Testosterone Levels Influence Mouse Fetal Leydig Cell Progenitors Through Notch Signaling

ABSTRACT

Leydig cells are the steroidogenic lineage of the mammalian testis that produces testosterone, a key hormone required throughout male fetal and adult life for virilization and spermatogenesis. Both fetal and adult Leydig cells arise from a progenitor population in the testis interstitium but are thought to be lineage-independent of one another. Genetic evidence indicates that Notch signaling is required during fetal life to maintain a balance between differentiated Leydig cells and their progenitors, but the elusive progenitor cell type and ligands involved have not been identified. In this study, we show that the Notch pathway signals through the ligand JAG1 in perivascular interstitial cells during fetal life. In the early postnatal testis, we show that circulating levels of testosterone directly affect Notch signaling, implicating a feedback role for systemic circulating factors in the regulation of progenitor cells. Between Postnatal Days 3 and 21, as fetal Leydig cells disappear from the testis and are replaced by adult Leydig cells, the perivascular population of interstitial cells active for Notch signaling declines, consistent with distinct regulation of adult Leydig progenitors.

Jagged 1-dependent Notch signaling in interstitial cells acts in conjunction with testosterone to regulate fetal Leydig cell differentiation and the transition from fetal to adult Leydig cells.

INTRODUCTION

The development of Leydig cells, the steroidogenic lineage in the testis, is critical for male-specific development throughout life, driving secondary sex differentiation and spermatogenesis. During fetal development, the population of Leydig cells is vital for initial virilization of the embryo, elaboration of the epididymis and vas deferens, and patterning of male external genitalia. Disruptions in endocrine function or androgen secretion are associated with malformations of male external genitalia, such as hypospadias, which is among the most frequent birth abnormalities [1]. Testosterone secretions from adult Leydig cells promote development of male secondary sex characteristics (such as body hair, voice changes, and musculature) at puberty. In addition, levels of testosterone control spermatogenesis and libido throughout the male reproductive lifespan.

Fetal Leydig cells (FLCs) and adult Leydig cells (ALCs) have distinct features that include unique morphology and gene expression profiles [2–4]. A transition from FLCs to ALCs occurs between the first and third weeks of postnatal life. Prior to the time when adult Leydig cells begin to appear, there is a nadir of testosterone production that results from the abrupt decline in the number of FLCs, which begins soon after birth [4–6]. However, the degree of postnatal FLC loss is still controversial: a morphometric analysis reported that a portion of FLCs persist in the rat testis even into adult stages [7], and another study recently demonstrated that a subset of steroidogenic cells in the adult mouse testis expresses an FLC-specific enhancer [8].

Differentiated (steroidogenic) Leydig cells are not in active cell cycle and rarely proliferate in fetal or adult life [9]. Instead, their numbers are replenished from a progenitor pool of undifferentiated cells within the interstitial compartment of the testis [2]. Although the Leydig progenitor population has not been specifically or directly identified, our recent work showed that both the coelomic epithelium of the gonad and the neighboring mesonephric tissue could give rise to cells that differentiate as steroidogenic Leydig cells [10]. It is currently accepted in the field that FLCs are distinct from and do not give rise to ALCs [2–4]. However, it is unclear (1) whether the two types of Leydig cells share a common progenitor cell present in the fetal testis, and (2) whether their progenitors are regulated by the same or different signaling pathways at fetal and adult stages.

During fetal life, the Notch pathway determines the balance between Leydig progenitors and differentiated Leydig cells [11]. Notch signaling, acting through the target gene Hes1, maintains interstitial cells in an undifferentiated state, thus, blocking the Notch pathway with DAPT (N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester) or through Hes1 mutation caused the differentiation of supernumerary mature FLCs at the expense of their progenitors. On the other hand, overexpression of the NOTCH1 intracellular domain in the gonad was sufficient to block Leydig cell differentiation [11].

Multiple Notch receptors and ligands are expressed in the vasculature or in vasculature-associated cells [11, 12]. While some Leydig cells form even in the absence of vasculature [13], the vasculature is critical for proliferation of interstitial cells and expression of multiple interstitial markers [10, 13]. Thus, we speculated that the vasculature might be part of the regulatory microenvironment that supports the maintenance of Leydig progenitors and the differentiation of their progeny.

To characterize the mechanisms driving progenitor maintenance and Leydig cell differentiation, we used a transgenic Notch reporter green fluorescent protein (TNR-GFP) mouse line to detect active Notch signaling within the interstitial compartment of the testis [14]. We identified a vasculature-associated TNR-GFP-positive cell population in the fetal and early postnatal testis. This population of cells represents a putative progenitor type population that gradually disappears from the testis during postnatal stages when the FLC population declines. Active Notch signaling in the interstitium disappears after puberty, suggesting that Notch is involved in the maintenance of only the FLC and not the ALC precursor population. We show that the Notch ligand JAG1 is specifically expressed in vasculature-associated cells in the interstitium of the fetal and postnatal testis. Jag1 conditional deletion in the interstitial compartment led to the appearance of supernumerary Leydig cells, indistinguishable from Notch pathway loss-of-function mutation in Hes1 mutants [11]. Artificial elevation of testosterone levels resulted in high levels of active Notch signaling within vasculature-associated presumptive Leydig progenitor cells of the postnatal testis. Our findings suggest a physiological feedback mechanism between circulating levels of testosterone and Notch signaling that can repress differentiation of mature Leydig cells through maintenance of a Notch-activated progenitor state.

MATERIALS AND METHODS

Mouse Lines

Wild-type expression was analyzed in CD-1 (Charles River) Mafb-GFP^+/− mice (a gift from S. Takahashi through L. Goodrich, maintained on a C57BL/6J [B6] background) [10, 15], and in the TNR-GFP mouse line (maintained as a homozygote in a CD-1 genetic background; see below).

To conditionally delete Jag1 from interstitial cells, we used a Jag1 floxed allele (Jag1^lox/lox) [16], a gift from F. Radtke, in combination with a smooth muscle-specific Tagln-Cre mouse line (also called SM22-Cre) [17]. This model of conditional Jag1 deletion from mesenchymal and smooth muscle cells has been previously described [18]. The expression of Tagln-Cre was verified using a Cre-responsive R26R-lacZ strain [19], obtained from Jackson Laboratories. Jag1^lox/lox, Tagln-Cre, and R26R lines are all maintained on a B6 background.

Transgenic TNR-GFP mice, a gift from B. Hogan but available from Jackson Laboratories (official strain name is Tg [Cp-EGFP] 25Gaia/J), was used to detect active Notch signaling in situ. This mouse line contains a CBF-1 response element with four CBF-1-binding sites and a minimal SV40 promoter followed by a GFP coding sequence [14]. All mice were housed according to National Institutes of Health guidelines, and experimental protocols were evaluated and approved by the Duke University Medical Center Institutional Animal Care and Use Committee or the Animal Research Committee at the University of California, Los Angeles.

Quantitative RT-PCR

Testes were dissected in phosphate-buffered saline (PBS) and homogenized by vortexing and sonication in TRIzol reagent (Invitrogen). Total RNA was extracted using a TRIzol/isopropanol precipitation method. cDNA synthesis was performed with the iScript cDNA kit (Bio-Rad), using 300–500 ng of total RNA.

Quantitative RT-PCR (qPCR) analyses of interstitial tissue separated from seminiferous tubules were carried out. To separate the interstitium and tubules, the tunica was removed, after which the testis was gently agitated in PBS to free tubules from intervening interstitial tissue. The tubules were teased apart further in a Petri dish by using fine forceps, thus releasing interstitial cells, and the PBS (containing interstitial cells) was centrifuged (at 3700 rpm for 5 min at 4°C) to harvest interstitial cells. RNA was then extracted from the interstitial cells, using the TRIzol/isopropanol extraction method.

qPCR was performed using StepOnePlus real-time PCR system (Applied Biosystems) with master mixture from Sensimix SYBR One-Step kit (Bioline), using the following parameters for 45 cycles: 95°C for 15 sec, 59°C for 30 sec, and 72°C for 30 sec. Threshold cycle (Ct) values were calculated by using StepOne software (version 2.2.2; Applied Biosystems). ΔCt values were calculated by using the housekeeping gene Gapdh as an internal control. All expression levels were normalized relative to those of Gapdh. Results are represented as means ± SEM. Statistical analyses were performed using a Student t-test with a P value of <0.05 considered statistically significant.

All reactions were run with primer sets specific for the following genes: Arx, Cyp17a1, Dll1, Dll4, Hes1, Jag1, Jag2, Lhx9, Nes, Notch1, Notch2, Notch3, Notch4, Pdgfra, Ptch1, Cyp11a1 (also known as Scc), Vcam1, and Cdh5 (also known as VE-Cad). Primer sequences are listed in Supplemental Table S1 (all supplemental data are available online at www.biolreprod.org).

Immunofluorescence and Confocal Microscopy

For stages prior to Embryonic Day 14.5 (E14.5), whole-mount immunofluorescence detection of testes was performed. Samples were dissected in PBS and fixed in 4% paraformaldehyde (PFA) overnight at 4°C. After several washes in PBTx (PBS plus 0.1% Triton X-100), samples were incubated in blocking solution (PBTx plus 10% FBS and 3% bovine serum albumin [BSA]) for 1–2 h at room temperature or overnight at 4°C. Primary antibodies were diluted in blocking solution and applied to samples overnight at 4°C. After several washes in washing solution (PBTx plus 1% FBS and 3% BSA), fluorescent secondary antibodies were applied for 4–5 h at room temperature or overnight at 4°C. Cy5- or Cy3-conjugated secondary antibodies (1:500 dilution; Jackson ImmunoResearch) or Alexa 647- or Alexa 488-conjugated secondary antibodies (1:500 dilution; Molecular Probes) were used for immunofluorescence. 4′,6-Diamidino-2-phenylindole (Sigma-Aldrich) was used to stain nuclei. Samples were mounted in 2.5% DABCO (1,4-diazabicyclo[2.2.2]octane; Sigma-Aldrich) and imaged using a Leica SP2 confocal microscope.

For stages between E14.5 and postnatal day 12 (P12), testes were fixed in 4% PFA overnight at 4°C and washed in PBTx. Samples were processed through a sucrose gradient (10%, 15%, 20%, 20% OCT [Sakura] 1:1 overnight at 4°C) before embedding in OCT medium at −80°C for sectioning at 20 μm. Cryosections were then stained as previously described [20]. For stages older than P12, the tunica of the testis was gently punctured 10–12 times with a 27-guage needle before being fixed in 4% PFA overnight at 4°C. Samples were then cut in half with a scalpel and fixed further for 1 h in PFA at 4°C to ensure fixative was able to access the center of the tissue. After several washes in PBTx, samples were processed as above before being embedded in OCT medium for sectioning at 20–25 μm.

Dilutions, codes, and sources of primary antibodies used were as follows: rat anti-PECAM1 (553370, 1:250; BD Pharmingen), chicken anti-GFP (GFP-1020, 1:1,000; Aves), rabbit anti-GFP (A11122, 1:1,000; Molecular Probes), rabbit anti-HSD3B1 (also known as 3β-HSD; 1:2000; a gift from K. Morohashi), rabbit anti-HSD3B1 (KO607, 1:100; TransGenic Inc.), rabbit anti-SOX9 (AB5535, 1:2000; Millipore), goat anti-AMH (also called MIS; sc-6886, 1:500 dilution; Santa Cruz Biotechnology), goat anti-VCAM1 (AF643, 1:2000; R&D), goat anti-JAG1 (AF599, 1:2000; R&D), rabbit anti-MKI67 (also known as Ki67; RM-9106-S, 1:500; ThermoScientific), rabbit anti-β-GAL (08559761, 1:10 000 dilution; MP Biomedicals), rat anti-NOTCH2 (C651.6DbHN, 1:200; Developmental Studies Hybridoma Bank), goat anti-CYP17A1 (sc-46081, 1:500; Santa Cruz Biotechnology), and rabbit anti-AR (sc-816, 1:300; Santa Cruz Biotechnology). Antibody-negative controls were processed in a manner identical to that of experimental samples, except that primary antibodies were omitted.

Testosterone Injections

TNR-GFP mice (n = 3) were injected subcutaneously with testosterone propionate (Sigma-Aldrich) dissolved in corn oil (5 mg/kg) once daily, from age P7 until P11 (total, 5 injections). The control group (n = 3) was injected with vehicle (corn oil). The mice were euthanized and dissected 24 h after final injection at P12. For each animal, one testis was fixed for immunofluorescent imaging and the other was used for separation of interstitial tissue (as described above), RNA extraction, and subsequent qPCR analysis.

Cell Counting

Using multiple tissue slices from different sections of the testis, we took 7–8 confocal images of each biological replicate. The number of bright Notch-GFP+ interstitial cells were counted and normalized to a unit area of 1 sq mm (image size was 375 μm × 375 μm). Statistical analysis was performed using a Student t-test.

RESULTS

Notch Signaling Is Active in Multiple Cell Types in the Fetal Testis

We have previously shown that Notch signaling drives the balance between progenitors and differentiated Leydig cells in the fetal testis by maintaining the progenitor population in an undifferentiated state [11]. However, the locations of the progenitor niche and the cell type in which active Notch signaling was taking place in the fetal testis were unknown. We examined expression of the transgenic Notch reporter, TNR-GFP, which expresses GFP in cells that are actively receiving a Notch signal [14].

At the stage when FLCs begin to differentiate in the testis (E12.5), TNR-GFP expression is visible in multiple cell types (Fig. 1A). Active Notch signaling was observed in some vasculature (as labeled by PECAM1), in Sertoli cells (which surround the germ cell population as we recently described in [21]), and interstitial cells, but TNR-GFP was very rarely detected in germ cells (Fig. 1, A and A'). GFP-positive cells in the interstitium were often localized immediately adjacent to blood vessels (Fig. 1A'). By E13.5, we detected patches of GFP expression throughout the interstitium and within a number of HSD3B1-positive cells (Fig. 1, B and B'). We anticipated that the Notch reporter would be limited to progenitor cells, based on previous results showing that Notch signaling was required to maintain cells in a progenitor state [11]. We examined TNR-GFP testes at E16.5 but still saw strong expression of GFP in a subpopulation of steroidogenic cells (Fig. 1, C and C'), although in fewer steroidogenic cells than at earlier stages. These data suggest that GFP expression in HSD3B1-positive cells was likely perdurance from cells that recently differentiated and were transitioning to a steroidogenic fate. Even though the pattern of interstitial GFP expression was mosaic, strong expression was consistently biased towards cells contacting or proximate to the vasculature (Fig. 1, A' and C').

FIG. 1.

TNR-GFP is expressed in multiple cells within the nascent fetal testis. Immunofluorescent images of fetal testes. Colors of immunofluorescent markers are indicated in all figures. A', B', and C' show higher magnification images of the boxes in A, B, and C, respectively. PECAM1 labels both endothelial and germ cells. In the E12.5 gonad (A), TNR-GFP expression is observed within blood vessels, Sertoli cells, and interstitial cells. A' shows highlighting of interstitial GFP-positive cells (arrows) tightly associated with vasculature (v), and lack of GFP expression in germ cells (gc). At E13.5 (B and B'), GFP is expressed in a subset of HSD3B1-positive Leydig cells. At E16.5 (C and C') TNR-GFP is still observed in interstitium and Sertoli cells, in addition to a small subset of Leydig cells. C') Note the proximity of GFP-bright cells (arrows) to vasculature (blue). Dotted lines define testis cord boundaries. Bars = 25 μm in all panels.

Notch Family Members Are Expressed in Specific Cells Within the Testis

To ascertain which Notch ligands may be involved in mediating Notch signaling, we examined the expression of Notch family members in our recent microarray analysis of sorted testis cell types [12]. Notch family members normally associated with vasculature, such as Dll4, Jag2, Notch1, and Notch4 [11, 22–26], were specifically enriched in endothelial cells of the gonad at E12.5 and E13.5 (Supplemental Fig. S1). Notch2 was enriched in Sertoli cells and interstitial cells at E12.5 and E13.5, while Notch3 was also enriched in interstitial cells but was expressed at much lower levels than Notch2. Consistent with previous in situ hybridization results [11], fluorescent immunostaining of NOTCH2 confirmed localization to the interstitial and supporting cell lineages in the testis at E13.5 (Fig. 2, A and B). The Notch ligands Dll1 and Dll3 were not detectable in the fetal gonad by our microarray criteria and were not explored further. The most interesting expression pattern revealed by our lineage-specific microarray data was for the Notch ligand Jag1, which was enriched within endothelial cells and interstitial cells.

FIG. 2.

Notch family members NOTCH2 and JAG1 are expressed in the fetal testis. Immunofluorescent images of fetal testes. A' and B') NOTCH2-channel-only versions shown in A and B. B and B') Higher magnification images of the yellow boxes in A and A', respectively. A and B) In the E13.5 fetal testis, NOTCH2 is detected in the cell membrane of Sertoli cells (SOX9-positive) and interstitial cells, whereas JAG1 is expressed only in the SOX9-negative interstitial compartment. B) Dashed line denotes testis cord-interstitium boundary; asterisks denote NOTCH2-dim germ cells in a cluster. (C) Mafb-GFP was used to label Leydig cells, as GFP expression is enriched in Leydig cells at E14.5 [10]. C' and C”) Higher magnification images of the boxed regions shown in C. C) At E14.5, JAG1 is expressed in vasculature-associated interstitial cells and occasionally weakly in Mafb-GFP-bright Leydig cells (C'). In the region near the coelomic vessel (cv) (C”), JAG1-bright cells (arrows) are adjacent to vasculature and are Mafb-GFP-dim, suggesting that they are not differentiated Leydig cells. tc, testis cord. Bars = 50 μm in all panels.

Immunostaining at E13.5 confirmed this pattern, and showed that JAG1 levels were highest in vasculature-associated non-Leydig interstitial cells (Fig. 2, C and C”). Faint expression was occasionally detected within endothelial cells themselves (Fig. 2C”). Very low levels of JAG1 were detected in Leydig cells, as marked by strong expression of Mafb-GFP [10] (Fig. 2C'). Given the important role of vasculature in testis morphogenesis [13, 27, 28] and our recent work showing that perivascular cells are progenitors for fetal Leydig cells [10], we further investigated the role of Jag1 in Leydig cell and progenitor cell development.

Jag1 Is Required for Maintenance of Leydig Progenitor Cells

The expression pattern of JAG1 suggested a role in maintenance of Leydig cell progenitors within the interstitial compartment of the testis. To determine a specific function for Jag1, we conditionally deleted Jag1 [16] by using Tagln-Cre, a Cre specifically active in mesenchymal and perivascular cells [17, 18]. We first characterized the expression of Tagln-Cre in the fetal testis by using the Rosa26R-lacZ (R26R) reporter line to identify Cre-expressing cells [19]. At E14.5 (a stage when fetal Leydig cells have been specified), we observed β-GAL (β-galactosidase) expression in virtually the entire interstitial compartment (Fig. 3A). Expression was especially strong in vasculature-associated cells and peritubular myoid cells surrounding testis cords (Fig. 3A”). We also saw some expression in the coelomic epithelium and in very few Sertoli cells, but never observed β-GAL in germ cells. Within the endothelium, β-GAL was rarely observed, although perivascular cells had strong β-GAL expression (Fig. 3A”).

FIG. 3.

Interstitial Jag1 is required for the balance between differentiated Leydig cells and their progenitors. Immunofluorescent images of fetal testes. A' and B') Single channel images are shown. (A” and B”) Higher magnification images of the boxed regions shown in A and B. A and A”) Tagln-Cre expression, as reported by R26R-lacZ, is observed in the testis interstitial compartment at E14.5. A”) β-GAL is detected within perivascular cells but is usually absent from the endothelium (cv, coelomic vessel) and Sertoli cells and germ cells within testis cords (tc). B and B”) VCAM1 expression is upregulated in Mafb-GFP-dim interstitial cells and is weakly expressed in Mafb-GFP-bright Leydig cells. In control E14.5 testes (C), VCAM1 is observed throughout the interstitial compartment; but in E14.5 Tagln-Cre, Jag1^lox/lox (smooth muscle [SM] Jag1 KO) testes (D), VCAM1 expression is nearly absent in the interstitium. D) Asterisk marks a large cluster of autofluorescent Leydig cells in SM Jag1 KO testes that is not normally observed in control testes. HSD3B1 staining confirmed an increased number of differentiated fetal Leydig cells in SM Jag1 KO testes relative to controls at E 14.5 (E and F). Bars = 50 μm in all panels.

To assess the differentiated status of interstitial cells, we used HSD3B1 to mark mature FLCs and VCAM1 to mark undifferentiated interstitial cells and early Leydig progenitors [10]. We observed that in later fetal stages (E14.5 and older), VCAM1 was downregulated in Mafb-GFP-positive Leydig cells, and strong VCAM1 expression was maintained in all other interstitial cells (presumably also in Leydig progenitors) (Fig. 3, B and B”).

In wild-type or Tagln-Cre; Jag1^lox/+ control testes, there was a balance between VCAM1-positive undifferentiated interstitial cells and VCAM1-negative differentiated Leydig cells (Fig. 3, B” and C). However, in Tagln-Cre; Jag1^lox/lox testes, there was a dramatic decrease in VCAM1-positive cells (Fig. 3, C and D), with the substantial VCAM1 staining detected only in the surface domain of the testis near the coelomic artery. Concomitantly, there was a substantial increase in the number of HSD3B1-positive cells (Fig. 3, E and F), suggesting that interstitial cells were shifted to a differentiated Leydig cell fate. These results showed that Jag1 expression in interstitial cells is required to maintain fetal Leydig progenitors in an undifferentiated state, and demonstrated that loss of Jag1 in interstitial cells recapitulates Notch pathway loss of function in the testis [11].

Notch Signaling Declines in the Testis Interstitium in Postnatal Life

To assess any potential role of Notch signaling in postnatal testis development, we undertook a temporal analysis of TNR-GFP [14] expression at various time points between perinatal stages (P3) and adulthood (P60). Similar to fetal stages, Sertoli cells were GFP-positive throughout all stages of development [21]. However, we saw dynamic changes in GFP expression within the interstitial compartment. At P3, numerous intensely bright GFP-expressing cells were present in the interstitium, and these cells were almost always located immediately adjacent to vasculature (Fig. 4, A and A'). These cells, unlike mature Leydig cells, were actively in cell cycle, as they strongly expressed MKI67 (Fig. 4B). Additionally, these bright GFP-positive cells were interspersed among JAG1-positive perivascular cells (Fig 4C). Some of these GFP-bright cells coexpressed HSD3B1 (Fig. 4D), likely a result of GFP persistence in FLCs that recently differentiated from Notch-active progenitors.

FIG. 4.

TNR-GFP expression declines in the postnatal testis interstitium. Immunofluorescent images of postnatal testes. (A', D', E', and F') Higher magnification images of the boxed regions in A, D, E, and F, respectively. Dashed gray outlines denote testis cord/seminiferous tubule boundaries. In the P3 testis (A–D'), there are TNR-GFP-bright cells immediately adjacent to PECAM1-positive vasculature (A and A', arrows), which are MKI67-positive (B, arrows). B) Insets show GFP (green) and MKI67 (red) channels of TNR-positive cells (arrows) next to a cross-section of vasculature (blue). P3 GFP- bright cells are adjacent to JAG1-bright perivascular cells (C). GFP-expressing cells are mostly undifferentiated, but some GFP-bright cells are HSD3B1-positive Leydig cells (D and D'). E) At P12 fewer interstitial cells are GFP-positive (arrow), but some vasculature is TNR-positive (arrowheads). F) At P18 very few strong GFP-positive interstitial cells remain (F', arrow), but vasculature expresses GFP strongly (F', arrowheads). By P30 (G) and P60 (H), GFP expression remains strong within Sertoli cells in seminiferous tubules [21] but is nearly completely absent from the interstitium. Bars = 25 μm in all panels.

At P12, there were fewer GFP-bright cells in the interstitium, but those present were still vasculature-associated (Fig. 4, E and E'). There was also sporadic GFP expression within PECAM1-positive endothelial cells (Fig. 4E'). At P18, there were even fewer remaining GFP-positive cells in the interstitium, and the only bright interstitial cells were vascular endothelial cells (Fig. 4F). By P30, there were very few interstitial GFP-positive cells remaining (Fig. 4G), and by P60, there was no GFP expression detected in the interstitium (Fig. 4H), apart from rare GFP-positive endothelial cells. This suggests that active Notch signaling plays a role in FLC development and in the transition to ALCs prior to puberty, but is unlikely to be required for maintenance of the ALC population or its progenitor pool after sexual maturity.

Age-Related Notch Pathway Expression Changes in the Testis

To quantitate molecular differences in the Notch pathway and other interstitial genes, we used qPCR to compare expression among the P3 perinatal testis (containing only fetal Leydig cells), the P12 testis (containing very few fetal or adult Leydig cells) and the P60 adult testis (containing only adult Leydig cells) (Fig. 5). To reduce any confounding results from seminiferous tubule gene expression, we mechanically removed the tubules from testis tissue, and isolated interstitial cells for qPCR analysis.

FIG. 5.

Expression of Notch genes and Leydig progenitor genes decrease in interstitial cells during postnatal development. qPCR expression analysis of Notch-associated genes (A) and interstitium-specific genes (B) at stages P3, P12, and P60. Expression is normalized to P3 expression level (set to 1). Results are shown as a means ± SEM of three independent homozygous TNR-GFP isolated testis interstitium samples for each stage.

In addition to Notch ligands expressed in the postnatal testis (Dll1, Dll4, Jag1, Jag2), Notch receptors (Notch1-4), and a Notch target gene (Hes1), we measured expression of other genes within interstitial or endothelial cell populations (Arx, Lhx9, Nes, Pdgfra, Ptch1, Vcam1, Cdh5) and steroidogenic cells (Cyp11a1, Cyp17a1). All Notch ligands except Jag2 decreased in expression from P3 to P12, and all decreased from P3 to P60 except Dll1 and Jag2. All Notch receptors and the Notch target gene Hes1 showed a reduction in expression during postnatal development. Genes encoding steroidogenic enzymes (Cyp11a1 and Cyp17a1) showed a decline between P3 and P12 as expected, but were expressed at their highest levels at P60.

Arx, Lhx9, Nes, Pdgfra and Ptch1, which are thought to be expressed in undifferentiated, Leydig precursor cells [11, 29–32], all significantly decreased in expression from P3 to P60. Interestingly, the expression of the endothelial-cell-specific gene Cdh5 was significantly decreased at P60, while Vcam1 increased at P12 and P60. Consistent with qPCR results, immunostaining revealed that VCAM1 expression expanded after birth and was detected in ALCs in the adult testis (Supplemental Fig. S2). These data suggest that progenitor cells of ALCs express different markers than progenitors of FLCs, and are likely regulated by a signaling pathway distinct from Notch.

Elevated Circulating Testosterone in the Testis Results in Increased Notch Signaling

The tight association of TNR-GFP-expressing cells with the vasculature raised the possibility that a circulating factor might affect Notch signaling, thereby regulating the progenitor cell population based on physiological requirements. One circulating factor that is critical for male development is testosterone, which is required for virilization of external genitalia and, later in adulthood, for maintaining spermatogenesis. We speculated that elevated circulating testosterone levels might stimulate Notch signaling to maintain the progenitor pool and reduce the number of differentiated, steroidogenic Leydig cells. Steroidogenic gene activity and intratesticular testosterone levels are high from E12.5 throughout fetal and early postnatal stages [4–6, 33], and difficult to manipulate in utero. However, between P10 and P25, there is a period in which fetal Leydig cell numbers diminish and the testis normally reaches a nadir of testosterone production prior to the differentiation of adult Leydig cells (Fig. 6A) [4–6]. To investigate whether exogenous high levels of testosterone affect Notch signaling, we injected P7 TNR-GFP mice with testosterone propionate for 5 days prior to euthanization at P12 (Fig. 6A). Analysis of control and experimental cohorts of mice showed that there was no statistical differences between body weight, testis weight, or percentages of testis weight (Supplemental Fig. S3), indicating that our testosterone treatment regimen had no physiologically deleterious effects.

FIG. 6.

Interstitial Notch signaling is increased in testosterone-treated testes. A) Cartoon shows time of development between late fetal stages to adulthood. A decline in testosterone levels and FLC numbers is followed by a decrease in active Notch signaling in the testis interstitial compartment during the transition to adult Leydig cells. Testosterone levels are modeled after data in reference 4 and represent relative intratesticular concentrations. Red bar represents duration of testosterone treatment experiments (from P7 to P11), which artificially maintained high testosterone levels, similar to that observed perinatally. B and C) Immunofluorescent images of postnatal P12 testes. B) In vehicle-treated testes, GFP-bright nonvascular interstitial cells are observed (arrowheads). C) In testosterone-treated samples, an increased number of interstitial nonvascular bright GFP-positive cells is observed (arrowheads). Bars = 50 μm. D) Quantitation of interstitial GFP-positive cells per unit area (1 square millimeter). Values are represented as means ± SEM. *P < 0.0001.

Confocal microscopy and immunofluorescent analysis of vehicle-injected and testosterone-injected testes revealed a significantly higher number of TNR-GFP-positive interstitial progenitor cells in the testosterone-injected group (Fig. 6, B and C). The vehicle-treated group had a mean 65.1 cells per sq. mm, while the testosterone-treated group had 101.0 cells per sq mm (P < 0.0001) (Fig. 6D).

To measure the effect of high testosterone levels on expression levels of Notch signaling pathway components and markers of the undifferentiated interstitial population, we compared qPCR results of isolated interstitial cells from testosterone-treated testes with those from vehicle-injected controls. Our analysis revealed that mice with elevated testosterone expressed significantly higher levels of Jag1 and Notch2 in interstitial cells than in controls, whereas expression levels of the other interstitial-enriched Notch receptor gene Notch3 were not significantly different between the two groups (Fig. 7A). Immunostaining for JAG1 revealed that in contrast to vehicle-injected testes in which JAG1 was strongly expressed only in perivascular cells (Fig. 7B), testosterone-treated testes had extensive JAG1 expression throughout the entire interstitial compartment (Fig. 7C). As expected, expression levels of the steroidogenic enzyme-encoding genes Cyp11a1 and Cyp17a1 were lower in testosterone-treated mice, indicative of a loss of endogenous steroidogenesis in the presence of high levels of exogenous steroids (Fig. 7D).

FIG. 7.

Testosterone treatment results in increased expression of Notch genes and undifferentiated markers and decreased endogenous steroidogenesis. A) qPCR expression analysis of Notch-associated genes at P12 after daily treatment with vehicle or testosterone (T) for 5 days from P7–P11. B and C) Immunofluorescent images show vehicle-treated P12 testes with only perivascular JAG1 staining (B), but testosterone-treated samples exhibit an expansion of JAG1 expression throughout the entire interstitial compartment (C). C) Arrow indicates the strongest expression of JAG1 in testosterone-treated testes is still in perivascular regions; dashed lines represent testis cord/seminiferous tubule boundaries. Bars = 50 μm. D) qPCR expression analysis of interstitium-specific genes. All qPCR results are shown as means ± SEM from three independent homozygous TNR-GFP isolated testis interstitium samples for each experimental condition (oil-treated control males versus experimental testosterone-treated males). *P < 0.05.

Another expected result of an increased number of TNR-GFP-bright progenitor cells was increased expression of Leydig progenitor genes, which under normal conditions, decrease in expression between P3 and P12. Consistent with this prediction, expression of Lhx9, Nes, Pdgfra, and Ptch1 increased in isolated interstitial cells from testosterone-treated testes. Of this group of progenitor markers, only Arx showed no significant difference (Fig. 7D).

Perivascular Cells Express Androgen Receptor

To assess whether postnatal perivascular cells can sense testosterone, we examined the expression of androgen receptor (AR), the receptor for testosterone. Early in postnatal testis development (at P3), virtually all interstitial cells except endothelial cells and differentiated FLCs expressed nuclear AR (Fig. 8, A and B). This may be associated with the fact that FLCs (due to a lack of Hsd17b3 expression during fetal and postnatal stages) synthesize only androstenedione, which is transferred to fetal/postnatal Sertoli cells. Sertoli cells then convert androstenedione to testosterone [34, 35]. Secretion of testosterone from Sertoli cells may expose all interstitial cells to testosterone before it enters the blood vessels.

FIG. 8.

Postnatal JAG1-positive perivascular cells strongly express androgen receptor (AR). Immunofluorescent images of postnatal P3 (A and B) and P12 (C–G) TNR-GFP wild-type testes (CD-1 genetic background). A', B', C', C,” F', and G') Higher magnification images are shown of the corresponding boxed regions in A, B, C, C', F, and G, respectively. Black and white images are single-channel images of the corresponding panel. Dashed lines indicate testis cord/seminiferous tubules boundaries. A and B) In the P3 testis, most interstitial cells have nuclear AR expression, including peritubular myoid cells (white arrowheads throughout figure) and perivascular cells (green arrows throughout figure). Differentiated Leydig cells (CYP17A1-positive [yellow arrows throughout figure]) and endothelial cells (PECAM1-positive [red arrows throughout figure]) weakly express cytoplasmic AR or are AR-negative. C–E) At P12, perivascular cells are strongly JAG1/AR double-positive (green arrows in C'), while endothelial cells (E) are AR-negative (D and E). Longitudinal (D) and cross-sectional (E) views of a blood vessel showing AR-positive (D) or JAG1/AR double-positive perivascular cells (E). F and F') Vehicle-treated control testes show restriction of JAG1 and AR expression to perivascular regions, while testosterone-treated testes (G and G') had expanded JAG1 and AR expression throughout the interstitium. Bars = 50 μm.

However, by P12, the stage at which we assessed the testis after our testosterone administration regimen, in wild-type testes, overall interstitial AR expression significantly decreased, but JAG1-positive perivascular cells maintained high levels of nuclear AR (Fig. 8C). P12 perivascular cells uniformly expressed nuclear AR (Fig. 8, D and E), but endothelial cells and other interstitial cells, such as FLCs did not. At this stage, perivascular cells are well situated and capable of responding to circulating factors (such as exogenous injected testosterone) coming from the bloodstream. After testosterone treatment, nuclear AR expression was expanded among ectopic JAG1-positive cells (Fig. 8, F and G; see also Fig. 7C), showing that nuclear AR expression in the interstitium can be stimulated by high testosterone levels.

Overall, these data provide evidence that JAG1-expressing vasculature-associated cells act as sensors for circulating testosterone levels and respond by elevating Notch signaling to adjust the balance between progenitors and differentiated cells in the testis.

DISCUSSION

Leydig cells are critical to drive the differentiation of the reproductive tract and maintain fertility throughout the male reproductive lifespan. However, how the testis maintains a balance between differentiated Leydig cells and their progenitors is unclear. We showed that the Notch ligand JAG1 is enriched in perivascular interstitial cells in the fetal testis and is specifically required in interstitial cells for the maintenance of FLC precursors. In the early postnatal testis, some perivascular cells undergoing active Notch signaling persist during the transition to ALCs. Given the close association of the vasculature with these precursor cells, we tested whether a circulating factor, such as testosterone, could regulate Notch signaling. During the early postnatal period, the neonate normally experiences a decline in FLC number and testosterone levels, as well as a substantial decrease in Notch signaling associated with the disappearance of FLC progenitors (Fig. 6A). We showed that during this transition, maintenance of high circulating testosterone levels resulted in increased Notch activation and Notch pathway gene expression, as well as an increase in interstitial and Leydig progenitor gene expression consistent with maintenance of Leydig progenitors. These data provide strong evidence that a circulating substance such as the hormone testosterone can influence a signaling pathway that controls a progenitor cell niche (Fig. 9).

FIG. 9.

A model for regulation of fetal Leydig cell precursors. Cartoon shows a region of the testis interstitial compartment highlighting interactions between JAG1/AR+ perivascular cells and TNR-GFP cells that maintain Leydig progenitors in an undifferentiated state via Notch signaling. Dashed arrows indicate diffusion of testosterone or other factors from the vasculature that influence signaling from perivascular cells.

Defining a Role for Notch Within the Leydig Progenitor Niche

The effect of Notch signaling on the maintenance/differentiation of Leydig progenitors was previously shown. The Notch target gene Hes1 is critically important downstream of Notch signaling to maintain the Leydig progenitor pool. The Hes1 mutant phenotype in the fetal testis is virtually identical to the phenotype seen after DAPT treatment, which represents complete Notch pathway loss of function. These genetic studies indicated that Notch signals through Hes1 within the FLC progenitors [11].

However, neither the ligand driving Notch signaling nor the signaling and receiving cells that constitute the FLC progenitor niche had been identified. To determine which Notch ligand and receptor are involved in Notch pathway activation, we took advantage of our transcriptome analysis of purified cell populations within the fetal testis [12] as well as a Notch reporter mouse [14]. This analysis indicated that Jag1, Notch2 and Notch3 are all expressed in interstitial cells and represent good candidates. Immunocytochemical analysis revealed that JAG1 was enriched in perivascular interstitial cells near cells where active Notch signaling was taking place. Conditional deletion of Jag1 in interstitial cells revealed a phenotype indistinguishable from loss of function of Hes1, suggesting that Notch2/3, Jag1 and Hes1 all belong to a pathway that regulates FLC progenitors.

Hes1 is broadly responsible for maintaining stem cells in an undifferentiated state to allow the progenitor pool to expand prior to differentiation. For example, loss of Hes1 function in the brain resulted in premature neurogenesis, thus exhausting the neural stem cell population and causing severe neural tube defects [36]. One paradigm that is similar to the FLC progenitor niche occurs in the cerebellar cortex, where Notch2 is responsive to activation by JAG1, and is specifically expressed in proliferating granule neuron precursors prior to their differentiation. In this instance, Notch acts to inhibit neurogenesis by keeping neural progenitors in a proliferative state when signaling is active [37].

The maintenance of stem cells in a proliferative and undifferentiated state by Hes1 may be partly due to its transcriptional repressive activity upon the cell cycle inhibitor gene Cdkn1b (also known as p27^Kip1) [38]. However, ChIP-Seq (chromatin immunoprecipitation followed by next-generation sequencing) experiments to determine HES1 targets in the testis would definitively determine whether Hes1 acts directly to repress Leydig cell differentiation genes, or if its influence on Leydig cell progenitors is secondary to its effect on cell cycle.

Interaction Between Vascular Microenvironment and the Stem Cell Niche

The proximity of stem cell populations to vasculature is a recurring theme in mammalian stem cell niches. Perivascular cells maintain the hematopoietic stem cell niche [39] and the neural stem cell niche in the brain [40, 41]. Specifically in the adult testis, experiments from Yoshida et al. [42] revealed there is a vasculature-associated niche for spermatogonial stem cells. In addition, Davidoff et al. [30] also suggested that there is a vasculature-proximal cell population that can serve as a progenitor pool after Leydig cell depletion by treatment with the alkylating agent EDS (ethane dimethanesulfonate). Whether these perivascular cells represent a stem cell population under steady-state conditions or whether only under injury conditions does a differentiation or transdifferentiation event take place remains an open question.

Vascular endothelial and mural cells may act directly to regulate stem cells through contact-based cellular communication or cell adhesion. Our results suggest that VCAM1 may be one molecule that coordinates extracellular contact between different interstitial populations. In our analysis, VCAM1 was excluded from differentiated FLCs (Fig. 3B), but was enriched in ALCs (Supplemental Fig. S2). It is possible that VCAM1 initially labeled an undifferentiated interstitial pool containing a potential ALC precursor population, which is distinct from FLCs (which do not express VCAM1) during late fetal stages. These data are consistent with the idea that FLCs and ALCs are unique lineages with different origins. Further characterization of VCAM1 function may yield additional insights into the cell-cell interactions taking place in the interstitium.

Hormonal Influence on Stem Cell Niches

In addition to regulation through cell contact, systemic circulating factors coming from the vasculature may also be involved. In vitro cell culture experiments showed that endothelial cells themselves can independently influence neural stem cell behavior [40], but in vivo, it is likely that perivascular cells and circulating levels of oxygen and metabolites may also be critical to maintain and regulate stem cell populations. One circulating factor that is critical for testis function is testosterone, which is required in fetal and adult stages for male reproductive tract development and for spermatogenesis. One study highlighted the importance of androgen receptor (AR) signaling in adult Leydig cell (ALC) development, by showing that AR-deficient mice have normal fetal Leydig cell (FLC) proliferation, but little to no ALCs [43]. In that study, the AR-deficient testis exhibited interstitial cell hyperplasia, but had reduced expression of differentiated ALC marker genes, leading the investigators to hypothesize that ALCs were specified but failed to differentiate normally in the absence of androgen signaling [43]. In this study, we showed that AR is expressed in perivascular cells and likely plays a role in maintaining the balance between progenitor cells and differentiated FLCs via sensing testosterone levels, but is likely not required for initial FLC differentiation since androgens are not present until FLCs differentiate.

Studies in EDS-treated adult rats that were given exogenous testosterone revealed that high levels of testosterone dramatically suppressed Leydig cell regeneration [44]. It was not definitively shown, but likely, in those EDS studies that Leydig progenitors were maintained in an undifferentiated state in the presence of high testosterone levels. These findings are consistent with this study showing that high testosterone levels leads to an increased number of undifferentiated Leydig progenitors, resulting from increased Notch signaling. This phenomenon could be the result of changes in LH levels, which are sensitive to testosterone in a negative feedback loop acting at the level of the pituitary [45, 46]. LH receptors are expressed on the surface of Leydig cells starting only in late fetal stages [33], and LH is critical for postnatal, but not fetal, testis and reproductive tract development [47, 48]. Additionally, LH was critical for Leydig cell recovery and function after EDS exposure [49], in accord with the established role of the hypothalamic–pituitary–gonadal axis in the differentiation of Leydig cell progenitors.

Our data links these systemic hormonal phenomena to Notch signaling, which keeps Leydig progenitors in a naïve state and maintains their expression of progenitor markers like Lhx9 and Nes. These results suggest a negative feedback mechanism that determines the rate at which Leydig cells are replaced. We show that Jag1 expression, normally limited to cells located near vasculature, is sensitive to changes in the level of testosterone: high levels induce high levels of Jag1 throughout the interstitium, and downstream Notch signaling that blocks differentiation of progenitors, whereas low levels of testosterone result in low levels of Notch signaling and the differentiation of more androgen-producing cells.

A drop in testosterone levels during the postnatal period precedes the decline in the FLC-specific progenitor population, and may be a mechanism to divert from Notch-maintained progenitors to another system responsible for the adult population. Whether the signaling niche that controls ALC progenitors also responds to testosterone levels remains to be determined. Other hormone-sensitive tissues, such as the mammary gland and brain, have been used as models for studying the influence of hormones upon stem cells [50, 51]. This paradigm also applies to other phyla, as a recent study demonstrated that ecdysone signaling is important in regulating germline stem cells in Drosophila [52], suggesting that aspects of systemic control over tissue stem cell niches are evolutionarily conserved.

Our study reveals the physiological complexity of the Leydig progenitor niche and links the systemic circulation to a signaling pathway controlling the maintenance/differentiation decision in a stem cell population. The regulation of Leydig cell progenitors is a critical issue in reproductive development, and understanding the mechanisms that drive the balance between self-renewal and differentiation in this stem cell system has clinical implications for the etiology of disorders of sexual development and male infertility.