Multiomics approach to profiling Sertoli cell maturation during development of the spermatogonial stem cell niche

Abstract

Spermatogonial stem cells (SSCs) are the basis of spermatogenesis, a complex process supported by a specialized microenvironment, called the SSC niche. Postnatal development of SSCs is characterized by distinct metabolic transitions from prepubertal to adult stages. An understanding of the niche factors that regulate these maturational events is critical for the clinical application of SSCs in fertility preservation. To investigate the niche maturation events that take place during SSC maturation, we combined different ‘-omics’ technologies. Serial single cell RNA sequencing analysis revealed changes in the transcriptomes indicative of niche maturation that was initiated at 11 years of age in humans and at 8 weeks of age in pigs, as evident by Monocle analysis of Sertoli cells and peritubular myoid cell (PMC) development in humans and Sertoli cell analysis in pigs. Morphological niche maturation was associated with lipid droplet accumulation, a characteristic that was conserved between species. Lipidomic profiling revealed an increase in triglycerides and a decrease in sphingolipids with Sertoli cell maturation in the pig model. Quantitative (phospho-) proteomics analysis detected the activation of distinct pathways with porcine Sertoli cell maturation. We show here that the main aspects of niche maturation coincide with the morphological maturation of SSCs, which is followed by their metabolic maturation. The main aspects are also conserved between the species and can be predicted by changes in the niche lipidome. Overall, this knowledge is pivotal to establishing cell/tissue-based biomarkers that could gauge stem cell maturation to facilitate laboratory techniques that allow for SSC transplantation for restoration of fertility.

Introduction

Lifelong male fertility is maintained by the spermatogonial stem cell (SSC) which is the basis for spermatogenesis. SSCs reside in a niche microenvironment at the base of the seminiferous tubules within the testis (Oatley and Brinster, 2012). High spatial and temporal control of SSC maintenance and differentiation by the niche is required to maintain spermatogenesis. Key players in this process are Sertoli cells and peritubular myoid cells (PMCs) (Hess and De Franca, 2008; Chen et al., 2014; Guo et al., 2020).

SSCs offer tremendous potential for the treatment of infertility. The bio-banking of testicular tissue from prepubertal boys prior to gonadotoxic cancer therapy may allow restoration of fertility later in life (Goossens et al., 2020). We previously reported that immature SSC precursors have a distinct metabolic phenotype compared to adult SSCs (Voigt et al., 2021, 2022). In humans, the majority of these immature SSC precursors migrate from the center of the testicular cords toward their adult location at the base of the seminiferous cords within the first year of life (Guo et al., 2018, 2020; Zhao et al., 2020); however, the cells do not acquire an adult SSC morphology until 11 years of age (Voigt et al., 2022). This morphological maturation of SSCs is followed by metabolic transition and downregulation of oxidative phosphorylation (OXPHOS)-associated genes (Voigt et al., 2022). Hence, during the prepubertal phase in humans, SSCs are functionally immature (Voigt et al., 2022). Preserved testicular tissue with an extremely limited number of SSCs may have an immature SSC pool, necessitating the development of biomarkers for SSC maturation. Identification of biomarkers will help to inform in vitro maturation strategies that can be targeted to individual developmental states. Therefore, we aimed to address how these transitions in SSC morphology and metabolism are associated with, and potentially regulated by, the development of surrounding niche cells, mainly focusing on Sertoli cells.

Developing SSCs are exposed to a dynamically evolving microenvironment. Sertoli cells differentiate from a common somatic cell progenitor pool around 7 weeks of age in humans to an immature epithelial cell type forming the seminiferous cords, which are subsequently colonized by the immature SSC precursors (Guo et al., 2021). In these cords, unpolarized and proliferative Sertoli cells form a simple niche environment, surrounded by an undifferentiated interstitium (Guo et al., 2020). Immature Sertoli cells do not yet control paracellular diffusion in the seminiferous tubule and are ultra-structurally and physiologically distinct from their adult counterparts (Vergouwen et al., 1991; Franca et al., 2000; Sharpe et al., 2003; Guo et al., 2020; Zhao et al., 2020). Adult Sertoli cells are a hyperpolarized epithelial cell type that compartmentalizes the seminiferous epithelium into basal and adluminal compartments by belt-like tight junctions (TJs) between adjacent cells as part of the blood–testis barrier (BTB) (Gilula et al., 1976; Russell, 1978; França et al., 2012). In humans, a thick basal lamina and several layers of PMCs are characteristic of the tubular appearance after puberty. The immature niche stands in stark contrast to the compartmentalized seminiferous tubules in adulthood, but the exact order of events during maturation is still uncharacterized.

Here, we investigated niche maturation to map which specific maturational changes are associated with the maturation of SSCs. We found that changes in the niche lipidome are associated with niche maturation at 8 weeks in pigs and 11 years of age in humans, coinciding with morphological changes in SSCs, that are followed by SSC metabolic transitions. This knowledge advances the understanding of the development of the SSC niche and identifies potential regulatory mechanisms that determine SSC maturation.

Materials and methods

Ethics statement

Human samples were processed with approval from the University of Pittsburgh Institutional Review Board (Protocol #STUDY09020220).

Experimental model and sample details

Pig (Sus scrofa) testes were obtained by castration from 1- and 8-week-old (wko) pigs from Sunterra Farms Ltd, Acme and the University of Alberta, respectively, AB, Canada. All animal experiments were carried out in compliance with the Canadian Council of Animal Care guidelines and approved by the University of Calgary’s Health Sciences Animal Care Committee (AC21-0017).

Human testis samples were obtained from 1-, 2-, and 7-year-old males, with informed consent from the parents, through the Fertility Preservation Program University of Pittsburgh Medical Center, Pittsburgh, PA, USA (Supplementary Table SI).

Seminiferous tubules and cell isolation

Cells were isolated from testes by a two-step enzymatic digestion, as previously described (Sakib et al., 2019). Sertoli cells were enriched using 1.5 h plating at 37°C with advanced MEM (Thermo Fisher Scientific, #12492013, Mississauga, ON, Canada), 5% FBS (Thermo Fisher Scientific #12483020), 1× P/S (Sigma, # P4333, Oakville, ON, Canada), followed by a washing step and overnight culture at 37°C. Sertoli cells were enriched to 91.1 ± 0.6 and 96.96 ± 1.5% Gata-4⁺ cells for 1- and 8-week-old Sertoli cells, respectively. For the seminiferous tubule isolation, the enzymatic reaction was stopped after collagenase IV treatment (Sakib et al., 2019).

Histology, immunohistochemistry, and BODIPY staining

Immunohistochemistry (IHC) was performed as previously described (Voigt et al., 2021). For lipid visualization, samples were fixed in 4% paraformaldehyde for 15 min at room temperature (RT) and subsequently frozen with liquid nitrogen and embedded in O.C.T. (Tissue-Tek^® O.C.T. Compound, Sakura Finetek USA, Inc., #4583, Torrance, CA, USA). Cryosections of 12 μm were cut and sections were permeabilized with 0.2% Triton X-100 in PBS for 10 min and blocked with CAS-Block for 15 min (Thermo Fisher Scientific, #008120). BODIPY dye (Thermo Fisher Scientific, #D3922) was diluted 1:1000 with antibody in a 10% CAS-Block in PBS solution and added to sections overnight (Supplementary Table SI).

Proteomics analysis and high-performance liquid chromatography and mass spectrometry

Sertoli cells (5 × 10⁶) from 1- and 8-week-old pig testes were washed twice with cold PBS. Cell preparation was performed as previously described (Voigt et al., 2022) and liquid chromatography and mass spectrometry (LC-MS) analysis was performed at the Southern Alberta Mass Spectrometry (SAMS) core facility (University of Calgary, Alberta, Canada). Phospho-proteomics analysis was performed by sequential enrichment using metal oxide affinity chromatography (SOMAC) following the manufacturer’s instructions (Thermo Fisher Scientific, Application note 65381).

Lipidomics

Lipidomics analysis on isolated seminiferous tubules was performed through the Metabolomics Innovation Centre (TMIC, University of Alberta, Edmonton, Canada).

Sample preparation: lipids extraction

Lipids were extracted following a modified Folch liquid-liquid extraction protocol (Buzatto et al., 2021; Zardini Buzatto et al., 2020). Aliquots of each sample were first evaporated to dryness with a nitrogen blowdown evaporator for 20 min at 25°C. The mass of dried samples was measured (0.91–5.47 mg) and used as a normalization factor for all reagents. Samples were homogenized with an internal standard mixture composed of 15 deuterated lipids and methanol (33.3 µl/mg of dried sample) using a bead homogenizer. The obtained mixture was extracted with dichloromethane (66.7 µl/mg of dried sample). A clean-up step was performed with water (24.0 µl/mg of sample). Samples were equilibrated at RT for 10 min and subjected to centrifugation at 16 000×g for 10 min at 4°C. An aliquot of the organic layer was evaporated to dryness with a nitrogen blowdown evaporator. The residue was immediately resuspended in mobile phase B (10 mM NH₄COOH in 95% 2-propanol/water), vortexed for 1 min, and diluted with mobile phase A (10 mM NH₄COOH in 50:40:10 methanol/acetonitrile/water) to 80.0 µl/mg of dried sample.

A pooled mixture composed of one aliquot of each sample was prepared for quality control (QC). The QC pooled mixture was divided into multiple vials and evaporated to dryness with a nitrogen blowdown evaporator for 20 min at 25°C. Six QC aliquots were extracted with the samples. Each QC extraction replicate was injected in triplicates, totaling 18 QC injections (i.e. one injection from each QC extraction replicate before, one in between, and one after the sample extracts).

LC–MS analysis conditions

The LC–MS analyses were performed in both positive and negative ionization (Buzatto et al., 2021; Zardini Buzatto et al., 2020) using a Thermo Vanquish UHPLC linked to Bruker Impact II QTOF Mass Spectrometer (m/z 150–1500) with a Waters Acquity CSH C18 column. A 26.4 min gradient separation was performed with mobile phase A (10 mM NH₄COOH in 50:40:10 methanol/acetonitrile/water) and mobile phase B (10 mM NH₄COOH in 95% 2-propanol/water) at a flowrate of 250 µl/min and 42°C. Sample extracts were injected between injection replicates of the QC pooled mixture. A total of 32 sample injections (extraction and injection duplicates of eight samples) and 18 QC injections (six aliquots of the pooled mixture injected in triplicates) were performed in each ionization polarity. MS/MS spectra were acquired for all samples for identification.

Data processing

LC–MS data from 50 injections (32 sample injections and 18 QC injections) were independently processed in positive and negative ionization. Lipid features were extracted and aligned using NovaMT LipidScreener (Nova Medical Testing Inc., Edmonton, Canada) with a retention time tolerance of 15 s and an m/z tolerance of 5.0 mDa. Only features detected in at least 80% of injections in any group (1wko, 8wko, or QC) were kept. The data acquired in positive and negative ionization from each sample experimental replicate were combined, i.e. the detected features from all samples were merged into one feature-intensity table. Missing values were substituted by: (i) the median intensity of the sample group for features detected in at least 75% of injections within the group (1wko, 8wko, or QC); (ii) the minimum intensity within the group for features detected in at least 50% of injections; or (iii) the global minimum for all sample and QC injections for features detected in <50% of injections within the group.

Lipid identification

A three-tier identification approach based on MS/MS spectral similarity, retention time and accurate mass match, was employed for lipid identification. For Tier 1, we employed MS/MS spectral similarity match with MS/MS match score ≥100, precursor m/z error ≤5.0 mDa, and a retention time filter based on lipid subclasses and fatty acyl group. For Tier 2, we employed a retention time match with error ≤30 s and precursor m/z error ≤10 ppm. For Tier 3, we performed a mass match with m/z error ≤5.0 mDa and 20 ppm. After Tier 3 identification, a six-tier filtering and scoring approach embedded in NovaMT LipidScreener was employed to restrict the number of matches and select the best identification option to determine the lipid sub-classes for normalization.

All compounds identified in tiers 1, 2, and 3 were combined for normalization and statistical analysis. Unidentified features were not employed for statistics.

Data normalization

Data normalization was performed using a set of 15 deuterated internal standards belonging to different lipid classes. The positively and putatively identified lipids were matched to one of the 15 internal standards according to lipid class similarity and expected retention time range for each class. Intensity ratios, i.e. intensity of each lipid divided by intensity of the matched internal standard, were calculated for normalization.

Single-cell RNA-sequencing

Isolated primary human testicular cells were processed using Chromium TM Single Cell 3′ v2 Chemistry library prep kit according to the manufacturer’s instructions (10× Genomics, San Francisco, CA, USA). cDNA libraries were prepared by the University of Pittsburgh HSCRF (Health Sciences Core Research Facilities) Genomics Research Core using the Chromium Single cell 3′ v3.1 library kit and QC was performed using an Agilent high sensitivity DNA kit. The cDNA libraries were sequenced with a targeted sequencing depth of ∼50 000 reads/cell using an Illumina NovaSeq 6000 sequencer (100 pb single end reads) (CHGI, University of Calgary, Alberta, Canada). The processing of raw reads was subsequently performed with a 10× Genomics CellRanger pipeline (version 3.1.0) using default settings. The outputs from the human samples were combined using the CellRanger aggr function to normalize for the sequencing depth and produce aggregated gene-cell UMI count matrix for the downstream analysis.

Quality control, pre-processing of individual matrices, and integration

Publicly available datasets spanning different groups of age were collected from the gene expression omnibus (GEO) of the National Center for Biotechnology Information database: GSE196819 (Voigt et al., 2022) for prepubertal samples, GSE134144 (Guo et al., 2020) for pre- and peripubertal samples, GSE120508 (Guo et al., 2018) for postpubertal samples, GSE186479 (Zhang et al., 2022) for porcine samples (7, 30, 60, and 90 days of age) and GSE174782 (Zhang et al., 2021) for porcine sample (150 days of age). Similar processing flow was applied to human and pig datasets. The filtered gene-cell barcode matrices were imported into Seurat (Hao et al., 2021) and subjected to a pre-filtration step to remove low quality cells using default settings. A pre-processing workflow was applied to individual samples to select and filter high quality cells based on the QC metrics (genes detected, UMI counts and mitochondrial genes). Only cells expressing >200 genes and with <25% reads mapped to mitochondrial genes were retained. To enable the comparison across different datasets, Seurat V4 integration workflow was used to merge and correct for technical batch effects using the ‘IntegrateData’ function. A new integrated Seurat matrix was generated and processed as follows. The ‘ElbowPlot’ function was used to identify the dimensionality of every dataset, and the first 15 principal components (PCA) were selected to run the ‘FindCellsNeighbors’, ‘FindClusters’ and ‘RunUMAP’ functions with a resolution of 0.2. The ‘FindMarkers’ function was applied to identify differentially expressed genes (DEGs) between clusters, with P values lower than 0.01 and fold change of 1 or higher. The ‘FeaturePlot’ function was used to visualize genes of interest. Cell clusters representative of Sertoli cells and PMCs were extracted from the integrated Seurat object using the ‘Subset’ function and imported to Monocle for further analysis.

Cell trajectory analysis

Single cell pseudotime and trajectory were constructed using the Monocle 2 package (version V2.20.0). The raw count matrix of Sertoli cells were processed according to the Monocle package vignette. The DDRTree method was used for dimensionality reduction and the ‘orderCells’ function was implemented to order cells in pseudotime. Cells were colored based on their state or age group. Subsequently, the ‘BEAM’ function was used to identify the genes with branch-dependent expression and analyze the fate decision of cells across trajectory branches. The ‘plot_genes_branched_heatmap’ function was used to generate a heatmap, with a q-value cutoff <1e−5, that displays the genes expression pattern along the trajectory path. The ‘plot_genes_branched_pseudotime’ function was used to plot genes that are expressed in a branch-dependent manner.

GO term enrichment and functional analysis

Functional enrichment analysis of DEGs along the trajectory was performed using ShinyGO (version 0.75), a web gene-set enrichment tool. GO for biological processes, cellular components, and molecular functions that are significantly enriched with P-value cutoff (false discovery rate (FDR)) <0.05 were selected. Pathways enrichment analysis between gene set clusters was performed using Metascape (Zhou et al., 2019) (https://metascape.org) (Supplementary Table SI).

qRT-PCR, western blot, transmission electron microscopy, and seahorse flux analyzer

PCR, western blot, transmission electron microscopy (TEM), and extracellular acidification rate with the Seahorse Flux Analyzer were performed as previously described (Voigt et al., 2021).

Quantification and statistical analysis

For all experiments, a minimum of four independent biological replicates were used. Results are stated in mean ± SEM. An unpaired parametric Student’s t-test and a non-parametric t-test (Mann–Whitney test) were used for normally distributed and not-normally distributed data, respectively. Welch’s correction was used for not equal standard deviation between samples. A P-value of P ≤ 0.05 was considered statistically significant. GraphPad Prism 8.0 software was used for all statistical analyses. For Proteomics and phosphoproteomics significant outlier cutoff values were determined using the BoxPlotR tool (Spitzer et al., 2014) after log(2) transformation by a bloxplot-and-whiskers analysis. Protein–protein interactions relationship was encoded into networks in the STRING v11 database (https://string-db.org) and data were analyzed with an FDR of P ≤ 0.05 using sus scrofa as the organism. For lipidomics, statistical analysis was performed with MetaboAnalyst 5.0 (https://www.metaboanalyst.ca/). The file with identified and normalized features was uploaded to the statistical analysis function. Non-informative features (i.e. internal standards and features with near-constant values between the groups determined by low RSD) and features with low repeatability (RSD >30% for QC samples) were filtered out. The dataset was further normalized by auto-scaling and summed (total) intensity. Finally, the sum-normalized and auto-scaled intensity ratios were used for statistical analysis.

Results

Serial transcriptomic analysis of human Sertoli cell maturation

We previously performed single-cell RNA sequencing (scRNA-Seq) on 1-, 2-, and 7-year-old testicular cells (Voigt et al., 2022), and combined our dataset with publicly available datasets (Guo et al., 2018, 2020). We defined that morphological maturation occurs at 11 years of age during human prepubertal SSC development, that is followed by a metabolic transition. In the present study, we combined datasets covering prepubertal (1, 2, 7 years) (Voigt et al., 2022) and peri/postpubertal ages (Guo et al., 2018, 2020) to map a serial transcriptomic analysis of Sertoli cell maturation to the changes observed in spermatogonia, including SSCs. We carried out the clustering analysis on 39171 cells (Supplementary Fig. S1A and Fig. 1A), which consisted of germ cells (DDX4+), Sertoli cells (AMH+, SOX9+), PMCs (ACTA2+), macrophages (CD14+), and endothelial cells (PECAM1+) (Supplementary Fig. S1A and B). The markers DLK1 and NR2F2 characterized Leydig cells and somatic cell precursors, respectively (Supplementary Fig. S1A and B). Early signs of germ cell differentiation occurred at 11 years, while no clear change in macrophages or endothelial cell distribution across ages was observed (Fig. 1A).

Focused clustering on Sertoli cells led to the identification of seven clusters (Supplementary Fig. S1C and C′). These cell clusters can be merged into two large cell clusters (1 and 2) connected by one small cluster (Cluster 3) (Fig. 1B), supporting previous observations (Guo et al., 2020). The contribution of each sample to these cell clusters is shown in Fig. 1B′. Using Monocle, we constructed the developmental trajectory of Sertoli cells and identified three distinct states that these cells exhibit during their development (Fig. 1C). Our data aligned with previous findings referring to these states as immature 1, immature 2, and mature state (Guo et al., 2020) or Stage a, Stage b, and Stage c, respectively (Zhao et al., 2020). All Sertoli cells from 1-, 2-, and 7-year-old males cluster within States 1 and 2, while those from 11-year-old and older males were identified as maturing or fully mature (State 3) (Fig. 1C′ and Supplementary Fig. S1D) (Zhao et al., 2020). Supplementary Fig. S1E illustrates the contribution of the seven clusters described in Supplementary Fig. S1C and C′ to States 1, 2, and 3. We next analyzed the expression of genes that significantly changed as these cells transit, through branch point 1, from State 1 toward State 2 or 3 (Fig. 1C). We identified four distinct gene clusters according to their expression pattern (Supplementary Fig. S2A).

A comparative analysis between the four gene sets (Supplementary Fig. S2A) using Metascape as illustrated in Supplementary Fig. S2B, revealed that Clusters 1 and 3 gene sets that were upregulated toward State 3 (maturing Sertoli cells), were enriched in genes related to OXPHOS, fatty acid metabolism, cholesterol homeostasis, estrogen response, and glycolysis. The Cluster 2 gene set, which was enriched in androgen response, mitotic spindle and E2F targets, was upregulated in State 1 compared to State 2 (States 1 and 2 immature Sertoli cell states), suggesting a highly proliferative state in State 1.

From the pre branch (State 1) toward State 3 (maturing Sertoli cells), upregulation of two subsequent gene sets, namely Clusters 1 and 3, was observed. These gene sets were enriched with the GO terms of spermatogenesis and cytoskeleton (Cluster 1) and mitochondrial activity (Cluster 3) (Supplementary Fig. S2C). Cells moving from State 1 (pre-branch) to State 2 displayed a drop in the gene set Cluster 2 expression, which was mainly enriched in GO terms related to focal adhesion and actin and cadherin binding (Supplementary Fig. S2C), suggesting that cells in States 1 and 2 may have different migratory and proliferative properties, as discussed above.

As we observed a heterogeneity in cell distribution within State 3 along the pseudotime trajectory (Fig. 1C′, blue cluster) (when using the label ‘age’), we sought to analyze the dynamic gene expression changes that occurred during puberty. We performed pseudotime trajectory analysis on Sertoli cells from 11-, 13-, 14-year-old males and adults, and identified four distinct states (Fig. 1D). The contribution of the individual samples to these states is shown in Fig. 1D′. We next analyzed the expression of genes that significantly changed as cells transit, through branch point 2, from State 1 to State 3 or 4 (Fig. 2A). As shown in Supplementary Fig. S2D, GO enrichment analysis of the Cluster 2 gene set showed enriched GO terms of focal adhesion and anchoring junction, while the Cluster 3 gene set was enriched for GO terms of growth factor binding and extracellular matrix, which is consistent with tissue morphogenesis during the peripubertal/pubertal phase. Furthermore, comparing the expression of gene sets using Metascape (Fig. 2B) showed an enrichment of genes related to several pathways such as fatty acid metabolism, cholesterol homeostasis, estrogen response, cytokine signaling (i.e. IL2_STAT5_signaling, Interferon gamma response), apical junction, OXPHOS and hypoxia. Both Clusters 2 and 3 gene sets were upregulated in the pre-branch (State 1; composed of 11- and 13-year-old samples) and remained toward State 3 (mainly composed of 14-year-old samples). Changes in cell structure, associated with enhanced cell metabolism and immune response observed in State 1 (composed of 11- and 13-year-old samples) point toward the initiation of Sertoli cell maturation at 11 years of age. These drastic changes at 11 years coincide with the morphological maturation of SSCs and therefore seem to indicate a critical point in niche maturation, despite Sertoli cells continuing to develop beyond 14 years of age.

Developmental trajectory of PMC development

UMAP visualization showed that PMCs emerge as a single cluster at 11 years, expanding from a combined Leydig cell and somatic cell precursors cluster (Fig. 1A), which was consistent with previous reports (Guo et al., 2020). Focused clustering of PMCs identified six clusters (Fig. 2C) and constructing the developmental trajectory using Monocle revealed eight individual states (Fig. 2D). Cells from 1-, 2- and 7-year-old males clustered at the beginning to the middle of the pseudotime trajectory, while those from 11-year-old and older males were positioned at the middle toward the end (Fig. 2D).

To define the molecular events involved with PMC maturation, we analyzed the expression of genes that were significantly altered along the pseudotime trajectory (Fig. 2E) and performed GO enrichment analysis (Supplementary Fig. S2E). The upregulation of genes related to cell–cell junction, cell migration and ATP-transport related gene sets (Cluster 1) and those associated with OXPHOS and fatty acid metabolism (Cluster 4) (Supplementary Fig. S2E) were observed at the beginning and toward the middle of the pseudotime, aligning with the position of prepubertal cells (1-, 2-, 7-year-old) (Fig. 2D). The Cluster 3 gene set was enriched for, among others, GO terms related to collagen-containing extracellular matrix, focal adhesion, and secreted vesicles. This gene set cluster was upregulated in the middle toward the end of the pseudotime (maturing PMCs), aligning with the position of peripubertal/postpubertal samples (11 years and older). Pathway analysis using Metascape showed an enrichment of genes related to cell metabolism (i.e. OXPHOS and lipid metabolism), hormone response, and stress response (i.e. DNA repair) (Clusters 1 and 4; from start to the middle of pseudotime) (Fig. 2F). Cluster 3, which was upregulated from the middle toward the end of pseudotime (11 years and older), was associated with, among others, inflammatory response and cytokine response (IL6-JAK-STAT3, Interferons) (Fig. 2F), pointing toward a development of local immunity around puberty.

A significant increase in androgen response around this age was observed in PMCs, which could play a role for indirect modulation of Sertoli cell function and germ cell development. All these coincide with the change in SSC morphology that precedes metabolic maturation at 11 years of age (Voigt et al., 2022).

Sertoli cell maturation in humans is accompanied by lipid droplet accumulation

We next investigated ultrastructural changes during human testicular development (Fig. 3). In males up to 5 years old, Sertoli cells appeared randomly distributed within the seminiferous cords, surrounding large circular putative SSCs (Fig. 3A–D). The Sertoli cell cytoplasm was small, whereas the nuclei were mostly round to oval and heterochromatic (Fig. 3, white circled nuclei). The basement membrane (BM), including PMCs, was generally under 5 $µ$m thick in males under 5 years of age, with a slight increase by 7 years of age (Fig. 3F). Similarly, we observed small seminiferous cords under 50 $µ$m in diameter until 8 years of age (Supplementary Fig. S3A–D). However, from 7 to 10 years of age onwards, the tissue appeared heterogenous in TEM, with some immature looking Sertoli cells (Fig. 3E and H) and some more organized and euchromatic nuclei with clefts (Fig. 3F and G), confirming an asynchronicity in the maturation of Sertoli cells similar to what has been described with scRNA-Seq analysis. Around 7 years of age, we also started to detect small clusters of lipid droplets (Fig. 3E; blue arrow). By 10 years of age, the lipid droplets were clearly accumulating to larger clusters and seemed to vary in abundance between tubules of the same individual (Fig. 3G–J; blue arrow), but increased consistently in cluster size and remained in high abundance after puberty. The accumulation of lipid droplets coincided with the increase of diameter and organization of tubule structure, including the thickening of the basal lamina, which was detectable in histology at 11 years of age (Supplementary Fig. S3E). From this point onward, tubules continuously increased in diameter with thickening of the basal lamina (Supplementary Fig. S3F–H). At 12 years of age, consistent euchromatic Sertoli cell nuclei with small clefts could be detected and the cytoplasm size increased drastically, such that a clear outline of the Sertoli cell plasma membrane could no longer be detected (Fig. 3I and L). Accompanying the increased width of the basal lamina was a change of nuclear shape of PMCs from oval/rounded to a thick-spindle like shape (Fig. 3, yellow arrow; Supplementary Fig. S4A–F), indicating the terminal differentiation of these cells at that stage.

Interestingly, in under 8-year-old testis we could mainly detect smooth cell–cell contacts between Sertoli cells and occasionally an electron-dense thickening between the adjacent Sertoli cell plasma membranes, referred to as desmosome-like (Dl) structures (Russell, 1977) (Fig. 4A–D). At 8 years, we started to detect smooth endoplasmic reticulum (sER) in the form of endoplasmic specializations associated to Sertoli cell junctions (Fig. 4E). This onset of maturational changes coincides with the appearance of lipid droplets observed around 7–8 years of age (Fig. 3E). We additionally observed a shift in the distribution of TJ protein occludin (OCLN) by immunofluorescence analysis after 7 years of age. OCLN was initially distributed between interstitial cells, before being increasingly detected in patches within the seminiferous cords at 7 and 8 years (Fig. 4K and L). By 11 years, OCLN was organized within the seminiferous tubules (Fig. 4M and N; Supplementary Fig. S4G–I), consistent with the formation of TJs and lumen formation at that age (Furuya et al., 1978; Guo et al., 2020). This organization of OCLN at 11 years of age coincides with the accumulation of lipid droplets, the increase in diameter and cellular organization of the seminiferous tubules, and the thickening of the basal lamina.

Sertoli cell maturation events are conserved between human and pig

We next aimed to analyze the content of lipid droplets. The inability to obtain a sufficient amount of human specimens for the downstream analysis prompted us to use 8-week-old porcine Sertoli cells as a translational model at the stage coinciding with the porcine SSC metabolic maturation (Voigt et al., 2021). To reconciliate the results between the two species, we took advantage of two recently published porcine scRNA-Seq datasets (Zhang et al., 2021, 2022) to validate previous data showing the age-dependent shift in metabolism in spermatogonia in pigs similar to that observed in humans (Voigt et al., 2022), and to determine which specific maturational changes in the niche are associated with the maturation of SSCs in pigs.

UMAP embedding representation of merged datasets showed 19 cell clusters (Supplementary Fig. S5A), which were subsequently categorized into eight cell types (Supplementary Fig. S5A–A′) based on known cell-type markers for testicular cells (Supplementary Fig. S5B). We then carried out a focused analysis of undifferentiated spermatogonia (Cluster 4, Supplementary Fig. S5A′) and Sertoli cells (Clusters 1, 2, and 3, Supplementary Fig. S5A′). The analysis of computed signature scores for OXPHOS, glycolysis, and hypoxia-related gene sets was performed across different age groups. The data showed enrichment of the OXPHOS-related gene set in 1-week-old spermatogonia (7D) compared to 8-week-old spermatogonia (60D), while a significantly higher score for glycolysis and hypoxia-related gene sets was detected in 8-week-old spermatogonia (60D) as compared to the 1-week-old spermatogonia (7D) (Supplementary Fig. S5C). These results aligned with the metabolic transition occurring from 1- to 8-week-old porcine spermatogonia, as we recently reported (Voigt et al., 2021), and highlighted for the first time the distinct metabolic differences in prepubertal SSCs relative to their adult counterparts, similar to the results obtained in humans (Voigt et al., 2022).

We then applied a similar Monocle workflow to generate the developmental trajectory of Sertoli cells (shown in Supplementary Fig. S6A and A′) and identified three distinct states that these cells exhibit during their development (Supplementary Fig. S6B and C). We next analyzed the expression of genes that significantly changed as these cells transit through branch point 1 from State 1 toward State 2 or 3 and identified four distinct gene clusters according to their expression pattern (Supplementary Fig. S6C). A comparative analysis of the four gene set clusters using Metascape (Supplementary Fig. S6D) showed that an enrichment of genes (Clusters 2 and 3) related to androgen response, OXPHOS, cholesterol homeostasis, hypoxia and cytokine signaling (i.e. IL2-STAT5-signaling, Interferon gamma response and TNF-alpha signaling) was observed when cells move toward State 3 (maturing Sertoli cells). These results are consistent with the human data analysis described above. Furthermore, the Cluster 1 gene set, mainly upregulated toward the end of the trajectory, was characterized by a drop in the expression, among others, of gene sets related to cell proliferation (CMyc and E2F targets), cytokine-related signaling and epithelial-mesenchymal transition; the latter may be an indicative of gain in cell polarity of Sertoli cells, thus marking cell maturation. Furthermore, we could confirm that Sertoli cell maturation is initiated at 8-weeks of age (60D) (Zhang et al., 2022). PRND was identified as a novel marker for mature Sertoli cells (Zhang et al., 2022). Plotting AMH and PRND along the pseudotime indeed showed a decrease in AMH expression and an increase in PRND expression in a small population of 8-week-old Sertoli cells (Supplementary Fig. S6E).

We then assessed the progression of structural changes in 8-week-old porcine Sertoli cells using histology and TEM compared to 1-week-old Sertoli cells. At 8 weeks of age, a trend toward a higher organization of the seminiferous epithelium in histology and an irregular nuclear envelope with nuclear clefts in Sertoli cells was observed by TEM (Fig. 5A–F).

In earlier stages (1-week-old), cell–cell interactions were predominantly smooth, and we rarely noticed Dl structures (Fig. 5G). In 8-week-old specimens, Dl structures could frequently be observed, together with the occurrence of TJs and an accumulation of intermediate filaments (IF) and sER, bordering these connection sites (Fig. 5H). Gene expression analysis, using qRT-PCR, detected a significant downregulation of cytokeratin-18 (KRT-18), anti-Müllerian hormone (AMH), proliferating cell nuclear antigen (PCNA), and upregulation of genes associated with increasing cell polarity and barrier formation. Specifically, we detected a 41.4-fold upregulation of androgen receptor (AR) expression (P = 0.03, n = 4) and a 45.6-fold upregulation of OCLN (P = 0.004, n = 4). Furthermore, the peroxisomal multifunctional protein-2 (MFP-2 or HSD17B4), essential for lipid homeostasis in adult mouse Sertoli cells (Huyghe et al., 2006), showed a 28.6-fold enhanced expression (P = 0.01, n = 4) (Fig. 5I, Supplementary Table SII).

Remarkably, in accordance with the increase of MFP-2 and similar to what we detected in humans, we observed an accumulation of lipid droplets in histology and TEM with Sertoli cell maturation (Fig. 5A–D; blue arrow). We confirmed those to be neutral lipid derivatives with BODIPY staining in IHC. This lipid accumulation was observed in all samples (n = 4) and concurred with the loss of AMH expression in 8-week-old Sertoli cells (Fig. 5J). Quantification of lipid droplets per Sertoli cells per tubule cross section revealed that 0.17 ± 0.08 vs 0.82 ± 0.05 lipid droplets/Sertoli cell can be detected in 1- versus 8-week-old porcine Sertoli cells (P = 0.0004, n = 4) (Fig. 5K) with a significantly higher surface area per droplet (2.66 ± 0.94 versus 17.90 ± 2.81 $µ$m², P = 0.007, n = 3) (Fig. 5L).

These ultrastructural changes were accompanied with decreased proliferative activity (Supplementary Fig. S7A) and enhanced glycolytic flux (Supplementary Fig. S7B and B′).

Our results indicate that changes in the lipidome may be associated with early Sertoli cell ultrastructural changes that accompany SSC maturation in pig and human testis.

Sertoli cell maturation coincides with a change in the niche lipidome in pigs

As we observed similar structural changes in early porcine Sertoli cell maturation, which also coincide with SSC metabolic maturation in humans and, in contrast to prepubertal human samples, porcine sample are readily available, we used porcine samples for further assessments. We performed LC–MS-based lipidomic profiling on whole isolated pig seminiferous tubules rather than enzymatically isolated cells to maintain cytoplasmic and plasma membrane structures, which otherwise collapse with isolation, as evident by vimentin staining in isolated Sertoli cells (Tang et al., 2018). The Sertoli cell is the predominant cell type in the seminiferous epithelium in these samples with no significant difference between the ages (germ cells 9.34 ± 0.54 versus 9.46 ± 2.06% and Sertoli cells 90.66 ± 0.54 versus 90.54 ± 2.07% in 1 versus 8  weeks, respectively).

A total of 6482 features were identified. The PCA two-dimensional (2D) scores plot with QC injections (purple) is shown in Fig. 6A. The 18 QC injections (purple, injection triplicates of six aliquots of a pooled mixture of all samples) are tightly clustered, displaying the reproducibility of the employed methods. The two groups (1- and 8-week-old) were fully separated, emphasizing deep changes in the lipidome of the samples (Supplementary Fig. S7C).

The identified and normalized lipids were mainly glycerophospholipids (35.4%), sphingolipids (29.0%), and glycerolipids (20.2%), and to a smaller percentage, lipids in the categories of fatty acyls, sterol lipids, prenol lipids, and polyketides (Fig. 6B with subclass distribution in Fig. 6B′). MS analysis of seminiferous tubules from 1- and 8-week-old pig testis revealed a significant change in 1654 lipids, with 830 lipids displaying lower intensity values and 824 displaying higher intensity values with maturation (fold change (FC) >1.5 or $≤$0.6667, raw-P < 0.05 and FDR < 0.25). We first analyzed the change of lipids per lipid category (Fig. 6C; Supplementary Fig. S7D–F). The classification of the lipids and abbreviation is presented in Supplementary Table SIII. We could detect 46 lipid subgroups that were significantly altered, 26 of which were on average upregulated in the 1-week-old Sertoli cells, and 13 of those were upregulated in the 8-week-old tubules.

Remarkably, 211 triglycerides (TGs) changed with cell maturation in the category of glycerolipids (Fig. 6C), which represent 16.1% of all detected glycerolipids, 49% of the significantly altered glycerolipids, and around 89% of the significantly changed TGs (Supplementary Fig. S7E and F). TGs were on average 2.63-fold upregulated with maturation (n = 4, P < 0.05). These non-membrane glycerolipids are neutral lipids that serve as storage in lipid droplets. Interestingly, these TGs were mainly composed of polyunsaturated-fatty acids (PUFA) (194 of 211) (Fig. 6D). These results align with the high expression of MFP-2, which is responsible for the production of PUFAs (Sprecher, 2000). The other component of the glycerolipids was the diacylglycerols (DGs) subclass, which was also on average upregulated with maturation, and which can serve as an intermediate for TGs biosynthesis.

Even though we observed this accumulation of lipid droplets with Sertoli cell maturation, a higher number of lipids were downregulated with maturation. The highest number of changed lipids were in the category of glycerophospholipids (Fig. 6C; Supplementary Fig. S5C). In this lipid category, the subclass glycerolphosphoinositol-monophosphates (PIP) decreased (on average 1.8-fold) with maturation, and a total of 18 PIP2 were significantly altered (Fig. 6D; Supplementary Table SIV).

In addition, 273 sphingolipids significantly decreased with Sertoli cell maturation, which represented 14.5% of all detected sphingolipids and 63% of all significantly changed sphingolipids (Supplementary Fig. S7F). Among these decreased sphingolipids were mostly ceramides (Cer-n-acylsphingosines) (105 versus 17 in 8 weeks), which consisted mainly of very long PUFAs (VLPUFA), and hexosyl ceramides (HexCer) (56 versus 13 in 8 weeks, also on average downregulated (Fig. 6D)).

Interestingly, of the sterol lipids detected, cholesteryl esters (CEs) were mostly upregulated with maturation whereas non-esterified sterol lipids (ST) were downregulated (Fig. 6C). CEs are an inactive form of sterol lipids often found in lipid droplets besides PUFA-TG in testis lipid droplets (Furland et al., 2003). Among those downregulated were, for example, vitamin D3 (8-fold lower), cholestenone (6.6-fold lower), cholest-4,6-Dien-3-One (2.5-fold lower), and other sterol lipids.

These data confirm the increase of storage lipids with maturation of porcine Sertoli cells as seen in TEM in the form of PUFA-TGs, DGs, and CEs. Furthermore, the data reveal that the early prepubertal niche lipidome is characterized by a high abundance of sphingolipids, especially ceramides, that decreases during maturation.

Quantitative proteomics revealed new phosphorylation sites associated with androgen signaling and cytoskeleton remodeling in porcine Sertoli cells

Sertoli cell development in humans and pigs seems to coincide with developmental events in prepubertal spermatogonial maturation.

We next aimed to analyze changes in phosphorylation and cell signaling. Phosphoproteins are a small proportion of the overall protein amount and require enrichment from a high cell number. We therefore performed quantitative proteomics and phospho-proteomics of isolated and enriched 1- and 8-week-old Sertoli cells at a purity of over 90% (see materials and methods) to allow for the detection of new phosphosites specifically in Sertoli cells with high confidence. Mass spectral peptide data was analyzed using the MaxQuant (Cox and Mann, 2008) software package v.1.6.10.23 with the Andromega algorithm (Cox et al., 2011) and peptides were matched to the porcine UniProt protein database, at a peptide-spectrum match FDR < 0.01. Protein–protein interactions were identified using STRING v11 (string-db.org) and the UniProt protein database. We identified 23 differentially expressed proteins and around 90 differentially phosphorylated peptides between 1-week and 8-week-old Sertoli cells (Fig. 7A; Supplementary Tables SIV and SV).

Figure 7A shows changed total proteins (blue and red boxes), phosphorylation sites and their known interactions with maturation. We detected a 10-fold upregulation of AR in 8-week-old Sertoli cells, which confirmed our qRT-PCR results. Heat shock protein-90 (HSP-90), that could associate with the AR in an unphosphorylated state and inhibit nuclear transfer (Dagar et al., 2019), showed around a 3-fold increase in phosphorylation. Interestingly, we also detected an upregulation of DNA damage checkpoint protein 1 (MDC-1) associated with lower phosphorylation, which is involved in enhanced androgen signaling in cancer cells (Wang et al., 2015), suggesting a similar contribution in maturing Sertoli cells.

Cytoskeletal reorganization, as detected in TEM, was accompanied with a downregulation of beta-actin (ACTB) and cofilin-2 (CFL2), both associated with increased cell motility. Interestingly, whilst CFL2 is associated with actin filament depolymerization, we detected an upregulation of the LIM domain and actin binding protein 1 (LIMA1) with maturation, responsible for actin filament bundle assembly and negative regulation of actin filament depolarization. In addition to higher total protein, maturing Sertoli cells showed a 4-fold lower phosphorylation of LIMA1 at Serine 682 and 695. Phosphorylation of LIMA1 at Serine 695 by MAPK1/3 has been described to reduce F-actin binding and enhance cell motility by facilitating actin reorganization (Han et al., 2007).

Furthermore, we detected changes in proteins associated with microtubule organization: for example, an increase of tubulin alpha 1B. We detected decreased phosphorylation of stathmin (STMN1; Serine 16 and 25) by 9-fold in maturing Sertoli cells, which negatively regulate microtubule depolymerization. STMN1 has been described to be phosphorylated at S16 by protein kinase A (PKA), of which we also detected a decreased phosphorylation at Threonine 196 and 198 with maturation. These results indicate that a decrease of cAMP-PKA pathway activity could contribute to cytoskeletal remodeling during Sertoli cell maturation, in addition to its known role in Sertoli cell proliferation (Meroni et al., 2019). These data align with the decrease of ceramides observed with maturation in our lipidomics analysis, as ceramides have been described to influence Sertoli cell function and work synergistically with the cAMP-PKA pathway (Meroni et al., 1999).

In addition, Sertoli cell cytoskeletal reorganization was accompanied with increased phosphorylation of myosin (MYH7), vimentin (VIM), sorbin and SH3 domain-containing binding protein (ARGbp or SORBS2). ARGbp is interacting with Wiscott-Aldrich syndrome protein family member 1 (WASF1) and palladin (PALLD); both proteins are pivotal for cell polarization (Fig. 7B). Notably, the initial formation of TJs as part of the BTB junctional complex between Sertoli cells was associated with higher phosphorylation of the broad substrate specificity ATP-binding cassette transporter (ABCG2) and 3-fold higher phosphorylation of gap junction alpha-1 protein (GJA1, connexin-43); these phosphorylation sites potentially enhance the assembly and maturation of the junctional complexes in porcine Sertoli cells (Fig. 7B).

Besides its importance for cytoskeletal remodeling, LIMA1 has been described to enhance cholesterol uptake and regulate cholesterol homeostasis. Similarly, we could detect a 4-fold decreased phosphorylation of the sterol-regulatory-binding protein (SERBF1) in maturing Sertoli cells, allowing its translocation into the nucleus in an unphosphorylated state and the expression of target genes associated with lipid and cholesterol biosynthesis (Yoon et al., 2009; Li et al., 2011). This potentially represents one of the molecular mechanisms underlying the increased accumulation of lipid droplets observed with Sertoli cell maturation in pigs and humans (Fig. 7B).

Overall, we observed a downregulation of protein phosphorylation associated with the cAMP-PKA pathway and SERBF1 and an upregulation of the AR-signaling pathway with the beginning of Sertoli cell maturation that coincide with changes in cytoskeletal reorganization and lipidome. Decreased cAMP-PKA activity could be related to decreased responsiveness to FSH signaling that occurs in maturing Sertoli cells and decreased FSH plasma levels at 2 months of age in pigs (Franca et al., 2000; Walker and Cheng, 2005).

Discussion

The characterization of events that coincide with the distinct metabolic transitions during SSC prepubertal maturation is necessary to develop predictive biomarkers to refine in vitro maturation strategies matched to an individual’s SSC maturation status.

During the prepubertal period, SSCs undergo a drastic change in morphology at 11 years of age prior to transitioning to an adult-like SSC metabolism in humans (Voigt et al., 2021, 2022). Therefore, we hypothesized that prepubertal development of spermatogonia occurs on interaction with the immature niche environment. We unraveled here that changes in the niche lipidome are associated with maturation events in Sertoli cells and PMCs. These events coincide with the morphological maturation of SSC and are followed by ongoing pubertal Sertoli cell maturation events and the metabolic maturation of SSCs after 11 years of age (Voigt et al., 2022).

Our scRNA-Seq data analysis supports the existence of two immature states and a single mature state, which Sertoli cells go through during testis development (Guo et al., 2020; Zhao et al., 2020). The maturation in mouse Sertoli cells has been described as stepwise (Tan et al., 2020), whereas the maturational heterogeneity observed in TEM points toward an asynchronous maturation in Sertoli cells during prepubertal development in humans. In agreement with earlier reports (Guo et al., 2020; Zhao et al., 2020), we observed the initiation of Sertoli cell maturation (entering State 3) at 11 years of age in humans, which coincides with SSC shape changes that precede metabolic maturation. In contrast, major aspects of Sertoli cell maturation occur within just 1 week during murine development (Tan et al., 2020), underlining the importance of a larger animal model to dissect out details of prepubertal development in humans.

Further analysis revealed that Sertoli cell maturation continues to evolve throughout puberty and is characterized by distinct stages and metabolic transitions. These changes were linked, among others, to increased glycolysis, an increase in anchoring junctions, and enhanced cholesterol homeostasis. Being a precursor of steroid synthesis, cholesterol, together with other lipids, serves as a ‘fuel’ for Sertoli cells (Shi et al., 2018). Cholesterol is also involved in remodeling the germ cell membrane during development (Potter et al., 1981) and is essential for the retention of claudins in the plasma membrane and the formation of TJs (Shigetomi et al., 2018). These results indicate that Sertoli cells initiate maturation at 11 years of age, continue to develop throughout puberty, and emerge into an adult-like profile at 13/14 years of age. Therefore, the completion of spermatogenesis occurs in association with the adult-like profile of Sertoli cells.

Similarly, PMCs undergo a structural cell reorganization and activate several cellular processes during the prepubertal phase. These include androgen response, OXPHOS, fatty acid metabolism and cell proliferation. Several of these cellular processes remained activated or further increased (i.e. androgen response, estrogen response) throughout the peri/postpubertal phase. Consistent with a highly proliferative and active metabolic state in PMCs, we observed an increased thickening of the basal lamina starting at 11 years of age. Interestingly, also in mice, major aspects of myoid cell maturation are aligned with Sertoli cell maturation and associated with similar gene networks such as estrogen receptor (Tan et al., 2020).

In summary, our data support a progressive change in various cellular processes initiated at 11 years of age that accompany changes in cellular structure in Sertoli cells and PMCs.

Porcine Sertoli cell analysis showed a progression of events similar to that in humans, evident by detecting distinct stages and metabolic transitions during cell maturation in both species. These similarities present the pig model as a suitable translational model to address questions related to the niche and SSC development in humans. By 2 months of age, coinciding with SSC maturation in pigs (Voigt et al., 2021), junctional formation is initiated. The formation of TJs, initiated at 8 weeks in pigs and at 11 years of age in humans, may significantly shift the niche metabolic microenvironment. TJs in Sertoli cells and their connections to the cytoskeleton are basally located within the seminiferous epithelium (Mruk and Cheng, 2015). This formation creates an intra- and extracellular polarization and a tight basal compartment, which serves as the basis for the adult SSC niche. Blood-derived hydrophilic molecules are unable to penetrate through this Sertoli cell barrier, thus creating gradients of blood-derived molecules between the basal and apical compartments (Setchell et al., 1969; Tuck et al., 1970; Griswold, 1995; Mruk and Cheng, 2004; Wong and Cheng, 2005; Zihni et al., 2016). Additionally, the final differentiation of PMCs, associated with a drastic thickening of the basal lamina at that stage specifically in humans, may change the diffusion to and from the niche. These aspects of niche polarization, by changing the paracellular diffusion potential, indicate a critical point in the formation of the SSC niche metabolic microenvironment.

Interestingly, these changes initiated at 11 years of age in human and 8 weeks in pigs were associated with an accumulation of lipid droplets, which are highly mobile and dynamic structures that interact with cell junctions and organelles to ensure the synthesis, uptake, usage and elimination of cholesterol (Shi et al., 2018). We therefore suggest that the accumulation of lipid droplets could serve as an indicator for ultrastructural changes and represent an indirect marker of SSC maturation.

With lipidomic analysis, we identified an increase of PUFA-TG and CE, previously described in rodent testis and seminiferous tubules (Bridges and Coniglio, 1970; Aveldano et al., 1993; Furland et al., 2003). PUFA-TG and CEs are both neutral storage lipid groups, found in lipid droplets of various tissues (Beckman, 2006), suggesting that the accumulation of lipid droplets represents the change in these neutral storage lipids detected with lipidomics. Lipids and cholesterol are essential for spermatogenesis. They play a critical role in Sertoli cell function and the membrane remodeling of germ cells during their development. It has been described that lipid droplets accumulate during Sertoli cell maturation (Flickinger, 1967; Tran et al., 1981) and differentiation (Liang et al., 2019). Lipid droplets may potentially function as building blocks for developing germ cells (Furland et al., 2003) or represent phagocytic bodies of apoptotic germ cells (Wang et al., 2006). In the drosophila testis, the incorporation of PUFA into lipid droplets serves as protection from lipid peroxidation, and therefore, it has an antioxidative role after apoptosis in the stem cell niche (Bailey et al., 2015). Furthermore, excessive accumulation of CE disturbs cholesterol homeostasis, which ultimately results in infertility (Kastner et al., 1996; Robertson et al., 2005). The upregulation of genes associated with fatty acid metabolism and cholesterol homeostasis during the maturation of Sertoli cells in humans, together with the observed increase in the number and size of the lipid droplets in pigs, support their role in maintaining an efficient lipid flux to prevent lipotoxicity.

Specific lipid compositions have been associated with TJ and OCLN proteins (Nusrat et al., 2000; Marchiando et al., 2010; Pelletier, 2011). It is possible that the lipidome shift during development plays a potential role for polarization and barrier formation, yet the mechanisms of those remain to be investigated.

Cell polarization, lipid droplet formation, and lactate production in Sertoli cells may be regulated by the peroxisome proliferator-activator receptor (Gorga et al., 2017). Here, we report a 4-fold decrease in the phosphorylation of SREBF-1 with maturation. SREBF-1 is a key transcription factor involved in regulating the main enzymes associated with lipid biosynthesis. Phosphorylation of SREBF-1 negatively regulates its nuclear transfer and action (Li et al., 2011) that, therefore, suggests a similar role of SREBF-1 in regulating lipid metabolism in Sertoli cells with maturation.

Notably, lipidomics revealed that a higher number of lipids decreased with maturation. While sphingolipids and VLCPUFA-Cer are important for spermatocytes and meiosis (Rabionet et al., 2008), we detected here a downregulation of sphingolipids with Sertoli cell maturation. A downregulation of 105 ceramides (Cer), encapsulating mostly PUFA-Cer in earlier Sertoli cell stages was observed. PUFA-Cer contribute to the cAMP-PKA pathway in Sertoli cells (Meroni et al., 1999). We further showed an increased phosphorylation of PKA in earlier Sertoli cell stages. Sphingolipids and ceramides are associated with extracellular vesicle formation and therefore cell–cell communication via exosomes (Wang et al., 2020). In this context, we previously reported increased expression of the exosome marker flotillin-1 and lipid raft formation in 1-week-old spermatogonia (Voigt et al., 2022), potentially responsible for the uptake of exosomes. Together, these results suggest that a specific lipid metabolism in the niche may be important for increased extracellular vesicle exchange between Sertoli cells and early prepubertal spermatogonia during their migration toward the basement membrane.

This study provides an extensive resource of the transcriptomic, (phospho)-proteomic, and lipidomic evolution of niche development that coincides with SSC maturation and, as such, could provide tissue-based biomarkers for the assessment of SSC developmental status. These findings serve as a basis to further characterize niche maturational events that determine SSC maturation.

Supplementary data

Supplementary data are available at Molecular Human Reproduction online.

Data availability

The proteomics data underlying this article are available via ProteomeXchange with identifier PXD037764. The scRNASeq data underlying this article are freely available through NCBI GEO with accession number GSE196819, or are publicly available online via the respective GEO accession numbers as cited. The lipidomics data underlying this article will be shared on reasonable request to the corresponding author.

Acknowledgements

We are grateful to Dr William Walker (University of Pittsburgh) for critically reviewing this manuscript. Transmission electron microscopy was performed at the Microscopy and Imaging Facility (MIF) at the University of Calgary (Calgary, AB, Canada). Lipidomics profiling was performed at The Metabolomics Innovation Center (TMIC) University of Alberta (Edmonton, AB, Canada), www.metabolomicscentre.ca.

Authors’ roles

Conceptualization: A.L.V.; Methodology: A.L.V., N.L.M.L., R.D., I.D., and A.D.; Formal Analysis: A.L.V., R.D., and T.H.; Investigation: A.L.V., T.H., R.D., H.S., and A.D.; Resources: I.D., K.E.O., and A.D.; Writing—Original Draft: A.L.V. and R.D.; Writing—Review and Editing: I.D., H.S., A.D., and N.L.M.L.; Visualization: A.L.V. and R.D.; Supervision: I.D.; and Funding Acquisition: I.D. and K.E.O.

Funding

This work was funded by NIH/NICHD R01 HD091068 and NIH/ORIP R01 OD016575 to I.D., K.E.O. was supported by R01 HD100197.

Conflict of interest

The authors declare no competing interests.

References