Competition for Mitogens Regulates Spermatogenic Stem Cell Homeostasis in an Open Niche

Animals

The following mice were as previously described: Fgf5^– (Fgf5^go-moja/Utr), BAC-Fgf5, Fgf4^flox, Fgf8^–, Ngn3-Cre, Gfrα1-GFP, and Ngn3-GFP. The background of Fgf5^–, Gfrα1-GFP, Ngn3-GFP, BAC-Fgf5, and the control wild-type mice was C57BL/6 (Japan SLC, Japan CLEA), while Fgf8^– mice was maintained in a CD-1 background. Fgf4 and Fgf8 mutants were obtained from the Mutant Mouse Regional Resource Centers. Intercross between Fgf5 and Fgf4 or Fgf8 mutants were done following the mating scheme (Figures S5J and S5K). Busulfan (10 mg/kg) was intraperitoneally injected to adult mice (2.5–4 months old) as described previously. All animal experiments were conducted with the approval of The Institutional Animal Care and Use Committee of National Institutes of Natural Sciences, or institutional committees for animal and recombinant DNA experiments at the Research Institute, Osaka Women’s and Children’s Hospital.

Immunofluorescence (IF)

Whole-mount IF of seminiferous tubules and IF on testis cryosections (10μm-thick) were performed as previously described (Hara et al., 2014, Nakagawa et al., 2010, Tokue et al., 2017). The following antibodies were used: anti-GFRα1, anti-FGF5, anti-CD34, anti-αSMA, anti-GFP, anti-RARγ, anti-c-Kit, anti-SOX9, anti-CSF1R, anti-StAR, anti-pH3, anti-Cleaved PARP, anti-FGF8, anit-FGFR2, anti-SDC4, anti-SDCBP, anti-CD63, anti-LAMP1, anti-FGFR3, anti-ID4. All secondary antibodies were Alexa Fluor conjugated from Life Technologies and used at 1:300 dilutions. Anti-NAH46 and anti-HepSS-1 antibodies (Seikagaku) were directly labeled with Alexa Fluor 488 5-SDP ester (Molecular Probes, A30052) and Alexa Fluor 647 NHS ester (Molecular Probes, A20006), respectively, as follows. Conjugation reaction was performed by mixing antibody solution (1.0 mg/ml in PBS) and 1/50 volume of the reactive dye (10 mg/ml) and incubating for 1.5 h at room temperature under dark conditions. Unconjugated dye was removed by gel filtration using Bio-Spin 6 Tris Columns (#732-6227; Bio-Rad) to collect the labeled antibody as a flow-through fraction. The nuclei were stained with hoechst33342 (Life Technologies). Slides were mounted in Fluoro-KEEPER Antifade Reagent (Nacalai). Observations and measurements were performed using an Olympus BX51 upright fluorescence microscope equipped with a DP72 CCD camera, a Nikon A1r confocal system, or a Leica TCS SP8 confocal system.

Clonal fate analysis of GFRα1⁺ cells

Fgf5^+/+ or Fgf5^–/–; GFRα1-CreER^T2; CAG-CAT-GFP mice were injected intraperitoneally with 0.25 mg of 4-hydroxytamoxifen per individual (sigma). After the tamoxifen treatment, the testes were removed and analyzed by IF, as described previously (Hara et al., 2014).

Bead preparation and transplantation

Affi-Gel blue beads (Bio-Rad) were soaked in a solution of recombinant FGF5 proteins (0.1 mg/ml) or 0.1% BSA (bovine serum albumin; 0.1 mg/ml) for 1 h at room temperature according to (Uchida et al., 2016). To mark the tubular wall adjacent to the transplanted beads, some beads were immersed in DiI (0.83 mg/ml; Thermo Fisher Scientific) solution for 15 min. For transplantation, the soaked beads were transplanted into the testicular interstitium (1 or 2 beads [one per site] were separated with appropriate intervals) via vitrified micro-capillary under a dissecting microscope.

Laser capture microdissection

Freshly isolated testes from 8 weeks-old C57BL/6 mice were cryosectioned with 7μm-thick sections, placed on slides, fixed, and stained with HistoGene before collection of the areas of interest using PixCell IIe (ArcturusXT), according to the manufactures’ protocol. The obtained tissue fragments were proceeded for cDNA microarray gene expression analysis.

cDNA microarray gene expression analysis

From tissue fragments collected by laser microdissection, RNA was purified using RNeasy micro kit (QIAGEN) and processed for two-round amplification to prepare fluorescence-labeled probes as described. Briefly, in the first round, RNA samples were reverse-transcribed with T7-(dT)₂₄ primer and made double-stranded, followed by cRNA synthesis using MEGAscript T7 kit (Ambion). After quality checked with an Agilent 2100 Bioanalyzer, cRNA was reverse-transcribed into ds-cDNA, and subjected to Cy3-labeled cRNA synthesis (Agilent Technology) by T7 reaction. From other materials (i.e., tissues, sorted cells, or cultured cells), RNA purification and preparation of Cy3-labeled cRNA probes were performed as described. Hybridization, scanning and data analysis were done as described. Briefly, Cy3-labeled probes were fragmented and hybridized to an Agilent whole mouse genome 4x44K or 8X60K array (Agilent Technology). Then, the image data was obtained using a G2505C scanner (Agilent Technology), analyzed using a Gene Spring software (Silicon Genetics). Data correction was performed with the threshold raw signals set to 1.0, percent shift to the 75th percentile as normalization algorithm, and no baseline transformation.

in situ hybridization (ISH) of histological sections

315 genes that showed high enrichments to vasculature-associated regions over tubule-bounding regions were selected based on the microarray data above. dsDNA fragments containing full length sequences of these transcripts were amplified from FANTOM clones (DNAFORM) using the primers of M13_Forward: CGACGTTGTAAAACGACGGCCAGTG and M13_Reverse: AGCGGATAACAATTTCACACAGGAAAC. Then, digoxigenin-labeled antisense RNA probes were synthesized using DIG RNA Labeling Kit (Roche) and processed for in situ hybridization on paraffin embedded sections, according to a protocol described previously (Yoshida et al., 2001).

ISH of dispersed testicular cells

ISH of dispersed single cells was carried out using an RNAscope Fluorescent Multiplex Kit (Advanced Cell Diagnostics). Freshly isolated testis from 8 weeks-old C57BL/6 mice were processed to generate cell suspensions of the interstitial cells and the peritubular cells on the seminiferous tubules by pipetting. The cell suspension was applied to MAS-GP-coated slide glass (Matsunami) and employed for detection of Fgf5, Cd34 and Des RNA using each specific target probe and HybEZ Hybridization system (Advanced Cell Diagnostics).

RT-qPCR

Total RNA was extracted form samples (tissues, sorted cell or cultured cells) using RNeasy kits (QIAGEN), reverse-transcribed using Superscript III first strand synthesis kit (Life Technologies), and processed for RT-qPCR using a LightCycler 480 system (Roche) with gene-specific primers (Table S3).

FACS

For microarray analysis, GFRα1⁺, NGN3⁺ and KIT⁺ cell fractions were sorted by FACS, as described previously (Ikami et al., 2015, Tokue et al., 2017). The GFRα1⁺ fraction was collected from Gfrα1-GFP mice as the GFP⁺ fraction, and the NGN3⁺ and KIT⁺ fractions were collected from Ngn3-GFP mice as GFP⁺/KIT^– and GFP⁺/KIT⁺ fractions, respectively, using an EPICS ALTRA instrument (Beckman Coulter). The data of FACS and the microarray were partly published previously (Ikami et al., 2015, Tokue et al., 2017), with the data-set deposited in the GEO (GSE75532). In Figure 3I, GFRα1⁺ cells were collected from 2.5 months-old-adult Gfrα1-GFP mice in wild-type and Fgf5^–/– backgrounds in the same manner to (Ikami et al., 2015, Tokue et al., 2017). In Figure S7, spermatogonial fractions were sorted from Gfrα1-GFP mouse testicular cell suspension, based on CD9, ECAD, KIT and GFP staining, using a FACSAria II cell sorter (BD Biosciences). Anti-ECAD antibodies were conjugated with PE by R-Phycoerythrin AffiniPure Fab Fragment Goat Anti-Rat IgG (Jackson ImmunoResearch).

In vitro culture of spermatogonia

Germline Stem (GS) cells derived from the C57BL/6 x ICR intercrossed mice were maintained according to (Tokue et al., 2017). For the quantification of mitogenic effect of FGF2 or FGF5, 2 × 10⁴ GS cells per well of 12-well plate were cultured in the respective concentration of FGF2 or FGF5 in supplement with 10 ng/ml GDNF. After 8 days, the cell numbers were counted (n = 3 independent experiments). For the gene expression analysis, GS cells were depleted for FGF2 and GDNF for 3 days, and then supplemented with or without FGF5 (100 ng/ml) for 24 hours, followed by cDNA microarray analyses. For the co-culture with CD34⁺ cells or MEF, GS cells were cultured in supplement with GDNF, and with or without FGF2. The passage was performed every 6 days.

Culture of CD34⁺ cells

Primary testicular cells expressing CD34 were prepared from 8 wk-old mice referring to (Seandel et al., 2007), which designated these cells as mouse testicular stromal, or MTS, cells. Seminiferous tubules were collected from detunicated testes and minced. The tissue was washed and then enzymatically dissociated with agitation at 37 °C in a buffer containing collagenase, hyaluronidase, and DNase I. The resultant cell suspension (non-filtered) was collected, plated in dishes coated with gelatin in a 50:50 mixture of alpha MEM/StemPro-34 (Thermo Fisher Scientific) supplemented with 20% FBS and expanded over two to five passages. Cells were then cryopreserved or plated in 12-well plates coated with gelatin, and treated with mitomycin-C for 3 h, before use for co-culture with GS cells.

Luciferase assay

Transient transfection of GS cells cultured in 48-well plates coated with laminin (BD Biosciences) was performed using Lipofectamine-3000 (Thermo Fisher Scientific), and the culture medium was changed after 24h. Analyses were performed 24h post-medium change with or without FGF2. The luciferase activity of cell lysates was measured using a Dual-Luciferase assay system and a GloMax 20/20n luminometer (Promega). The activity of the firefly luciferase reporter pGL4-Ngn3 was normalized to that of Renilla luciferase expressed from co-transfected pGL4.7 plasmid as described (Tokue et al., 2017).

Copy number determination of the BAC transgene

Genomic DNA was extracted from tail clips of WT and mutant mice according to a standard protocol, including phenol-chloroform extraction after lysis in buffers containing Proteinase K. To quantify the BAC transgene including the Fgf5 gene (Figure S2A), the real time TaqMan PCR method using universal Probe Library probes (Roche) and gene-specific primers was used (Table S3). Copy number of Fgf5 was determined using Ppia as a standard.

Histology, evaluation of degenerating tubules, and measurement of the tubule area

Testes of WT and mutant mice were fixed with Bouin’s fixative and processed for paraffin-embedded section preparation (7μm 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). Chronological sequence of these stages represents the periodic change of seminiferous epithelium, known as “seminiferous epithelium cycle.” In the testes of Fgf mutants, 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. The area of the cross sections of seminiferous tubules were measured (n = 255 and 195 tubule sections in WT and Fgf5^–/– mice, respectively). The round shape tubule cross sections were photographed under bright-field illumination, then measured the areas by a CellSens Standard software of an Olympus BX51 microscope.

Counting the density of GFRα1⁺ or RARγ⁺ cells

The densities of GFRα1⁺ or RARγ⁺ cells were measured on immunofluorescenced whole mount seminiferous tubules or cryo-sections of the testes. Figures 1C, 3A, right, 6A, 6B, 6J, S2E, S6M, and S6N were counted by the whole mount seminiferous tubules. For counting the cell density in 1 mm tubule length, the 1mm tubule length was measured, then the GFRa1⁺ cell number was counted by an Olympus BX51 fluorescence microscope equipped with a DP72 CCD camera or a Leica TCS SP8 confocal system. For counting the GFRα1⁺ cell density in > 10 mm tubule length, the GFRα1⁺ cell number in long continual segments over 10 mm was counted. The total length of counted tubule segments were 105 mm (Figure 1C), 140 mm of BAC-Fgf5^Tg/+, 78 mm of Fgf5^+/+, 94 mm of Fgf5^–/– mice (Figure 3A right), 122 mm (day10), 105 mm (day15), 94 mm (day20), 90 mm (day25), 160 mm (day30), 103 mm (day40), 100 mm (day50), 90 mm (day60), 100 mm (day80), 73 mm (day110) of WT mice (Figure 6A), and 61 mm (day0), 106 mm (day10), 116 mm (day15), 86 mm (day20), 91 mm (day25), 102 mm (day30), 104 mm (day40), 109 mm (day50), 104 mm (day60) of Fgf5^–/– mice (Figure 6B) after busulfan treatment. In Figure 6J, the total lengths of counted tubule segments were 46 mm (+DiI in FGF5 beads), 79 mm (–DiI in FGF5 beads), 17 mm (+DiI in BSA beads), 17 mm (–DiI in BSA beads). In Figure S2E, the densities of GFRα1⁺ and SOX9⁺ cells were counted in total 14- and 18-mm tubule segments of WT and Fgf5^–/– mice. In Figures S6M and S6N, the GFRa1⁺ or RARγ⁺ cell densities were counted in the following segments, 53 mm (day0), 32 mm (day3), 28 mm (day6), 29 mm (day9), 30 mm (day14), 39 mm (day20), 21 mm (day30). The densities of GFRα1⁺ or RARγ⁺ cells were counted in the cryo-sections at Figures 3A left, 3B–3H, 6C, S2D, S3D, S3H, S4I, S4J, S4M, S6K, and S6L were counted by the cryo-sections. For counting the cell density per tubule section, the GFRα1⁺ or RARγ⁺ cells per more than 150 tubule sections per testes (N ≥ 3 mice) were counted by Olympus BX51 microscopy.

To assess whether spermatogonia are distributed uniformly, randomly or in a clustered fashion, we performed a statistical test, comparing the homeostatic frequencies of spermatogonial unit numbers in bins of 1mm length along the tubule axis to a Poisson distribution with the same mean (Figure S1A). A standard ${\chi }^2$-test yielded a p value smaller than ${10}^{-5}$, indicating a significant deviation from spatial randomness. Furthermore, a variance-to-mean ratio of the bin population of $(\langle n^2\rangle -{\langle n\rangle }^2)/\langle n\rangle \approx 3$ indicated that spermatogonial units were more clustered rather than random or uniform. However, we could not detect any spatially regular patterns associated with this clustering.

Measurement of Fgf5 RNA-positive cells on tubule circumference

Testicular sections were stained for Fgf5 by in situ hybridization. The tubule cross sections were photographed under bright-field illumination, then measured the Fgf5–positive signals on tubule circumference.

Measurement of FGF5-positive signals adjacent to interstitial area or tubule bounding area

Testicular sections were double stained for FGF5 and CD34 by IF. The tubule cross sections were photographed under a confocal microscope (Nikon A1r). The data were then analyzed to determine whether the FGF5-positive or -negative signals distributed in the interstitial-tubule or tubule-tubule bounding regions, and then measured their lengths of FGF5- and CD34-double positive or FGF5-negative and CD34-positive, respectively.

Scoring the distribution of GFRα1⁺, RARγ⁺, KIT⁺ and ID4⁺ spermatogonia for Fgf5–positive area

Testicular sections were stained for Gfrα1 or Fgf5 by ISH, or RARγ, KIT, or ID4 by IHC. All the tubule cross sections were photographed. The spatial correlation with Fgf5–positive signals was judged from whether the distributions of Gfrα1, RARγ or KIT-positive spermatogonia were adjacent to Fgf5⁺ signals on the adjacent section. When their spermatogonia were (or were not) adjacent to Fgf5⁺ signals, they were categorized as “positive” (or “negative”). When their spermatogonia were partially adjacent to Fgf5⁺ signals, they were categorized as “boundary.” The category of “boundary” was estimated as an intermediate between “positive” and “negative,” and then the each “boundary” parameter was assigned to 0.5 of “positive” and 0.5 of “negative” in the determination whether the spermatogonia distributions were adjacent to Fgf5 mRNA in the category between positive or negative for the calculation. Then, expected positive numbers were calculated assuming their non-biased distributions, and the ‘preferences’ (actual positive cell number/expected positive cell number) were determined; these were statistically evaluated by chi-square test between the actual and expected cell numbers adjacent or not to Fgf5⁺ area. For each data point, more than 40 seminiferous tubule cross-sections in 3 testicular slices were examined. The data were then analyzed to determine whether the distributions of Gfrα1, RARγ, KIT or ID4-positive spermatogonia were adjacent to Fgf5⁺ signals on the adjacent section. Statistical evaluations were performed as explained below, using the data of the Gfrα1-positive spermatogonia as an example. The 3 testis specimens used for analyses contained 91 cross-sections of the seminiferous tubules in total, and 121 Gfrα1-positive spermatogonia were observed. These 121 spermatogonia were classified as “positive” or “negative” for adjacent to Fgf5⁺ signals on the adjacent section. The lengths of the circumference of the tubule cross-sections were also summarized according to their levels of ISH staining of Fgf5. Then, the expected numbers of Gfrα1-positive spermatogonia in the category of positive or negative area were calculated on the basis of the null hypothesis: the Gfrα1-positive spermatogonia evenly distribute without bias. The expected numbers of Gfrα1-positive spermatogonia were obtained by multiplying the total number of Gfrα1-positive spermatogonia by the percentage of tubule sections in the category. The preferences of positive cells in each category were calculated as the ratio of the observed number of positive cells in each category to the expected number in the same category. Thus, a preference of 1 indicates that the actual number of Gfrα1-positive cells is the same as expected, i.e., a non-biased distribution. Values greater or smaller than 1 suggest preference or avoidance, respectively. The differences between the observed and expected numbers of Gfrα1-positive spermatogonia were statistically evaluated by chi-square tests. The P-value (0.00002) was small enough to reject the null hypothesis and indicated a non-random distribution of Gfrα1-positive spermatogonia. This was also true for RARγ but not KIT-positive spermatogonia, whose P-values were 0.00052 and 0.13494, respectively.

Scoring the distribution of GFRα1⁺ and ID4⁺ spermatogonia for the interstitium or tubule-tubule bounding area

Testicular sections were stained for GFRα1 and ID4 by IF. The tubule cross sections were photographed by Nikon A1r confocal system. The spatial correlation was judged from whether the distributions of GFRα1 or ID4-positive spermatogonia were adjacent to the interstitium or tubule bounding region. When their spermatogonia were partially adjacent to the intermediate area between the interstitium and the tubule bounding region, they were categorized as “boundary.”