perform more investigation to further improve and modify this model to gain a complete understanding of the connections between different primitive SSCs subpopulations via lineage tracing assays in the testes of adult mice.

Based on our observation, FOXC2 seems nonessential for the transformation from gonocytes to SSCs in infant mice, in contrast to its requirement for adult spermatogenesis. A recent study showed that FOXC2 was present in a fraction of A8 and A16 cells in the postnatal mouse testis (<5 weeks), however, this FOXC2⁺ subpopulation appeared more active in proliferation than the adult counterpart (56). Such differential functionality might reflect the difference in the physical nature of spermatogenesis between developmental stages. For example, the maturity of spermatogenesis is still under development during the juvenile period with a focus on expanding the SSCs pool. Therefore, it would be interesting to explore differences in individual functional contexts as well as the underlying regulatory mechanisms. Meanwhile, FOXC2, highly conserved between mice and humans with 94% identity in amino acid sequence (57), is also expressed in a subset of human adult SSCs, raising the possibility of an evolutionarily conserved mechanism governing SSCs homeostasis in humans. Further work following this direction might be of great clinical significance specifically to patients who suffer from infertility. Moreover, the developmental correlation between FOXC2⁺-SSCs and other SSCs subpopulations proposed previously should be revealed via biological methods such as multiple lineage tracing and live imaging. Collectively, our work here provides new insights into the investigation of adult SSCs and serves as a reference for studying the homeostasis and regeneration of other stem-cell systems.

Materials and Methods

Data Availability

All data are available in the main text or supplementary materials. The scRNA-seq and CUT&Tag sequencing data have been uploaded to the GEO with accession codes GSE183163, GSE180729, and GSE180926. All of the R packages were available online and the code was used according to respective R packages documentation as described in the Methods. The MSigDB (v.7.0) used in this study is available at https://www.gsea-msigdb.org/gsea/msigdb.

Additional Experimental Procedures

The procedures for mice, magnetic-activated cell sorting (MACS), single-cell RNA-seq, single-cell RNA-seq data processing, CUT & TAG sequencing and analysis, enrichment analyses, transplantation assay, fluorescence-activated cell sorting (FACS), immunofluorescence, RNA isolation and quantitative RT-PCR analysis, tamoxifen inducible, analyses of cell density, sperm counts, histology, evaluation of degenerating tubules, and statistical analysis are presented in the Supplemental Materials and Methods.

Acknowledgements

This work was supported by the National Key Research and Development Program of China grant (2022YFA0806302, 2018YFC1003500, 2019YFA0801800), CAMS Innovation Fund for Medical Sciences (CIFMS, 2021-I2M-1-019, 2017-I2M-3-009), National Natural Science Foundation of China grant (31970794, 32000586, 31725013, 32200646), and State Key Laboratory Special fund from the Ministry of Science grant (2060204).

Supplemental Figures

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Figure S5.

Figure S6.

Supplemental Tables (separate files)

Supplemental Table S1. List of the top30 differentially expressed genes of different clusters.

Supplemental Table S2. List of the differentially expressed genes found by CUT&Tag sequencing and scRNA-seq respectively and their respective enriched Gene Ontology terms.

Supplemental Table S3. List of the Gene Ontology terms of the 306 crossed candidates.

Supplemental Table S4. Primers and antibodies used in this study.

Supplemental Materials and Methods

Mice

Animal experiments were approved by the Committee on Animal Care of the Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences and Peking Union Medical College. The 8-week-old C57BL/6J wild-type mice were used for magnetic-activated cell sorting. The Rosa26mTmG^flox mice (stock no. 007676), Ddx4-Cre mice (stock no. 000692) and EGFP^Tg/+ mice (stock no. 021930) were bought from the Jackson Laboratory. The Foxc2^iCreERt2 mice and the Foxc2^flox/flox (Foxc2^f/f) mice were constructed and bought from the Biocytogen. The Rosa-eGFP-DTA (R26^DTA/⁺) mice were bought from GemPharmatech. All mice were housed and bred under specific pathogen-free conditions (temperature: 22-26°C, humidity: 40-55%, 12-h light/dark cycle) in the animal facility at the Institute of Basic Medical Sciences. DNA was isolated from the tails, and the genotypes of the mice were checked using PCR with specific primers (Supplemental Table S4). All mice were randomly assigned to experiments and no statistical methods were used to predetermine sample size. The person performing the experiments did not know the sample identity until after data analysis. No data were excluded from analyses and the data displayed included a minimum of three independent experiments and a minimum of three biological replicates for each independent experiment. The 8-week-old C57BL/6J WT mice were treated with busulfan (40 mg/kg) and used as recipient mice 1 month later.

Magnetic-activated cell sorting (MACS)

The testes from 8-week-old C57BL/6J wild-type mice or 4-month-old Foxc2^f/+ and Foxc2^f/-; Ddx4-cre mice (n=4) were minced and digested in the collagenase type IV (1mg/mL, Sigma) and DNase I (500μg/mL, Sigma) at 37°C for 15 min. The cell suspension was pipetted up and down once every 5 minutes and the digestion process was stopped with DMEM (containing 10% FBS). The cell suspension was filtered through a 40-μm nylon mesh, and after centrifugation, the cells were resuspended in 8mL PBS. The 15 mL conical centrifuge tubes were slowly overlayed with 2 mL of 70% Percoll solution, 2 mL of 30% Percoll solution, and then 2 mL of testicular cell suspension and centrifuge at 600 × g for 10 min at 4 °C without using the centrifuge brake. After centrifugation, the cells at the interface between the 70% and the 30% Percoll solution were carefully removed into the new conical centrifuge tubes, washed with PBS, and then centrifuge at 600 × g for 10 min at 4 °C. After centrifugation, the cells were resuspended in 360μL MACS buffer, added with 40μL of magnetic microbeads conjugated with anti-Thy-1 antibody (Miltenyi Biotec 130–049–101, Auburn, CA), and mixed well. Incubate the cell suspension containing Thy-1 microbeads for 20 min at 4 °C. Mix gently by tapping every 10 min. Add 20 mL of MACS buffer to the tube to dilute Thy-1 microbeads and centrifuge at 300 ×g for 10 min at 4 °C. Remove the supernatant completely and resuspend in 2 mL of MACS buffer. Place the separation columns (MS Column; Miltenyi Biotec 130-042-201) in the magnetic field of the mini MACS Separation Unit (Miltenyi Biotec 130-142-102) and rinse with 0.5 mL of MACS buffer. Apply the cell suspension to the columns (500μL/ column). After the cell suspension has passed through the column and the column reservoir is empty, wash the column with 0.5mL of MACS buffer three times. Remove the column from the MACS Separation Unit and elute the magnetically retained cells slowly into a 50 mL conical centrifuge tube with 1mL of MACS buffer using the plunger supplied with the column. Centrifuge the tube containing the cells at 600 × g for 10 min at 4 °C and resuspend the cell pellet with 10mL of MACS buffer for rinsing. Repeat this step once. After the final rinsing step, resuspend cells in 0.04% BSA and count the cell number.

Single-cell RNA-seq

The MACS-sorted Thy1⁺ cells were used for loading onto the Chromium Single Cell 3’ Chip kit v2 (10x Genomics, PN-120236) according to the instructions. Cell capturing and library preparation was performed following the kit instructions of the Chromium Single Cell 3’ v2 Library and Gel Bead Kit (10x Genomics, PN-120237). In brief, 5000 cells were targeted for capture, and after cDNA synthesis, 10-12 cycles were used for library amplification. The libraries were then size-selected, pooled, and sequenced on a Novaseq 6000 (Illumina). Shallow sequencing was performed to access the library quality and to adjust the subsequent sequencing depth based on the capture rate and the detected unique molecular indices (UMI).

Single-cell RNA-seq data processing

Raw sequencing reads were processed using the Cell Ranger v.3.0.1 pipeline of the 10x Genomics platform. In brief, reads from each sample were demultiplexed and aligned to the mouse mm10 genome, and UMI counts were quantified for each gene per cell to generate a gene-barcode matrix. Default parameters were used. The UMI counts were analyzed using the Seurat R Package (58) (v.3.0.1) following the Seurat pipeline. Cells with more than 200 detected genes or less than 10% mitochondria reads were retained. Genes not detected in at least 10 cells were removed from subsequent analysis. The resulting matrix was normalized, and the most variable genes were found using Seurat’s default settings, then the matrix was scaled with regression against the mitochondria reads. The top 2000 variable genes were used to perform PCA, and Jackstraw was performed using Seurat’s default settings. Variation in the cells was visualized by UMAP for the top principal components. Cell types were determined using marker genes identified from the literature (59). We used the Seurat function CellCycleScoring to determine the cell cycle phase, as this program determines the relative expression of a large set of G2-M and S-phase genes. After removing the undefined cells, the spermatogonia were used for trajectory analysis, and the single-cell pseudotime trajectory was constructed with the Monocle 2 package (v2.12.0) (60–62) according to the provided documentation. The Monocle function clusterCells was used to detect cell clusters between clusters. The Seurat function FindAllMarkers with default settings was used to find DEGs upregulated in each cluster compared to the other cells. The top200 DEGs of cluster1 were used for ordering cells, and the discriminative dimensionality reduction with trees (DDRTree) method was used to reduce the data to two dimensions. The dynamic expression patterns with the spermatogonial developmental trajectory of specific genes were visualized using the Monocle function plot_genes_in_pseudotime and plot_pseudotime_heatmap. The procession data of the adult human single-cell dataset was downloaded from Gene Expression Omnibus (GEO): GSE112013 and the UMI counts were analyzed using the Seurat R Package (v.3.0.1) following the Seurat pipeline with the same parameters and functions as mentioned previously. According to the known markers, the germ cells characterized was used for trajectory analysis, and the single-cell pseudotime trajectory was constructed with the Monocle 2 package (v2.12.0) as mentioned previously.

CUT & Tag sequencing and analysis

CUT&Tag assay was performed using CUT&Tag 2.0 High-Sensitivity Kit (Novoprotein scientific Inc., Cat# N259-YH01). The detailed procedures were described in (40, 63). In brief, cells were harvested by trypsin and enriched by ConA-magnetic beads. 10,000 cells were re-suspended in 100 mL Dig-wash Buffer (20 mM HEPES pH 7.5; 150 mM NaCl; 0.5 mM Spermidine; 13 Protease inhibitor cocktail; 0.05% Digitonin) containing 2 mM EDTA and a 1:100 dilution of primary FOXC2 antibody. The primary antibody was incubated overnight at 4°C. Beads were washed in Dig-wash Buffer 3 times and incubated with secondary antibody for 1 hour at a dilution of 1:200. After incubation, the beads were washed 3 times in Dig-Hisalt Buffer (0.05% Digitonin, 20 mM HEPES, pH 7.5, 300 mM NaCl, 0.5 mM Spermidine, 13 Protease inhibitor cocktail). Cells were incubated with proteinA-Tn5 transposome at 25°C for 1 h and washed 3 times in Dig-Hisalt buffer to remove unbound proteinA-Tn5. Next, cells were re-suspended in 100mL Tagmentation buffer (10 mM MgCl2 in Dig-Hisalt Buffer) and incubated at 37°C for 1 h. The tagmentation was terminated by adding 2.25 mL of 0.5 M EDTA, 2.75 mL of 10% SDS and 0.5 mL of 20 mg/mL Proteinase K at 55°C for 1 hour. The DNA fragments were extracted by phenol chloroform and used for sequencing on an Illumina HiSeq instrument (Illumina NovaSeq 6000) to generate 2 × 150-bp paired-end reads following the manufacturer’s instructions.

Raw reads were analyzed by removing low-quality or adaptor sequences using Trim_galore (v0.5.0) and cleaned reads were mapped to the reference genome mm10 using Bowtie2 (v2.2.5). We used MACS2 (v2.1.2) to call peaks found in different groups. Homer (v4.11.1) de novo motif discovery tool was used for finding the binding motifs of Foxc2 with the findMotifsGenome.pl command. The binding potential of candidate target genes at the binding motif was predicated using the JASPAR Scan function (binding potential >0.8). The peaks filtered by fold change more than 5 and transcription start site (TSS) less than 3000 bp were annotated by R package Chip Seeker for gene category analysis. R package Cluster profiler was used for gene function annotation such as KEGG and GO analysis.

Enrichment analyses

Gene Ontology (GO) and KEGG pathway enrichment analyses were conducted using the ClusterProfiler package (v3.12.0) (Yu et al., 2012) and the ClueGO app (v2.5.7) in Cytoscape (v3.8.1) with default settings and a p-value cut-off of 0.05. GSEA enrichment analysis was assessed using the GSEA (v4.0.2) algorithm with MSigDB (v7.0) with default settings. The signaling pathways enriched by niche-derived paracrine factors and undifferentiated SPG-derived membrane proteins in the DEGs of the four samples were characterized. Then for each niche cell type, the niche-derived signaling pathways in all four samples were crossed with the SSC-derived signaling pathways to identify the candidate signaling pathways pivotal to SSCs maintenance.

Transplantation assay

The 8-week-old C57BL/6J WT mice were treated with busulfan (40 mg/kg) and used as recipient mice 1 month later. SSCs were transplanted into the testis of recipient mice (1 x 10³ cells/testis), and two months after transplantation, the testes were harvested and observed under a fluorescence microscope.

Fluorescence-activated cell sorting (FACS)

Single-cell suspensions were generated from testes or in vitro cultured SSCs. FACS was performed using an SH800 machine (Sony Biotechnology) to isolate the GFP⁺ cells. Briefly, the GFP⁺ gating area was based on the point of the fluorescence intensity axis where cells were considered as being GFP⁺, set based on the background fluorescence intensity of a non-transgenic control testis cell population.

Immunofluorescence

Mouse testes were fixed in 4% Paraformaldehyde (PFA) at 4°C overnight, dehydrated, embedded in paraffin, and cut into 5-μm thick sections. The rehydrated mouse or human testis sections were subjected to antigen retrieval, blocked in 5% BSA with 0.1% Triton X-100, and incubated with primary antibody (Supplemental Table S4) at 4°C overnight, including the germ cell marker DDX4, undifferentiated spermatogonia markers ZBTB16, LIN28A, ECAD (64), GFRA1, EOMES, PAX7, progenitor marker NEUROG3, and spermatocyte marker SYCP3 (65). After three 5-min washes in PBS, the sections were incubated with secondary antibodies (Supplemental Table S4) and DAPI (Sigma) at 37°C for 1 h. After three 5-min washes in PBS, coverslips were then mounted on glass slides using anti-quencher fluorescence decay (Solarbio). Images were captured using a Zeiss 780 laser-scanning confocal microscope. Whole-mount immunofluorescence of seminiferous tubules was performed as previously described (66). Briefly, seminiferous tubules were disentangled from testicular biopsies and immediately fixed in 4% PFA at 4°C for 12 h. After fixation, the seminiferous tubules were permeabilized with 0.5% Triton X-100 in PBS and treated with 5% BSA in PBS overnight at 4°C. After three 30-min washes, the seminiferous tubules were incubated with primary antibody (Supplemental Table S4) overnight at 4°C. After three 30-min washes, the seminiferous tubules were incubated with species-specific secondary antibodies and DAPI at 4°C for 12 h. After three 30-min washes, the seminiferous tubules were mounted on slides with anti-quencher fluorescence decay (Solarbio) and observed with a Zeiss 780 laser-scanning confocal microscope.

RNA isolation and quantitative RT-PCR analysis

Total RNA was extracted from the testes or cultured cells using the RNeasy kit (Qiagen), reverse-transcribed using RevertAid First Strand cDNA Synthesis kit (Thermo), and processed for qRT-PCR using PowerUp SYBR Green Master Mix (Applied Biosystems) and a LightCycler 480 system (Roche) with gene-specific primers (Supplemental Table S4). Reactions were run in triplicate and the mRNA levels were normalized to Gapdh and quantified using the delta-delta Ct method. The values shown are mean ± s.e.m. from three biological replicates.

Tamoxifen inducible

According to a previous report for activation of iCre (9), the mice were fed with TD.130859 (TAM diet) for three days. The food was formulated for 400 mg tamoxifen citrate per kg diet, which would provide ~40 mg tamoxifen per kg body weight per day.

Analyses of cyst length

The cyst length was obtained according to the previous report (67). Briefly, to determine the cyst length, after immunofluorescence staining with anti-E-CAD antibody, the whole mount seminiferous tubule specimens were observed under a fluorescence microscope. The E-CAD staining coupled with staining for FOXC2 enabled us to reliably identify syncytial cysts of FOXC2⁺ cells.

Analyses of cell density

The cell density was counted according to a previous report (68). Briefly, the densities of the ZBTB16⁺, GFRA1⁺, LIN28A⁺, or FOXC2⁺ cells were measured on the seminiferous tubules with whole-mount staining, the numbers of which per 1000 Sertoli cells were determined.

Sperm counts

Total sperm counts were obtained according to the previous report (69). Briefly, epididymal caput and cauda were minced and incubated in prewarmed M16 medium (Sigma-Aldrich) at 37°C in air containing 5% CO2 for 30 min to allow the sperm to swim out. Then, the sperm were diluted in water and counted using a hemocytometer.

Histology, evaluation of degenerating tubules

Testes of WT and mutant mice were fixed with PFA fixative and processed for paraffin-embedded section preparation (5 μm thick) and hematoxylin and eosin staining, according to standard procedures. The percentage of degenerating seminiferous tubules was calculated based on the cross-sections of seminiferous tubules (n > 200) that appeared on one transverse section for each testis. In normal (WT) mouse testes, four generations of germ cells, each synchronously progressing through spermatogenesis, form cellular associations of fixed composition (called seminiferous epithelial stages). In the testes of Foxc2^flox/-; Ddx4-cre mice, a few tubule cross-sections lacked one or more out of the four germ cell layers, which was defined as “degenerative tubules” in this study.

Statistical analysis

All statistical analyses were performed using GraphPad Prism (v7.0). All experiments were repeated at least three times, and data for evaluated parameters are reported as mean ± s.e.m. The p-values were obtained using two-tailed unpaired Student’s t-tests or one-way ANOVA followed by Tukey test (ns represents p-value > 0.05, * represents p-value < 0.05, ** represents p-value < 0.01, *** represents p-value < 0.001, and **** represents p-value < 0.0001).