Hierarchical differentiation competence in response to retinoic acid ensures stem cell maintenance during mouse spermatogenesis

Abstract

Stem cells ensure tissue homeostasis through the production of differentiating and self-renewing progeny. In some tissues, this is achieved by the function of a definitive stem cell niche. However, the mechanisms that operate in mouse spermatogenesis are unknown because undifferentiated spermatogonia (Aundiff) are motile and intermingle with differentiating cells in an 'open' niche environment of seminiferous tubules. Aundiff include glial cell line-derived neurotrophic factor receptor α1 (GFRα1)(+) and neurogenin 3 (NGN3)(+) subpopulations, both of which retain the ability to self-renew. However, whereas GFRα1(+) cells comprise the homeostatic stem cell pool, NGN3(+) cells show a higher probability to differentiate into KIT(+) spermatogonia by as yet unknown mechanisms. In the present study, by combining fate analysis of pulse-labeled cells and a model of vitamin A deficiency, we demonstrate that retinoic acid (RA), which may periodically increase in concentration in the tubules during the seminiferous epithelial cycle, induced only NGN3(+) cells to differentiate. Comparison of gene expression revealed that retinoic acid receptor γ (Rarg) was predominantly expressed in NGN3(+) cells, but not in GFRα1(+) cells, whereas the expression levels of many other RA response-related genes were similar in the two populations. Ectopic expression of RARγ was sufficient to induce GFRα1(+) cells to directly differentiate to KIT(+) cells without transiting the NGN3(+) state. Therefore, RARγ plays key roles in the differentiation competence of NGN3(+) cells. We propose a novel mechanism of stem cell fate selection in an open niche environment whereby undifferentiated cells show heterogeneous competence to differentiate in response to ubiquitously distributed differentiation-inducing signals.

Keywords: Retinoic acid receptor gamma; Spermatogenesis; Spermatogonia; Stem cell niche.

Fig. 1.

Testis anatomy and spermatogonial populations and their kinetics in the VAD model. (A) Anatomy of seminiferous tubules and seminiferous epithelium. A_undiff spermatogonia, including GFRα1⁺ (magenta) and NGN3⁺ (green) cells and KIT⁺ differentiating spermatogonia (blue), reside in the basal compartment (between the basement membrane and the tight junction of Sertoli cells). (B) Hierarchical and heterogeneous composition of spermatogonia. PL, preleptotene spermatocytes. (C) Representative whole-mount IF images of spermatogonia derived from an Ngn3-EGFP mouse triple stained for GFRα1, GFP and KIT. (D) The experimental schedule for E and F. Wild-type VAD mice were injected with VA and fed a normal (VA-sufficient) diet thereafter, before analysis at the indicated times. (E) Representative images of in situ hybridization analysis of Gfra1, Ngn3 and Kit expression in testis sections. Arrowheads indicate spermatogonia expressing these genes. Note the persisting Kit expression in interstitial cells (asterisks). (F) Counts of Gfra1⁺, Ngn3⁺ and Kit⁺ spermatogonia. Raw counts are summarized in supplementary material Table S1. Scale bars: 100 μm.

Fig. 2.

Fate of pulse-labeled NGN3⁺ spermatogonia following administration of VA. (A) Experimental design. Two days after the TM pulse, Ngn3-CreER^TM; CAG-CAT-EGFP transgenic mice maintained in VAD were injected with VA to resume spermatogenesis. Testis samples were harvested at the indicated times. (B) IF analysis of GFP and KIT expression in whole-mount seminiferous tubules 0, 2, 4 and 6 days after VA injection. Arrowheads indicate GFP⁺/KIT⁺ double-positive cells. Note that KIT immunostaining exhibits a punctate pattern at these early stages. (C) The number of GFP-labeled GFRα1⁺ A_undiff (magenta), GFRα1⁻ A_undiff (green), KIT⁺ (blue) and total (black) cells in the testes of Ngn3-CreER^TM; CAG-CAT-EGFP mice. Shown is the mean±s.e.m. of three, three, five and five testes on days 0, 2, 4 and 6, respectively. *P=0.041, **P<0.002 (t-test) compared with the values for day 0. Scale bar: 50 μm.

Fig. 3.

Fate of pulse-labeled GFRα1⁺ spermatogonia during VAD/VA administration. (A) The experimental schedule for B and C. Two days after the TM pulse, Gfra1-CreER^T2; CAG-CAT-EGFP transgenic mice maintained in VAD were injected with VA and then fed a normal diet. Testis samples were harvested at the indicated times. (B) Representative IF images of whole-mount seminiferous tubule 0, 2, 6 and 10 days after VA injection stained for GFP and GFRα1. (C) The number of GFP-labeled GFRα1⁺ A_undiff (magenta), GFRα1⁻ A_undiff (green), KIT⁺ (blue) spermatogonia and total labeled (black) cells. Shown is the mean±s.e.m. of five, five, seven, six, four and three testes for days 0, 2, 4, 6, 8 and 10, respectively. Data, except for those of GFRα1⁺ cells, on days 2, 6, 8 and 10 were significantly different compared with the values on day 0 (P<0.03, t-test). (D) Schedule for E and F. After the TM pulse, Gfra1-CreER^T2; CAG-CAT-EGFP transgenic mice were continually fed the VAD diet. (E) Representative IF images of whole-mount seminiferous tubule 2, 14 and 30 days after TM pulse stained for GFP and GFRα1. (F) Number of GFP-labeled GFRα1⁺ A_undiff (magenta) and GFRα1⁻ A_undiff (green) spermatogonia and total labeled cells (black), shown as the mean±s.e.m. of eight, three and five testes on days 0, 14 and 30, respectively. *P=0.032, **P<0.002 (t-test) compared with the values for day 2. Scale bars: 50 μm.

Fig. 4.

Expression of genes in the RA signaling pathway and the specificity of RARγ expression by NGN3⁺ spermatogonia. (A) Scatter plot comparing the levels of transcripts expressed by GFRα1⁺ and NGN3⁺ cells according to microarray analysis. For each fraction, average values from three independent RNA samples sorted from different animals are shown. The middle line indicates a difference of 0, and the outer lines represent ratios {log₂([NGN3⁺]/[GFRα1⁺])} of 1.0 and −1.0. Red dots indicate genes expressed at significantly different levels (P<0.05, t-test with Benjamini–Hochberg correction). Members of the Rar and Rxr gene families are indicated. A list of the genes in this panel is shown in supplementary material Table S2. (B) qRT-PCR analysis of Rara and Rarg mRNA expression in GFRα1⁺, NGN3⁺ and KIT⁺ spermatogonia. Relative copy numbers are shown in arbitrary units after normalization to the amount of Actb mRNA. The mean±s.e.m. values of three independent cell preparations from different animals are shown. *P=0.043, **P=0.011 and ***P=0.0065 (t-test). (C) Representative IF images of whole-mount seminiferous tubules from Ngn3-EGFP mice stained for EGFP, GFRα1 and RARγ. (D,E) Frequency of single- and double-positive spermatogonia for the expression of RARγ and NGN3 (recognized as RARγ⁺/KIT⁻ and NGN3-EGFP⁺/KIT⁻, respectively) (D) and of RARγ and GFRα1 (E). Total counts are shown above each bar. Scale bar: 50 μm.

Fig. 5.

Ectopic RARγ expression by GFRα1⁺ spermatogonia. (A) The CAG-CAT-3xFLAG-Rarg transgene. When CAT between the loxP sites is deleted by TM-activated Cre, FLAG-tagged RARγ is constitutively expressed under the control of the CAG promoter. (B) Experimental design of the fate analysis of GFRα1⁺ cells with enforced FLAG-RARγ expression upon VA readministration in VAD mice, as shown in C-F. Gfra1-CreER^T2; CAG-CAT-3xFLAG-Rarg transgenic mice were maintained in VAD and VA was administered 2 days after TM injection, as indicated. Testes were then processed for IF. (C,D) IF images of whole-mount seminiferous tubules of the mice described above, 2 days after VA injection, stained for FLAG-RARγ (green) and KIT (magenta) (C), and cell number relative to the number of initial induced cells (D). Data for GFP-labeled NGN3⁺ and GFRα1⁺ cells are reproduced from Fig. 2C and Fig. 3C, respectively, for comparison. The mean±s.e.m. value of three testes is shown. *P<0.003 (t-test), compared with the values of FLAG-RARγ⁺ GFRα1⁺ cells at day 2. (E,F) Representative confocal images of the same field of whole-mounts of seminiferous tubules of mice treated as described above, at 2 days after VA injection; staining was performed for GFRα1, KIT and FLAG (E). Open arrowheads, white arrowheads and small arrows indicate FLAG⁺ cells that are GFRα1⁺/KIT⁺, GFRα1⁺/KIT⁻ and GFRα1⁻/KIT⁺, respectively. (F) Quantitation of GFP⁺ and FLAG-RARγ⁺ cells showing different patterns of GFRα1 and KIT expression in Gfra1-CreER^T2; CAG-CAT-EGFP and Gfra1-CreER^T2; CAG-CAT-3xFLAG-Rarg mice, respectively. Cell numbers are shown above each bar. (G) Experimental design of the fate analysis of GFRα1⁺ cells with enforced FLAG-RARγ expression under normal conditions, as shown in H-J. Gfra1-CreER^T2; CAG-CAT-3xFLAG-Rarg transgenic mice were pulsed with TM at 13-17 weeks of age, and after 2 and 10 days their testes were processed for IF. (H) IF images of whole-mount seminiferous tubules 2 and 10 days after TM injection, stained for FLAG-RARγ and GFRα1. (I,J) Numbers of GFRα1⁺ A_undiff (magenta), GFRα1⁻ A_undiff (green), KIT⁺ (blue) spermatogonia and total cells (black) in either GFP-labeled (I) or FLAG-RARγ-expressing (J) cells of Gfra1-CreER^T2; CAG-CAT-EGFP and Gfra1-CreER^T2; CAG-CAT-3xFLAG-Rarg mice, respectively, following the schedule shown in G. The mean±s.e.m. of four (I) and three (J) testes are shown. *P<0.05 (t-test), compared with the values on day 2. Scale bars: 50 μm.

Fig. 6.

Kinetics of GFRα1⁺, NGN3⁺ and KIT⁺ spermatogonia during the seminiferous epithelial cycle. (A) The frequencies of Gfra1⁺, Ngn3⁺ and Kit⁺ spermatogonia observed during each stage of the seminiferous epithelial cycle as determined by in situ hybridization of three testis sections. Counts from 894 tubule transverse sections were summed and normalized to the number of Sertoli cells. Data for Kit⁺ spermatogonia are reproduced from Sugimoto et al. (2012). The same data are shown twice here to highlight the periodicity. (B) Model showing the behavior of GFRα1⁺ and NGN3⁺ spermatogonia during the seminiferous epithelial cycle that maintains the A_undiff pool while periodically producing differentiating KIT⁺ spermatogonia (see text for details).