Single-cell resolution uncovers neighboring cell subtypes that share steroidogenic capacity during fetal testis development

Significance

These data uncover the exquisite timing and the means by which individual cells differentiate, mature, and cooperate within microenvironments of the fetal testis to produce a critical level of testosterone necessary for optimal male embryo development. We uncovered the neighborhoods and individual cells that are at first responsive to stimulation, and then repression of steroidogenic pathway gene expression within a tightly regulated window of time. Further, complementary use of single-cell RNA sequencing and single-molecule fluorescent in situ hybridization data allowed us to visualize the potential for individual cells to cooperate with neighboring cells to shuttle and convert steroid metabolites into testosterone.

Abstract

Historically, endocrine cells were perceived to coordinate their output in a uniform manner. Recently however, single-cell technologies have uncovered heterogeneity within these populations, indicating that individual cells may operate as independently regulated units. Using high-resolution tools such as single-molecule fluorescent in situ hybridization (sm-FISH) and single-cell RNA sequencing (scRNA-seq), we investigated the contributions of individual and the collective of fetal Leydig cells to androgen production over time during mouse testis development. Temporal profiles of intratesticular androgens alongside the expression of steroidogenic pathway genes (Star, Cyp11a1, Cyp17a1, and Hsd3b1) from prenatal to perinatal testes demonstrated that the peak in gene expression preceded the peak in androgen production. Spatially, steroidogenic cells were initially observed to be concentrated toward the anterior–posterior poles along the center of the dorsal–ventral axis of the fetal testis at embryonic day (E) 13 and then expanded to a uniform distribution by E16. Next, sm-FISH using probes for individual steroidogenic pathway genes exposed the following findings: gene transcription and processing of individual and combinations of steroidogenic pathway genes are not synchronized among fetal Leydig cells; and some fetal Leydig cells express incomplete sets of genes. Further, sm-FISH and scRNA-seq data corroborated the presence of fetal Leydig and other interstitial cell types harboring incomplete sets of steroidogenic pathway genes throughout developmental stages. Taken together, these findings highlight that fetal steroidogenic gene expression is tightly regulated and that transcript presence among interstitial cell types promotes the possibility that optimal androgen biosynthesis results from a cooperative effort among neighboring steroidogenic cells.

Certain organs comprise functional units, such as the hepatic acinus in the liver (1), renal glomerulus in the kidney (2), the islet of Langerhans in the pancreas (3), and steroidogenic cell zones or clusters within the adrenal gland or gonad (4–6). Each unit operates independently but its regulation is coordinated for optimal organ function. Historically, it was postulated that individual cells within these functional units, such as Leydig cells in the testis (7, 8) or β cells in the islet of Langerhans (9), synchronized their output. The advent of single-cell RNA sequencing (scRNA-seq) has increased our appreciation for greater cellular heterogeneity by demonstrating distinct transcriptional profiles for each cell, even within a functional unit, suggesting the possibility that each cell can also be independently regulated (10). The challenge is to visualize and understand individual cell contributions, their coordinated output, and their functional significance within local microenvironment units as they relate to the optimal output of the whole organ.

Gonadal hormones, particularly androgens, play a vital role in the establishment of sex-dimorphic phenotypes in mammals during pre- and perinatal development that determines vitality, behavior, and fertility later in life (11–15). Androgens facilitate the masculinization of the male reproductive system, including the testes, the reproductive tract, and the external genitalia during a crucial developmental phase known as the masculinization programming window (MPW), which occurs between E15.5 and E18.5 in rats (16), gestational week (GW) 7 to 14 in humans (17), and is anticipated to transpire between E13.5 and E16.5 in mice (16). After the MPW has been achieved, fetal androgen output is rapidly suppressed until a transient resurgence occurs in the neonate, termed minipuberty (14, 15). Thereafter, androgens remain suppressed until just before the onset of puberty (15, 18). Male fetuses that are exposed to inadequate levels of androgens experience variations in sex differentiation (VSD), with phenotypes including cryptorchidism, micropenis, and hypospadias (14, 19).

Leydig cells are the primary source of androgens and are organized in clusters intermixed with other cell types within the interstitium of the testis. Two distinct types of Leydig cells have been identified: fetal and adult Leydig cells (20). The pathways for fetal and adult Leydig cell differentiation have been debated, but recent investigations utilizing scRNA-seq and lineage tracing analyses indicate that these cells, along with other interstitial cells, populate the fetal testis from coelomic epithelial and mesonephros origins to initiate differentiation from a common precursor cell (10, 21–24). A myriad of local signals, including Desert Hedgehog (DHH), fibroblast growth factor 9 (FGF9), Retinoic acid, Platelet Derived Growth Factor (PDGF), Wnt family member 5a (WNT5a), and NOTCH nudge interstitial cells toward differentiation into fetal Leydig cells, peritubular myoid cells, smooth muscle cells, and pericytes while constraining others as progenitor adult Leydig or other cells (21, 22, 24–27). Notably, studies also indicate that cell status is fluid over developmental time, suggesting the potential for individual cells to transition their differentiation state to play related but distinct roles as the testis develops and matures (22, 28, 29).

The present research utilizes murine fetal Leydig cells as a model to investigate their development and their interactions with neighboring cells to understand how they collectively contribute to the final androgen output. We profiled fetal Leydig cell activity throughout development, noting that intratesticular androgen levels and the expression of steroidogenic pathway genes reach their peak during mid to late gestation, followed by a decline around the time of birth. Three time points, representing the onset (E13), peak (E16), and nadir (P0) of androgenesis, were selected to investigate the steroidogenic activity of individual cells through single-molecule fluorescent in situ hybridization (sm-FISH) targeting Star. This investigation confirmed heterogeneity present among transcript-positive cells as three distinct transcriptional stages for Star were identified, demonstrating that Star expression was not synchronized among cells. Costaining sm-FISH analyses of Star and Cyp17a1 further established that steroidogenic pathway gene members can be, but are not always, synchronized within individual cells. Using scRNA-seq analysis, we also observed that multiple cell types, including those annotated as fetal Leydig cells, can harbor an incomplete profile of steroidogenic pathway genes, a finding validated using sm-FISH. In sum, the combination of scRNA-seq and sm-FISH data provides corroborating evidence that distinct, neighboring interstitial cell types, including fetal Leydig cells, express complete or subsets of steroidogenic pathway genes that collaborate to provide a microenvironment that likely optimizes androgen output from the whole fetal testis.

Results

Intratesticular Androstenedione and Testosterone Levels Peak at E18.

In fetal Leydig cells, de novo testosterone synthesis begins with the translocation of cholesterol from the cytoplasm to the inner mitochondrial membrane by the steroidogenic acute regulatory protein (STAR). Cholesterol is converted to pregnenolone by CYP11A1 and subsequently to progesterone by HSD3B1. The process then transitions to the smooth endoplasmic reticulum, where CYP17A1 catalyzes two reactions: the conversion of progesterone to 17-OH progesterone and its subsequent transformation into androstenedione. Fetal Leydig cells do not complete conversion into testosterone; instead, androstenedione must reach the Sertoli cell, where HSD17B3 manages the final conversion (Fig. 1A) (30–32). To assess intratesticular steroid accumulation along the testosterone synthesis pathway during development, we used LC/MS/MS on lysates from fetal and perinatal testes from embryonic day (E) 12 to postnatal day (P) 6 and adult testes from mice 6 wk or older. All steroid measurements were below the limit of detection at E12; therefore, results are presented starting at E13. The concentrations of progesterone and 17-OH progesterone were consistently low across all measured time points, which could be attributed to the rapid conversion into androstenedione. Meanwhile, androstenedione and testosterone were detected at substantially higher levels, with both increasing between E13 and E15, stabilizing until E16, followed by a significant increase until E18. Thereafter, concentrations rapidly declined and reached their lowest point at P1 and then increased, as expected, in the adult testis (Fig. 1 B–F).

Fig. 1.

Temporal profiles of intratesticular steroid levels from prenatal, perinatal, and adult testis. (A) An illustration of the testosterone biosynthesis pathway. Fetal Leydig cells produce androstenedione, which is converted to testosterone in the Sertoli cell. Steroid intermediates are color coded to match results presented in B–F. Steroid pathway mediators are color coded to match transcript data presented in Fig. 2 A and C. (B) Intratesticular progesterone, 17-OH progesterone, androstenedione, and testosterone levels from n = 3 biological replicates each of E13–P3 (pooled 2 to 10 testes each replicate), P6 (pooled a pair of testes each), and adult testes (n = 6, pooled a pair of testes each). Pooled testes were harvested from at least three different litters. (C–F) Individual steroid levels; (C) Progesterone, (D) 17-OH progesterone, (E) Androstenedione, and (F) Testosterone. Mt: mitochondria, ER: endoplasmic reticulum, 17-OH Progesterone: 17-hydroxyprogesterone. Data are presented as the mean from the biological replicates ± SEM. Unit: ng of steroid per gram testis wet weight.

Steroidogenic Pathway Gene Expression Peaks at E16.

The dynamic production of testicular androgens over time suggests corresponding changes in the expression of steroidogenic pathway genes. To examine the expression of steroidogenic pathway genes during embryonic and early postnatal testis development, bulk RNA was extracted from urogenital ridges from E11 to E12.75, and from testes only from E13 to P6. To capture underrepresented transcripts during early development and minimize normalization biases introduced by housekeeping genes, RNA was converted to cDNA using gene-specific primers within a gBlock for copy number real-time qPCR (RT-qPCR). This approach also enabled direct comparisons of transcript levels across different genes. Transcript copy numbers were determined using their C_T values and an external standard. The analysis focused on several factors including Star and all steroidogenic enzymes expressed in fetal Leydig cells: Cyp11a1, Cyp17a1, and Hsd3b1 (SI Appendix, Fig. S1). The final conversion enzyme, Hsd17b3, is expressed in Sertoli cells. The steroidogenic pathway genes exhibited similar temporal expression profiles characterized by a gradual increase between E11 and E13, a rapid escalation during the MPW (E13 to E16), and a sharp decline by P0 (Fig. 2A and SI Appendix, Fig. S1 A–E).

Fig. 2.

Temporal profiling of steroidogenic pathway genes in prenatal and perinatal testes. (A) Copy number RT-qPCR result of Star, Cyp11a1, Cyp17a1, and Hsd3b1 (n = 3) from bulk RNA harvested from testes spanning E11.5 to P6. (B) Fetal Leydig cell number per testis quantified using Star-positive interstitial cells using smFISH at E13, E16, and P0 (birth) (n = 4; Figs. 3 and 4). (C) Values represent the mean ± SEM copy number transcript results of Star, Cyp11a1, Cyp17a1, and Hsd3b1 from each biological replicate normalized to the overall average fetal Leydig cell numbers reported in B for E13, E16, and P0. (D) An illustration of randomized fetal Leydig cells selected for FISH-quant transcript quantification analysis. Intersected rays were drawn from the coelomic vessel at 15 to 30° angles. (E) Values represent the average number of Star transcripts, calculated using FISH-quant, of 60+ Star-positive cells detected along the rays drawn in D for each of four testes at each time point. The raw data from each testis are presented in SI Appendix, Fig. S2 (n = 4). All values are presented as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. E: embryonic day, P: postnatal day, FLC: fetal Leydig cell; testis samples were collected from at least two litters.

To assess whether changes in steroidogenic gene expression were due to changes in fetal Leydig cell numbers, we quantified these cells at E13, E16, and P0 using sm-FISH for Star as outlined in detail in Fig. 4. We chose to evaluate Star as a representative marker for fetal Leydig cells because it is the established gatekeeper for the steroidogenesis pathway (33, 34). We counted an average of 4,262 ± 762 fetal Leydig cells at E13, 14,672.5 ± 1,775 at E16, and 16,492 ± 1,300 at P0 (Fig. 2B, n = 4 each timepoint). Next, copy number PCR levels were normalized to the average number of Star-positive interstitial cells counted at each time point (Fig. 2C). Results mirrored the temporal expression trends observed in Fig. 2A, suggesting that changes in gene expression were not related to changes in Star-positive cell numbers.

Fig. 4.

Spatiotemporal distribution profiles of Star expressing cells in developing testes. (A) An illustration of the sectioning and imaging processes and classification of Star-transcript stages in transcript-positive cells identified within testis interstitium. Transverse cryosections (5 μm) were obtained every 50 μm from the anterior to the posterior poles of each testis and each section was further segmented into 50 μm (E13) or 100 μm (E16 and P0) concentric arcs. (B) Proportions of different Star transcript Stages at E13, E16, and P0. All Star-positive cells within the interstitium were counted. (C, D) Bar graphs of Star-positive cell density [cell/area (10,000 μm2)] of E13 testes along the anterior-posterior axis (C) and the dorsal-ventral axis (D). (E-H) 3D Heatmaps of Star-positive cell density of E13 testes. Each datapoint represents the mean of one concentric arc of four different testes. Legend unit: cell/area (10,000 μm2). Color bar representing low (purple) to high (yellow) density along the right side of the heatmap. (E) Stage I; (F) Stage II; (G) Stage III; (H) All stages together. (I, J) Star-positive cell density of E16 testes along the anterior-posterior axis (I) and the dorsal-ventral axis (J). (K-N) 3D Heatmaps of Star-positive cell density of E16 testes. (K) Stage I; (L) Stage II; (M) Stage III; (N) All stages together. (O, P) Star-positive cell density of P0 testes along the anterior-posterior axis (O) and the dorsal-ventral axis (P). (Q-T) 3D Heatmaps of Star-positive cell density of P0 testes. (Q) Stage I; (R) Stage II; (S) Stage III; (T) All stages together. Note scale ranges differ at each time point. A: anterior, P: posterior, D: dorsal, V: ventral, Mid: Middle; data at each time point represents four testes from different embryos of at least two litters.

While the number of Star-positive, fetal Leydig cells increases from E13 (Fig. 2B), testicular volume and total cell number also increase substantially during this period. Consequently, the apparent decrease in fetal Leydig cell gene expression detected by copy number qPCR after E16 could reflect a dilution effect rather than a reduction in transcript levels. To evaluate this possibility, we used sm-FISH to directly quantify Star transcript numbers within individual cells at E13, E16, and P0. The number of Star-transcripts per cell was quantified from 92 cells at E13, 123 cells at E16, and 91 cells from P0 using FISH-quant (35). Positive cells that intersected lines radiating from the coelomic vessel were chosen for transcript quantification within sections from four testes (Fig. 2D). Results showed the average number of Star transcripts per cell corresponded to both raw and normalized copy number qPCR results (Fig. 2E and SI Appendix, Fig. S2). Together, these data confirm that steroidogenic pathway gene expression in fetal Leydig cells precedes the testosterone production profile by approximately 2 d. Furthermore, the temporal expression pattern aligns with the MPW and rapidly declines shortly before birth. The observation that this transcript profile is independent of fetal Leydig cell number supports the notion that steroidogenic pathway genes are actively and tightly regulated during development.

Steroidogenic Pathway Gene Expression Is Not Synchronized Among Neighboring Fetal Leydig Cells.

Based on the observation that steroidogenic pathway genes are regulated, we hypothesized that the fetal Leydig cell microenvironment influences their expression. To test this, we conducted sm-FISH analysis of Star as a representative marker for fetal Leydig cells, to understand the transcription status of fetal Leydig cells within their microenvironments. Guided by the temporal steroidogenic pathway gene expression profile, spatial expression analysis was conducted on E13, E16, and P0 testes, covering the onset, peak, and end of steroidogenic pathway gene expression, respectively. Interstitial cells with Star signals were included in this analysis. Representative images are shown in Fig. 3 A and B.

Fig. 3.

Star mRNA sm-FISH of developing testes. (A and B) Representative transverse section from E13, E16, and P0 (newborn) testis. [Scale bar, (A) 100 μm and (B) 25 μm.] [yellow arrow: Stage I; filled triangle: Stage II; white arrow: Stage III.] (C) Representative images for cells with distinct transcript stages of Star: Stage I (signal confined to nucleus only), Stage II (transcript localized to both nucleus and cytoplasm), and Stage III (transcripts present in cytoplasm only). (Scale bar, 10 μm.) Cyan: Star mRNA, Gray: DAPI nuclear counterstain.

The high resolution of sm-FISH revealed three distinct types of fetal Leydig cells, categorized into transcriptional stages based on the subcellular localization of Star transcripts: 1) Stage I: sm-FISH signal confined to the nucleus, marking active transcription at gene loci and the initiation of Star transcription; 2) Stage II: sm-FISH signal observed in both the nucleus and cytoplasm, indicating active transcription and processed mRNA available for translation; and 3) Stage III: sm-FISH signal detected only in the cytoplasm, suggesting transcription suppression but sustained mRNA availability for translation (Figs. 3C and 4A).

To gain three-dimensional spatial perspective and avoid sampling bias, 5 μm cryosections were collected from the anterior to the posterior poles of each testis (n = 4) at 50 μm intervals. These transverse section images were further segmented into concentric arcs to evaluate fetal Leydig cell distribution along the dorsal–ventral axis (Fig. 4A).

The proportions of each Star-positive cell stage within the interstitial compartment of testes were quantified using sections from E13, E16, and P0 testes (n = 4 each stage). Stage I cells, representing the initiation of transcription, declined sharply from 15% at E13 to 2% at E16, and 0.2% by P0. In contrast, Stage II cells, indicative of active transcription and translation, fluctuated over time, representing 35% at E13, peaking at 54% at E16, and declining to 38% by P0. Stage III cells, characterized by transcriptional suppression but sustained translation access, represented 51% at E13, 44% at E16, and then 62% at P0 (Fig. 4B). Notably, all three stages of fetal Leydig cells were observed within the same microenvironment. Together, these observations suggest that while Star expression is regulated, its transcription stage is not synchronized among neighboring steroidogenic cells of the interstitium.

Fetal Leydig Cell Spatial Distribution Evolves Over Developmental Time.

Next, to investigate the spatial distribution of fetal Leydig cells and their microenvironments, we assembled a three-dimensional spatial distribution using sm-FISH probes for Star to identify cells. The number of Star-positive cells detected within testis interstitium at each transcriptional stage (Stages I, II, and III) were quantified based on their location and plotted as the average cell density per 10,000 μm² per testis (n = 4; Fig. 4 C–T and SI Appendix, Figs. S3 and S4). At E13, fetal Leydig cells, particularly those in Stages II and III, showed a higher density near the middle of the dorsal–ventral (D–V) axis and accumulated toward the anterior and posterior poles (Fig. 4 C and D). By E16 and P0, the spatial distribution of all stages of fetal Leydig cells had shifted, becoming uniformly distributed along both the D–V and anterior–posterior (A–P) axes (Fig. 4 I, J, O, and P and SI Appendix, Figs. S3 and S4). To better visualize these patterns, we generated three-dimensional heatmaps representing the density of fetal Leydig cells across the testis. In each heatmap, the X-axis corresponds to the D–V axis, while the Y-axis represents the A–P axis (Fig. 4 E–H, K–N, and Q–T). Note the heatmap density range is highest at E13 and decreases as the testis grows, corresponding to the expansion of the testis cords and overall testis size over time.

Neighboring Cell Subtypes Express Steroidogenic Pathway Genes.

STAR is considered the gatekeeper for steroidogenesis, with its transcription tightly regulated in steroidogenic cells with rapid on–off transcriptional control (33, 34, 36). This is supported by the observation that Star transcript levels change with lower amplitude compared to other steroidogenic enzymes, as demonstrated in Fig. 2 A and E. In contrast, Cyp17a1 transcripts exhibited profound fluctuations in expression over time (Fig. 2A and SI Appendix, Fig. S2). To investigate whether the transcription and RNA processing of steroidogenic pathway genes are synchronized among steroidogenic pathway genes, we conducted sm-FISH on E13 testis sections using probes for Star and Cyp17a1, the two with the most divergent transcript profiles. sm-FISH resolution allowed us to evaluate the subcellular localization of each probe and assess transcriptional coordination between both genes within individual cells (Fig. 5A).

Fig. 5.

Costaining of Star and Cyp17a1 mRNA probes on E13 testis. (A) Representative costaining image. Arrow: Cyp17a1 not present + Star Stage II. Arrowhead: Cyp17a1 Stage II + Star Stage III. Filled triangle: Cyp17a1 Stage I + Star Stage I. Open triangle: Cyp17a1 Stage II + Star Stage II. (B) Average numbers of cells containing each possible staging result from both probes. (C) Average percentages of cells containing each possible staging result from both probes. Shaded cells: cells with transcripts at the same stage for Star and Cyp17a1. Data were generated from four testes from different embryos from two litters and presented as mean ± SEM counts per testis.

For this analysis, interstitial cells expressing Star and/or Cyp17a1 were identified as fetal Leydig cells. The majority (91.33%) coexpressed both genes, predominantly in Stages II or III (shaded cells in Fig. 5B, 65.54%). As expected, all three stages of Star transcription were observed, but Stage I Cyp17a1 cells were rare. Of note, a small proportion of cells expressed only one of the two genes, with 6.18% expressing Cyp17a1 alone, and 2.49% expressing only Star (Fig. 5 B and C). Thus, sm-FISH exposes transcriptional activity that is not synchronized among fetal Leydig cells for a single gene or among genes within the steroidogenic pathway.

Classic models suggest that steroidogenic cells autonomously express the full complement of enzymes required to convert cholesterol into steroid hormones as steroidogenic pathway genes were used as markers of steroidogenic cells (37–39). To further explore whether cells in fetal testes expressed only subsets of steroidogenic pathway genes, we reanalyzed publicly available single-cell RNA sequencing (scRNA-seq) data from E13.5 testes (10). This analysis included all cell types, avoiding annotation biases when choosing specific cell clusters (Fig. 6A). Cells annotated as fetal Leydig cells expressed a suite of genes, including the set of all four steroidogenic pathway genes: Star, Cyp11a1, Cyp17a1, and Hsd3b1, consistent with their role in androstenedione synthesis (40). The final step, conversion of androstenedione to testosterone, was confirmed to occur in Sertoli cells via HSD17B3 activity (Fig. 6 B–F) (30, 31, 40). Steroidogenic gene expression, however, was not exclusive to fetal Leydig cells as these transcripts were also detected in other cell types including interstitial progenitor, Sertoli, and mesenchymal cells (Fig. 6 B–E and SI Appendix, Figs. S6–S9). In addition, numerous germ cells were observed to express Star according to scRNA-seq; a representative sm-FISH image validates this observation (SI Appendix, Fig. S10). Of note, many cells, including some annotated within the fetal Leydig cell cluster, expressed only a subset of steroidogenic genes (Fig. 6 G and H and SI Appendix, Figs. S5–S9).

Fig. 6.

E13.5 scRNA-seq analysis identified steroidogenic cells expressing an incomplete set of steroidogenic pathway genes (from GSE184708). (A–F) Uniform manifold approximation and projection (UMAP) representations of E13.5 testis cells colored by (A) annotation; (B) Star expression; (C) Cyp17a1 expression; (D) Cyp11a1 expression; (E) Hsd3b1 expression; and (F) Hsd17b3 expression. (G and H) Numbers of cells expressing different combinations of steroidogenic pathway genes in all E13.5 testis cells (G) or within the fetal Leydig cell cluster (H).

Cells that did not express the complete set of steroidogenic pathway genes at E13.5 were hypothesized to be differentiating progenitor cells (SI Appendix, Figs. S6–S9). Our results showed that fetal Leydig cell numbers did not change between E16 and P0 (Fig. 2B), suggesting that most fetal Leydig cells had reached a mature state by E16. To test whether the observed incomplete gene expression profile persisted in cells at this later stage, we analyzed scRNA-Seq data from E16.5 testes from the same publication source (10). Consistent with E13.5 findings, fetal Leydig and other cells expressed subsets of the steroidogenic pathway genes (Fig. 7 and SI Appendix, Figs. S11–S15).

Fig. 7.

E16.5 scRNA-seq analysis identified steroidogenic cells expressing an incomplete set of steroidogenic pathway genes (from GSE184708). (A–F) UMAP representations of E16.5 testis cells colored by (A) annotation; (B) Star expression; (C) Cyp17a1 expression; (D) Cyp11a1 expression; (E) Hsdb1 expression; and (F) Hsd17b3 expression. (G and H) Numbers of cells expressing different combinations of steroidogenic pathway genes in all E16.5 testis cells (G) or within the fetal Leydig cell cluster (H).

To validate scRNA-Seq data and explore the spatial relationship among cells expressing incomplete sets of steroidogenic pathway genes, we performed sm-FISH costaining for Star and Cyp17a1 in E16, P0, and P6 testes. Results show interstitial cells that were positive for Star, Cpy17a1, or both (Fig. 8A). While scRNA-seq data suggest that most of these positive cells are fetal Leydig cells, it is also possible that the sm-FISH probes detect transcripts within other interstitial cell types, such as peritubular myoid cells (Fig. 8B). Together, these observations provide evidence suggesting that fetal androgen production may not be confined to traditional steroidogenic Leydig cells. Instead, our data uncover the possibility of a collaborative model, where androgen biosynthesis is the result of a collective effort of neighboring, closely related interstitial cell types.

Fig. 8.

Star and Cyp17a1 dual sm-FISH show the heterogeneity of steroidogenic pathway gene expression. (A) Representative images of cells that express either Star or Cyp17a1 in the E13, E16, P0, and P6 testis. Filled triangles: cells that are positive only for Star or Cyp17a1. (B) Cells with peritubular myoid cell morphology (arrowhead) exhibit Star and Cyp17a1 expression. (Scale bar, 20 μm.)

Discussion

High-resolution detection methods, including copy number qPCR, scRNA-seq, and sm-FISH, provided detailed insights into the mechanisms and locations of individual cells that together, contribute to fetal testis androgen production over time. This study used these advanced tools to reveal the spatiotemporal expression dynamics of steroidogenic cells in the developing fetal testis, showing that 1) the expression profile of steroidogenic pathway genes results from a combination of cell differentiation and transcriptional regulation; 2) Star expression begins at the anterior and posterior poles and in the center of the testis; 3) steroidogenic pathway gene expression is not synchronized among neighboring steroidogenic cells; and 4) individual steroidogenic cells do not always express the complete set of steroidogenic pathway genes, but together, can provide all necessary components for androgen synthesis. These findings suggest that fetal androgen output arises from the coordinated effort of neighboring steroidogenic interstitial cells with differentiation trajectories similar to fetal Leydig cells.

Efficient Intratesticular Fetal Androgen Production.

This study established a comprehensive temporal profile of androgen synthesis intermediates during fetal and perinatal testis development (E13–P6), expanding on previous research that measured androgens at E14.5, E16.5, E17.5, and E18.5 (37, 41, 42). We observed consistent detection of progesterone and 17-OH progesterone, though their levels were negligible compared to androstenedione and testosterone. The persistent accumulation of androstenedione across all time points suggested a lower conversion rate for the last step of androstenedione to testosterone. Testosterone synthesis is known to require a two-cell pathway in fetal testis as HSD17B3 catalyzes the final conversion in Sertoli cells (30–32, 43). Additional reports indicate that androstenedione can also be metabolized via the 11-keto androgen pathway to 11-ketotestosterone (11-KT). Although this androgen was not included in our evaluation, it has been reported at similar levels as androstenedione and testosterone in E17.5 mouse testes (42). Conversion to 11-KT requires other enzymes, including HSD11B and AKR1C6, which are present in interstitial cells not including fetal Leydig cells. It is not clear how much, or whether, 11-KT contributes to the androgen requirements during fetal development; however, these data support the concept that differentiating cells of potentially divergent origins collaborate within discrete microenvironments to optimize output of different androgens during testis development.

Temporal androstenedione and testosterone profiles revealed a significant increase between E16 and E18, peaking at E18, 2 d after the peak in steroidogenic pathway gene expression. In mammalian cells, the protein translation rate occurs at approximately 10 amino acids (aa) per second (44), with CYP17A1, the longest protein in this pathway, at 507 amino acids, completing the translation and folding in under 3 min. Posttranslational modifications, however, are essential for most steroidogenic conversion factors to become fully functional. For example, phosphorylation is required for STAR to be active (45), and the presence of cytochrome b₅ enhances CYP17A1’s enzymatic activity in converting 17-OH Progesterone into androstenedione (46). Taken together, the delay between gene expression and peak androgen production likely reflects the combined effects of translation and posttranslational modification. Additional studies are underway to examine these events.

Genes Required for Fetal Testicular Androgen Synthesis Are Regulated.

Our copy number RT-qPCR data revealed a significant increase and peak of transcripts, corresponding to the MPW, followed by a rapid decline in expression thereafter. Meanwhile, quantification of Star-expressing cells at E13, E16, and P0 demonstrated cell numbers that increased between E13 and E16 and then remained stable at P0. Despite normalizing transcript copy numbers to cell numbers, relative expression retained a profile showing a dramatic rise and fall of steroidogenic pathway gene expression at these time points. This outcome emphasizes the likelihood that fetal Leydig cell steroidogenesis is actively stimulated and then suppressed during fetal development.

In adult Leydig cells, testosterone production is stimulated by luteinizing hormone (LH) from the pituitary gland. In mice, Lhcgr gene expression initiates at E16.5 (41), coinciding with the peak of steroidogenic pathway gene expression. Furthermore, mice with null mutations in Lhcgr and Lhb have been shown to be fully masculinized at birth (47, 48). Together, these data imply that mouse fetal testosterone synthesis is independent of the LH/LHCGR pathway; however, recent studies opened the door for collaborative stimulation via a common Gα_s-protein coupled receptor pathway. We and others have shown that fetal Leydig cells launch androgen synthesis in response to LH, human chorionic gonadotropin (hCG), corticotropin releasing hormone (CRH), adrenocorticotropin hormone (ACTH), and vasoactive intestinal peptide (VIP) (32, 49–52). Further, recent studies comparing fetal to adult Leydig cell transcripts identified fetal Leydig cell-specific expression of the corticotropin releasing hormone receptor 1 (Crhr1) and melanocortin 2 receptor (Mc2r), along with the Lhcgr, suggesting the potential for redundant regulation via three related hypothalamic–pituitary hormones, LH, ACTH, and CRH (40). Altogether, we conclude that fetal Leydig cell function is regulated at single-cell and organ levels by a combination of local autocrine and/or paracrine factors and hormones sourced from the developing hypothalamic–pituitary axis.

Star 3D Spatial Expression Profiles Highlight the Two Waves of Fetal Leydig Cell Differentiation.

Fetal Leydig cell progenitors originate from the coelomic epithelium (dorsal) and the mesonephric mesenchyme (ventral) (22–24, 53–56), undergoing two distinct waves of differentiation. The first wave includes cells responding to Sertoli cell-sourced DHH to undergo immediate differentiation between E11.5-12.5 (26, 57, 58). The second wave of differentiation includes other cells within a perivascular niche that retain progenitor characteristics until the release of the NOTCH signal after E13.5 (22) (reviewed in ref. 25). Our initial observations at E13 may capture events that span the time between the two waves of differentiation.

The spatial context for cell differentiation is instrumental in the development of testis morphology. Sertoli cells have been shown to differentiate in a pattern radiating from the central testis starting around E11 (24, 59). Once Sertoli cell identity has been established, they release morphogens, including DHH, that stimulate interstitial cell differentiation and other factors that promote massive cell migration from the ventral, gonad–mesonephros border toward the dorsal testis surface. These include endothelial and other cell types that eventually create the perivascular niche with NOTCH ligands restraining fetal Leydig cells until the second wave of differentiation after E13.5 (22).

Previous studies have also shown that somatic cells mature as they approach the center of the gonad (24, 56); therefore, our observation that Star-expressing cells are denser in the central region of the D–V axis supports the hypothesis that fetal Leydig cells mature as they migrate toward the center of the testis. Further, testes at early developmental stages are an elongated oval shape with the near-pole regions possessing smaller volumes relative to the middle region. Thus, one possible explanation of Star-positive cells accumulating near the A–P poles could be that they correspond to morphological changes and remodeling of the vasculature being completed first within the smaller volume areas, therefore allowing steroidogenic cell maturation to emerge sooner near the poles (60). The proliferation of endothelial cells in the center of the testis during vascular remodeling may also inhibit the differentiation of fetal Leydig cell progenitors. Further analyses must be completed to test these hypotheses.

Individual Cell Steroidogenic Pathway Gene Expression Is Regulated but Not Synchronized.

The steroidogenic pathway genes Star, Cyp11a1, Cyp17a1, and Hsd3b1 constitute the complete set required for androstenedione synthesis in interstitial steroidogenic cells. We chose to examine Star and Cyp17a1 via smFISH because not only is Star considered the gatekeeper of steroidogenesis (33, 34), but also, when comparing expression profiles of all genes together, they represent the low and high end of transcript accumulation, respectively.

Star transcript levels have been shown to be strictly regulated at transcription initiation, elongation, splicing, and posttranscription steps in MA-10 immature Leydig and Y-1 adrenal mouse cell lines (36, 61, 62). Under basal conditions, Star transcription in MA-10 cells is completely quiescent while Y-1 cells are constitutively active at a low level. High-resolution sm-FISH was used to evaluate the kinetics and subcellular localization of both primary (pRNA) and spliced (spRNA or mRNA) Star transcripts in both MA-10 and Y-1 cells before and immediately following Br-cAMP treatment. Similar to findings reported herein, the amount of Star transcripts present in nuclei and cytoplasm were not synchronized within Y-1 cells under basal conditions (36, 63) Upon addition of Br-cAMP, pRNA transcripts show acute stimulation but with asynchronous activation at gene loci with MA-10 cell activity slower than that of Y-1 cells. Timed imaging using both cell types documented p- and spRNA appearance within minutes first at one locus followed by activation of the second locus after several minutes; both loci harbor primary and spliced RNA within 60 min. In general, sp- or mRNA spreads from the nuclear envelope by 90 min and remains abundant within the cytoplasm up to at least 180 min post stimulation (36, 63). This brief interval between nuclear and cytoplasmic localization of transcripts explains the challenge in capturing Stage I fetal Leydig cells, making them the least prevalent stage documented in our study.

There is a significant decrease in nuclear pRNA detection among cells between 60 and 180 min representing cessation of transcription initiation at Star loci; spRNA transcripts exhibit a similar but delayed trend (63–65). Star-positive cells in fetal testes were also observed with transcripts exclusively located in the cytoplasm, indicating transcriptional suppression and labeled as Stage III. Initially, we hypothesized that Stage III represented the conclusion of transcription; however, we were surprised to find the predominance of Star-positive cells in the earliest stages of androgen synthesis (E13 testis) were already in Stage III (51%). By E16, however, Stage II cells became the most abundant (54%). Taken together, we propose the following possible explanations for changes in transcript processing over time: 1) Stage III cells may resume transcription and revert to Stage II, with cells able to adjust transcriptional status depending on local cues; 2) newly differentiated steroidogenic cells emerging post-E13 contribute to the Stage II population; or 3) a combination of these scenarios.

Additional studies colocalized Star mRNA and STAR protein with mitochondria suggesting direct targeting of transcripts for translation with the intent of rapid protein activity at the site of action (63). Of note, the increases of Star mRNA transcripts detected by sm-FISH lag appreciably behind previous reports of RNA increases using Northern blots and with STAR protein, suggesting the most readily translatable RNA is protected from sm-FISH probes by ribonucleoprotein or polyribosomal complexes (62, 63, 66). Thus, it is important to consider multilevel regulation for transcription, transcript access for translation, and posttranslational modifications in controlling the rise and fall of androgen production within an intricately defined window of developmental time.

sm-FISH and scRNA-seq.

To assess whether asynchronous transcript processing was unique to Star, we also performed sm-FISH for Cyp17a1 mRNA. Our investigations showed that both Star and Cyp17a1 transcription activities are asynchronous in the endogenous state. In addition, sm-FISH analysis uncovered interstitial cells that expressed only one of the two factors. To gain perspective and verify whether individual cells might harbor incomplete sets of steroidogenic pathway genes that contribute to steroid metabolism, we analyzed scRNA-seq data from E13.5 and E16.5 fetal testes (10). Our analyses focused on detecting steroidogenic pathway genes among all cells and within those that were annotated as fetal Leydig cells (10). As expected, scRNA-seq exposed significant heterogeneity in gene expression among cells, with many expressing incomplete complements of the steroidogenic pathway genes. Steroidogenic pathway genes were identified in cells annotated as fetal Leydig cells, but also in many other cell types. Even among fetal Leydig cells, most expressed the complete set of genes, but many did not. It must be noted, however, that scRNA seq does have limitations. With respect to detecting subsets or individual steroidogenic pathway genes within cells, the phenomenon of technical dropout must be considered. In this case, the transcript profile of individual cells may not be accurately represented due to stochastic sampling during the library preparation process or insufficient amplification of low-abundance transcripts, causing a failure to detect genes. Thus, one possibility for the detection of incomplete steroidogenic pathway gene sets is that one or more genes dropped out of the analysis. While this is undoubtedly the case in many cells, individual steroidogenic pathway genes and combinations of genes were detected within similar cell types, including fetal Leydig cells at both E13.5 and E16.5; therefore, we took advantage of the high resolution of sm-FISH to validate these data.

The combined outcomes from scRNA-seq and sm-FISH elevate confidence in our observations that individual cells harbor varying complements of steroidogenic pathway genes. sm-FISH images provide understanding of the microenvironments that contribute to total testicular output. In both scRNA-seq and sm-FISH techniques, transcripts were detected in many cell types with sm-FISH tagging potential cells besides fetal Leydig cells, including interstitial progenitor cells, peritubular myoid cells, pericytes, germ cells, and others in pre- and postnatal testes. In support, previous reports with immunostaining images highlight the presence of steroidogenic pathway factors in cells other than fetal Leydig cells based on location and shape (32, 38, 67).

In response to DHH and the attenuation of NOTCH, steroidogenic progenitors differentiate into fetal Leydig cells and other cell types, such as peritubular myoid cells (21, 22). Costaining of Star and Cyp17a1 revealed that cells deficient in either gene were in close proximity to other steroidogenic cells. Notably, as observed via transmission electron microscopy, fetal Leydig cells are clustered within a basement membrane (5). The transport of lipophilic steroid intermediates between cells in the interstitium, including fetal Leydig cells, likely facilitates the production of androstenedione, enabling cells missing some steroidogenic genes to still contribute to steroidogenesis. Although not examined here, this includes the possibility for conversion to other steroids including 11-KT or other corticosteroids (42). In the fetal testis, it is established that testosterone synthesis requires the cooperation of both fetal Leydig cells and Sertoli cells where Hsd17b3 is expressed (30–32). The absence of the blood-testis barrier in the fetal testis (68) facilitates Sertoli cell conversion of androstenedione to testosterone, although with some lag suggesting that the increased distance into the testis cord is relevant.

Summary.

In summary, utilizing fetal steroidogenic cells as an example, we have demonstrated that cells within aggregates are capable of functioning independently. These cells cluster together to perform biological functions, these studies suggest this likely includes the production of androgens. The cells that surround this aggregate create a niche that regulates the activity of the cluster. Collectively, these niches enable the entire testis to secrete optimal amounts of testosterone necessary for the differentiation and development of male reproductive organs.

Materials and Methods

Animals.

C57BL6/J (Strain Number: 000664; Jackson Laboratory, Bar Harbor, MD) mice were set up for timed matings and animals were collected with testes harvested at the indicated timepoints. Animals were euthanized by CO₂ asphyxiation followed by cervical dislocation. Animal housing and all described procedures were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Wisconsin–Madison and were performed in accordance with NIH Guiding Principles for the Care and Use of Laboratory Animals (https://policy.wisc.edu/library/UW-4090).

Steroid LC/MS/MS.

Testes were collected and then analyzed by liquid chromatography–tandem mass spectrometry (LC–MS/MS) using a QTRAP 6500+ triple quadrupole (Sciex) adapted from a previously published method (69).

sm-FISH.

Testes were harvested, fixed, processed through sucrose gradients, and embedded in O.C.T. for sectioning. 5 μm transverse cryosections were harvested every 50 μm (every 10 sections) from the anterior to the posterior of each testis using Leica CM3050 S Cryostat. Sections were then incubated with Star and/or Cyp17a1 mRNA sm-FISH probes (Biosearch Technologies, SI Appendix, Table S4) in hybridization buffer containing 10% deionized formamide at 37 °C overnight. Images were captured using a Nikon Yokogawa W1 CSU Spinning Disk Confocal Microscope and a Hamamatsu Quest camera at 100× magnification and 0.9 μm Z series.

Bioinformatics.

Normalized single-cell RNA-seq (scRNA-seq) counts were obtained from GSE184708, a publicly available dataset from Mayère et al. (10). The datasets were filtered to remove XX and germ line cells. Only the time points E13.5 and E16.5 were used. Dotplots were generated with the R package Seurat v. 5.0.1. Cells were classified based on which steroidogenic pathway genes they expressed (Star, Cyp11a1, Cyp17a1, Hsd3b1). For example, a cell was placed in the “Star only” category if it expressed Star (>1 count) and also did not express Cyp11a1, Cyp17a1, and Hsd3b1; a cell would be placed in the Star and Cyp11a1 category if it expressed Star and Cyp11a1 and did not express Cyp17a1 and Hsd3b1, and so forth. The categories are mutually exclusive such that a cell can only belong to a single category. Please see SI Appendix for further details.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix. Previously published data were used for this work (10).