FGF9 promotes mouse spermatogonial stem cell proliferation mediated by p38 MAPK signalling

Abstract Objectives Fibroblast growth factor 9 (FGF9) is expressed by somatic cells in the seminiferous tubules, yet little information exists about its role in regulating spermatogonial stem cells (SSCs). Materials and Methods Fgf9 overexpression lentivirus was injected into mouse testes, and PLZF immunostaining was performed to investigate the effect of FGF9 on spermatogonia in vivo. Effect of FGF9 on SSCs was detected by transplanting cultured germ cells into tubules of testes. RNA‐seq of bulk RNA and single cell was performed to explore FGF9 working mechanisms. SB203580 was used to disrupt p38 MAPK pathway. p38 MAPK protein expression was detected by Western blot and qPCR was performed to determine different gene expression. Small interfering RNA (siRNA) was used to knock down Etv5 gene expression in germ cells. Results Overexpression of Fgf9 in vivo resulted in arrested spermatogenesis and accumulation of undifferentiated spermatogonia. Exposure of germ cell cultures to FGF9 resulted in larger numbers of SSCs over time. Inhibition of p38 MAPK phosphorylation negated the SSC growth advantage provided by FGF9. Etv5 and Bcl6b gene expressions were enhanced by FGF9 treatment. Gene knockdown of Etv5 disrupted the growth effect of FGF9 in cultured SSCs along with downstream expression of Bcl6b. Conclusions Taken together, these data indicate that FGF9 is an important regulator of SSC proliferation, operating through p38 MAPK phosphorylation and upregulating Etv5 and Bcl6b in turn.

colony-stimulating factor 1 (CSF1), which increases stem cell number. 4 However, there is still much unknown about niche regulating SSC functions.
The fibroblast growth factor (FGF) family plays an important role in the maintenance of SSCs. In humans, dysregulation of signalling pathways involved in SSC regulation by FGF receptor mutations produces a striking paternal age effect through disproportionate self-renewal of harbouring cells increasing the proportion of mutant spermatozoa. [5][6][7] FGF2 is required for normal spermatogenesis and is typically included in germ cell culture media as it increases SSC proliferation. [8][9][10][11] FGF4 enhances regeneration of testis after damage, 12 and FGF5 also promotes proliferation in culture of spermatogonia expressing GDNF family receptor alpha 1 (GFRα1). 13 However, much remains unknown about the effect of other FGFs expressed in testes on SSC regulation, such as FGF9. FGF9 expression is observed in the interstitial regions of mice from 14 to 18 days post coitum, and from postnatal days 35 to 65 in Leydig cell cytoplasm, spermatid nucleus and spermatogonium cytoplasm. 14 Immunohistochemical staining shows FGF9 presence in Leydig cells of 9-month-old sheep 15 and adult humans. 16 Infertile human patients with Sertoli cell-only syndrome show decreased FGF9 expression, 16 suggesting a potential link between aberrant FGF9 expression and normal germ cell maintenance. In culture, FGF9 increases growth of rat 17 and mouse 18 germ cells. FGF9 has been suggested to be an inhibitor of meiosis in vitro. 19,20 However, none of these studies have analysed the effect of FGF9 on SSCs via transplantation. Much remains to be understood about the role of FGF9 within the SSC niche, and as FGF9 knockout results in a complete male-to-female sex reversion, 21 its effects on spermatogonia in vivo on the adult testis have not been determined experimentally.
Determining SSC niche growth factors is vital to understand SSC homeostasis and spermatogenesis. In this paper, we determine FGF9 as an SSC growth factor using in vitro culture and transplantation. We show that FGF9 acts as a stem cell renewal factor, inhibits differentiation in vivo when overexpressed and investigate pathways involved in FGF9-mediated regulation of SSC proliferation.

| Cell culture
THY-1 + germ cell cultures were established as described previously. 8 Briefly, male C57 LacZ pups (5-8 days post-partum) were sacrificed and testes were digested using Trypsin-EDTA (Gibco, USA). Cells were magnetically sorted using CD90. Feeder cells were identified and discounted from counts based on morphology. 22 SSC number was calculated using the following formula, adapted from reference 4 : Here, 'cells seeded' indicates the number of cultured THY-1 + germ cells seeded at the beginning of each time period and 'cells harvested' is the cell count at the end. 'Colonies' indicates the number of distinct colonies counted per testis and is divided by the number of cultured THY-1 + germ cells transplanted per testis.

| Transplantation
Transplantation was performed as described previously. 23 Briefly, cultured THY-1 + germ cells were prepared and injected into the tubules of testis. 2 months later, mice were sacrificed and testes were fixed and stained with X-gal solution. Tissue processing, sectioning and staining with nuclear fast red was performed by the histology services core at the School of Veterinary Medicine, University of Pennsylvania.

| Single-cell RNA-seq and bioinformatics
Established THY-1 + germ cells (p12) were plated onto STO feeders 3 days before the start of the experiment and cultured in mSFM with normal growth factors. At hour 0, the cells were switched to mSFM with 20 ng/mL of FGF9 or 0.1% BSA. Cells were cultured for 48 hours, and germ cell clumps were removed by gentle pipetting. Clumps were treated with 0.25% Trypsin for 2 minutes to digest to single cells and strained in a 40 μm filter and blocked in PBS-S for 10 minutes. Cells were incubated at 4℃ with BioLegend TotalSeq c-KIT antibody (CITE-sEquation 24 ) following the manufacturer's protocol. Cell viability following antibody incubation was 89% and 95% for control and FGF9-treated samples, respectively. Cells were encapsulated and libraries generated using a Chromium Next GEM Single Cell 3' Kit v3.1 with Feature Barcoding (10X Genomics) per manufacturer's protocol.
One replicate of each treatment was encapsulated using cells derived from three individual mice. Libraries were sequenced on a NextSeq 500 sequencer (Illumina) using a 75-cycle high-output sequencing kit to a depth of 30k reads per cell. Data were processed using Cell Ranger (10X Genomics) using Mouse reference mm10 (GENCODE vM23/Ensembl 98). Gene counts were analysed with Seurat v3.1 25 for clustering, integration and differential gene expression and Monocle v3 26

| Supplementary methods
Additional methods are shown in supplementary material.

| Fgf9 overexpression increased PLZF-positive cells in vivo
To investigate the role of FGFs in governing SSC self-renewal, we constructed overexpression lentiviruses containing Fgf3, Fgf5, Fgf8 and Fgf9 and injected into recipient testes ( Figure 1A). Only Fgf9 overexpression plasmids for 11 weeks were substantially reduced in size relative to the uninfected testis from the same animal ( Figure 1C, Figure S1B and Figure S2). Testes injected with empty vector showed normal histology and PLZF staining of undifferentiated germ cells along the basement membrane ( Figure 1D).
Gdnf overexpression produced an increase in PLZF + cells around the basement membrane but also in aggregates within the tubule ( Figure 1D). PLZFdifferentiating cells were visible within the tubule up to round spermatids, including some SYCP3 + cells ( Figure 2A). When both Fgf9 and Gdnf were overexpressed, a mixture of both phenotypes was seen: all cells showed strong PLZF staining ( Figure 1D), no differentiating or SYCP3 + cells were seen and undifferentiated spermatogonia form clumps (Figure 2A).
When FGF9 was overexpressed, far more whole-mount staining of PLZF + cells was visible, whereas GDNF overexpression resulted in even more PLZF + cells with sporadic clumps ( Figure 2B and Figure   S3). These results suggest that excess FGF9 promotes undifferentiated spermatogonia accumulation and leads to a lack of differentiating cell types in vivo.

| Stimulation of SSCs proliferation by FGF9
In order to determine the effect of FGF9 on SSCs, we established THY-1 + germ cultures which enriched SSCs (p7-p12) and plated
The germ cell cultures showed exponential increases over time ( Figure 3A). FGF2-treatment resulted in more cell numbers over time compared with control although this did not rise to statistical significance. 1 ng/mL FGF9 showed lower levels of cell growth compared with FGF2, largely indistinguishable from controls.

F I G U R E 2
Immunofluorescence of seminiferous tubules exposed to Fgf9 and Gdnf overexpression. A, Cross-sections of seminiferous tubules stained with antibodies against PLZF (marker for undifferentiated spermatogonia), SYCP3 (essential for meiosis), DAPI (DNA stain) following injection of vector, Fgf9 and Gdnf overexpression plasmids in addition to both overexpression plasmids injected together.  Figure 3B). To assay the number of SSCs, cultured THY-1 + germ cells enriched for SSCs were transplanted into busulfan-treated mice, where SSCs produce countable colonies of germ cells carrying the LacZ marker gene. [29][30][31] For a given number of germ cells transplanted, we observed no significant differences in colony count between any of the treatments.
This indicated that while FGF9 treatment increased the total cell number it did not alter the stem cell concentration (ie, stem cells per 10 5 cultured THY-1 + germ cells, Figure 3C). Moreover, testes of all groups showed normal spermatogenesis after transplantation ( Figure S4). Taking these observations together, SSC number in culture at the three time points was calculated as described in the Methods section. FGF9 20 ng/mL showed significantly more SSCs after 4-week culture as compared with 1 ng/mL and control ( Figure 3D,E). We can conclude that FGF9 exposure leads to a greater cultured THY-1 + germ cell number and SSC number over time.

| p38 MAPK signalling is required for FGF9activated SSC proliferation
To explore how FGF9 promotes SSC proliferation, RNA-seq was performed (Table S3) and validated by qPCR ( Figure S5). The pathways that showed the most significance included interferon signalling, p38 MAPK signalling and TGF-β pathways ( Figure 4A). We (p38 MAPK pathway inhibitor) reduced the growth of control cells by 25.5 ± 5.3%. FGF9-treated cultured THY-1 + germ cells grew faster than control, but when FGF9-treated cells were exposed to SB203580, the growth was reduced by 45.2 ± 3.2%, significantly more than control ( Figure 4B,C). FGF9 treatment with and without SB203580 did not substantially alter the colony number per 10 4 cultured THY-1 + germ cells as compared with unexposed controls ( Figure 4D). However, the total number increase of SSCs with FGF9 treatment was largely abolished by SB203580, indicating that disruption of p38 MAPK pathway eliminated the growth advantage conferred by FGF9 ( Figure 4E). We found that the phosphorylated p38 MAPK protein significantly increased with FGF9 treatment while the total amount of p38 MAPK did not change ( Figure 4F,G). These results show FGF9 promotes SSC proliferation via p38 MAPK pathway.

| Regulation of Etv5 by FGF9
While  Figure S7J). Figure 6A shows the effect of FGFs on the genes known to be involved in self-renewal or differentiation of spermatogonia (  Figure S8A). Of these, Etv5 showed the most dramatic response following FGF9 treatment ( Figure 6B and Figure S8A). It is notable that Etv5 did not show significant upregulation in RNAseq data ( Figure 6A), possibly because the cultures used for RNAseq included GDNF and GFRα1 that are known to regulate Etv5 expression. 9,33 Results also showed a significant decrease of Etv5 expression in cells exposed to SB203580 ( Figure 6C). To further confirm Etv5 is required for the effect of FGF9 on cell growth, small interfering RNA (siRNA) was applied. THY-1 + germ cell number in the Etv5 knockdown group was significantly lower than the control ( Figure 6D,E). SSC number per 10 4 cultured THY-1 + germ cells was reduced by a small, but significant amount ( Figure 6F).

| FGF9 activates Bcl6b through Etv5
We selected LIM homeobox 1 (Lhx1), brachyury (T) and B-cell CLL/ lymphoma 6 member B (Bcl6b) as candidates to investigate the downstream effect of Etv5 ( Figure S8B). 34,35 Of these, only Bcl6b showed a concomitant upregulation with Etv5 when cultured THY-1 + germ cells were exposed to FGF9 ( Figure 7A). There was a significant decrease of Bcl6b expression following Etv5 knockdown ( Figure 7B). When treated with FGF9, Bcl6b expression was indeed reduced when Etv5 was knocked down ( Figure 7C). In addition, we saw a significant decrease of Bcl6b expression in cultured THY-1 + germ cells when treated with SB203580 ( Figure 7D). These results demonstrated that p38 MAPK phosphorylation induced by FGF9 regulates Bcl6b expression via Etv5 in germ cells.

| D ISCUSS I ON
The FGF family plays an important role in diverse physiological processes, both in embryonic development and as growth factors throughout life. 36

F I G U R E 5
Single-cell RNA-sequencing analysis of FGF9 effect on cultured THY-1 + germ cells in vitro. RNA-sequencing analysis of 4,143 cells cultured for 48 h in mSFM with 20 ng/mL FGF9 integrated with 4482 control cells cultured with no growth factors for the same time. A, Cells clustered and scored for cell cycle expression (see Figure S7C). B, Clustering of cells after regressing out cell cycle genes for S and G 2 M phases. C, Pseudotime trajectory. D, Assignment of unbiased clusters to cell identities indicated by gene expression profile (see Figure  S7G for dying cells and STO). Note that 'SSCs' are used to designate the cluster containing SSCs, it is likely that not all cells are true SSCs. E, Module scores using sets of marker genes. For SSC: Gfra1, Ret, Etv5, Id4, Tspan8 and Esrp1. For progenitor spermatogonia: Upp1, Lhx1, Nanos3, Sox3 and Galnt12. For differentiating spermatogonia, Kit, Sohlh1, Crabp1 and Lmo1. F, Violin plots of gene expression by cell type and treatment. Three genes were selected as representative markers of gene expression for each of the three cell identities The most well-known growth factor for SSC self-renewal is GDNF. 41 In mutant mice containing a Ret-inactivating mutation that disrupts GDNF action, a subset of spermatogonia persist via FGF signalling from Sertoli cells. 18 This suggests FGFs can stimulate SSC self-renewal independent of GDNF signalling. PLZF is a marker for undifferentiated spermatogonia, 42 and in our experiments, WT testes displayed few PLZF + cells, consistent with undifferentiated spermatogonia comprising only 0.3% of germ cells. 43 Following Fgf9 stimulation, PLZF + cells were spread around the basement membrane and tubules lacked differentiating cell types, as determined by morphology and SYCP3 expression, a protein involved in the synaptonemal complexes of meiosis. 44 The observed increased number of PLZF + cells in tubules overexpressing FGF9 along with loss of chains of cells suggests an increase in progenitor spermatogonia and potentially an increase in SSCs. 45 In order to dissect the effect of FGF9 on SSCs, we turned to our in vitro SSC culture system. SSC number increased with FGF9 dose in a similar manner as FGF2. 10  p38 MAP kinase, see Figure 8B. 52 36 Considerable evidence from animal models and cell culture support FGF2 as a niche factor for SSCs. 56 Given that FGF9 is expressed in the testis and the similarity in germ cell response to FGF2, it is reasonable to suggest that FGF9 is also an SSC niche factor.
Despite the similarities in mechanistic response to FGF2 and FGF9 stimulation, certain differences between the two factors exist. Firstly, FGF2 is expressed via the Sertoli cells and differentiating germ cells, 56 while FGF9 is mainly expressed in Leydig cells, [14][15][16] see Figure 8A.  36 and we show in this study this leads to p38 MAPK phosphorylation, which activates Etv5 gene expression as well. After Etv5 activation, Bcl6 expression is upregulated, thereby regulating SSC proliferation. Simultaneously, FGF9 exposure inhibits differentiation, ultimately by affecting the expression of pro-differentiation genes and spermatogenesis. We show FGF9 is a potent niche factor that promotes SSC proliferation. Determining the intracellular mechanisms by which growth factors achieve their control over SSC fate determination is equally important. Our results indicate FGF9 induces phosphorylation of p38 MAPK and activate its signalling cascade within the SSCs. This results in an upregulation of Etv5 expression, which in turn increases Bcl6b expression, ultimately leading to an increased population of stem cells.

ACK N OWLED G EM ENTS
We thank James Hayden for micrographs and John Tobias for RNA-seq analyses. We also thank R. Naroznowski and D. Lee for animal maintenance and Nilam Sinha for helpful advice. This re-

CO N FLI C T O F I NTE R E S T
None declared.