Background
Cannabis is the most widely used drug in the world. It is also the commonest recreational drug consumed by pregnant women. In western countries, up to 20% of pregnant women are exposed to cannabis during pregnancy [
1]. In the USA, while fetal exposure to tobacco decreases, cannabis use during pregnancy rises along with the enhanced perception that there are no risks with cannabis consumption during pregnancy [
2]. Thus, an increase of cannabis use by 3.4–7% of pregnant women in the USA has been reported during this last decade [
3]. However, data on pregnant women consumption remain rare and the use of self-reported questionnaires likely underestimates real exposures. Indeed, phytocannabinoid compounds have been evidenced in 5.3% of newborn meconiums, whereas only 1.7% of the mothers self-reported cannabis use [
4]. Cannabis components and their metabolites are well known to cross the placenta barrier [
5]. Use of cannabis during pregnancy can induce negative birth outcomes, such as reduced weight, increased risk of prematurity, cognitive deficits, and behavioral and neurocognitive impairment. To date, the effect of this consumption on the human genital tract is unknown [
6‐
9].
The main chemicals present in recreational cannabis are Δ9-tetrahydrocannabinol (THC), which is responsible for the psychoactive effects of marijuana, and cannabidiol (CBD). The presence of THC in fetal blood after maternal ingestion and THC and CBD in meconium between 0 and 48 h after delivery indicate these two molecules cross the placenta [
10‐
13]. THC and CBD act on specific cannabinoid receptors called CB1 and CB2. Both of these receptors also respond to naturally synthesized cannabinoids, namely anandamide (AEA) and 2-arachidonyl-glycerol (2-AG), so-called endocannabinoids. In addition to the endocannabinoids and their receptors, this endocannabinoid system (thereafter called ECS) comprises synthesis and degradation enzymes and transporters [
14]. AEA and 2-AG, which both derive from membrane phospholipid precursors and arachidonic acid, are synthesized and metabolized by different pathways from the ECS. The ECS, which is sensitive to phytocannabinoids, plays a fundamental physiological role in the brain and peripheral tissues, as well as in many pathological conditions [
14]. ECS has been involved in reproductive function at both the central and gonadal levels in human and animal models [
15,
16]. Although cannabinoid receptors have been described in the mouse fetal testis [
17], there is no description of the ECS in the human fetal testis. To date, the potential effect of cannabis exposure on the developing testis is unknown. We recently showed that the endocannabinoid system is present in the adult human testis [
18]. In young men, several studies have demonstrated an association between cannabis consumption and testicular germ cell tumors (TGCTs) development [
19‐
21], which could reflect the sensitivity of testicular cells to cannabis compounds.
In this context, we aimed at investigating the expression and localization of the ECS components in the human fetal testis and determining whether phytocannabinoids can directly affect fetal testicular cells, in order to unveil a potential implication of cannabis exposure in utero in impaired testis development and subsequent deleterious consequences, including testis cancer. Indeed, the testicular dysgenesis syndrome hypothesis implies that the alteration of the development and functions of any type of fetal testicular cell can have a negative impact on other distant cells and tissues [
22,
23]. This is especially true in the case of impaired steroidogenesis by Leydig cells, since testosterone is key for fetal testis development and other organ masculinization. [
22,
23]. In this study, we demonstrate that cannabinoid receptors CB1 and CB2 as well as endocannabinoid synthetizing and degrading enzymes are present in the human fetal testis. Moreover, our data uncover deleterious effects of direct exposures of THC and CBD onto the developing human testis and its endocrine functions ex vivo.
Methods
Chemicals
CBD (Axon Medchem, Netherlands) was diluted in dimethylsulfoxide (DMSO) solution, while THC (Sigma-Aldrich; CAS number 1972–08-3, USA) was diluted in ethanol (EtOH) solution. The same amount of DMSO, EtOH, and a mixture of DMSO and EtOH were used as controls for CBD, THC, and THC/CBD mixture, respectively.
Human fetal testis collection and ethic approval
First trimester (7–12 developmental weeks (DW), i.e., 9–14 weeks from last menstrual) human fetuses were obtained from pregnant women following legally induced abortion performed in Rennes University Hospital (France). None of the pregnancy terminations were due to fetal abnormalities. All women received information about the project and gave informed written consent in accordance with national guidelines enacted by the French Agency for Biomedical Research (Agence de la Biomédecine, #PFS09-011). The local ethic committee of Rennes hospital (advice #11–48) approved the protocol. The aspiration products were placed at 4 °C after collection and rapidly processed. The testes were recovered from the aspiration products under a binocular microscope (Olympus SZX7, Lille, France) and immediately placed in cold phosphate-buffered saline (PBS) solution.
Organotypic cultures of fetal testis
The fetal testes were cut into explants of 1mm
3 and placed in cell culture inserts (0.4 µm pores, Falcon, Becton–Dickinson, Le Pont de Claix, France) put in 24-well companion culture plates (Falcon, Becton–Dickinson) according to a standardized protocol previously described [
24,
25]. On each insert, 3 or 4 testicular explants were placed and cultivated in 350 µl of phenol red-free medium 199 (Invitrogen Life Technologies, Saint-Aubin, France) supplemented with 50 µg/mL gentamicin (Sigma-Aldrich, Saint-Quentin, France) and 2.5 µg/mL Fungizone (Sigma-Aldrich). As determined in previous experiments, human chorionic gonadotropin (hCG, Sigma-Aldrich) was added to a final concentration of 0.1 IU/mL in order to maintain steroidogenic responsiveness of the testicular tissue. The cultures were incubated at 37 °C for 96 h in a humidified atmosphere of 95% air and 5% CO
2. The medium was removed every 24 h and divided into 2 aliquots that were immediately snap-frozen on dry ice and stored at − 80 °C. Explants were cultured in medium 199 supplemented with hCG for the first 24 h of culture (D0) without any drug, in order to define a baseline for hormone production in each well. This allows normalization of the explant secretory capacity when the drugs are added [
24]. After D0, the explants were exposed to treatments for 3 days by adding CBD and/or THC at concentrations of 10
−5 M, 10
−6 M, or 10
−7 M to the medium, or to dimethylsulfoxyde (DMSO, Sigma-Aldrich) and/or ethanol, the solvents of CBD and THC, respectively. Medium was changed every 24 h. The range of concentrations of THC and CBD was chosen based on peak plasma concentrations (Cmax) in adults, which differ depending on the mode of administration of cannabis. When smoked, the most common consumption method, THC Cmax, usually reaches 10
−7 M [
26]. Phytocannabinoid pharmacokinetics may differ from occasional to heavy users and according to the user’s age [
27]. Since only few studies assessed the quantitative relation between mother consumption and the distribution of phytocannabinoids in the placental and fetal compartment, we decided to study a large range of concentration, from 10
−7 M to 10
−5 M.
For the 2-week-long culture, insulin, transferrin, and selenium (ITS, Corning, Ref 354352) were added to the previously described culture medium, as we found it entitled better preservation of fetal testis morphology and functions in longer culture. In this case, explants were cultured for the first 24 h of culture without drugs and then with either control, CBD 10−5 M, THC 10−5 M, CBD/THC 10−6 M, or CBD/THC 10−5 M. The medium was changed every 48 h and kept frozen at − 80 °C at day 4 (D4), D8 and D14 for subsequent analysis.
MALDI imaging of 2-AG
A 11 DW human fetal testis was embedded in carboxymethylcellulose (M-1 embedding matrix, Thermo Scientific) then frozen on dry ice and stored at − 80 °C. Sections 10-µm-thick were prepared with a CM 3050S cryostat (Leica) at − 17 °C and thaw-mounted onto a glass slide coated with indium tin oxide (Bruker). The MALDI matrix solution was composed of 2,5-dihydroxybenzoic acid (DHB) at 50 mg.mL−1 in methanol/water/trifluoroacetic acid (70/30/0.1). An HTX M5 sprayer (HTXImaging) was used to deposit the DHB solution on the slide by performing 8 passes at a spray temperature of 75 °C and a flow rate of 0.1 mL.min−1. The tissue section was analyzed in positive ion mode on a 7 T MALDI-FT-ICR SolariX XR mass spectrometer (Bruker). Spectra were acquired in the m/z 150–900 mass range with a resolving power of 130,000 at m/z 400 and a lateral resolution of 10 µm. Every spectrum was calibrated with internal lock masses. MALDI images were reconstructed and normalized on root mean square using the fleximaging 5.0 software.
Endocannabinoid measurements in fetal testes
Testes were removed, flash frozen, and stored at − 80 °C until analysis. Measurements of anandamide (AEA) and 2-arachidonoylglycerol (2-AG) were carried out as previously described (Lourenço et al. 2011). Briefly, tissues were homogenized and extracted by liquid–liquid extraction with chloroform/methanol/Tris–HCl 50 mM, pH 7.5 (2:1:1, vol/vol) containing internal deuterated standards (AEA-d4, PEA-d4, OEA-d4, and 2-AG-d5). Phase separation was facilitated by centrifugation at 10,000 g for 2 min. The lipid-containing organic phase was concentrated on an N2 stream evaporator, while the protein-containing aqueous phase was treated with trichloroacetic acid (TCA) in order to precipitate the protein. Protein amounts were determined by the Lowry method using the BCA protein Pierce (ThermoScientific, CA) assay with bovine serum albumin (BSA) as standards. The dried lipid extracts were then subjected to isotope-dilution liquid chromatography-chemical ionization-tandem mass spectrometric analysis. Mass spectral analyses were performed on a TSQ Quantum Access triple quadrupole instrument (Thermo-Finnigan) equipped with a APCI source (atmospheric pressure chemical ionization) and operating in positive ion mode. The TSQ Quantum Access triple quadrupole instrument was used in conjunction with a Surveyor LC Pump Plus (Supelco C18 Discovery Analytical column) and cooled autosampler. The amounts of anandamide and 2-AG were determined by isotope-dilution using a calibration curve and expressed as pmol by mg of proteins.
Hormone measurements in fetal testes
Hormone levels were measured in the culture medium of testis explants at D0 and D3 for short-, and at D0, D4, D8, and D14 for long-term culture experiments. Each sample was assayed in duplicate for all hormone measurements. Testosterone levels were assayed with a specific radioimmunoassay (RIA) kit (Beckman Coulter, Ref IM1119), according to the kit manufacturer’s instructions. Insulin-like factor 3 (INSL3) levels was measured using a specific RIA kit according to the kit manufacturer’s instructions (Phoenix, Ref RK-035–27). Anti-Müllerian Hormone (AMH) levels were assayed by Enzyme-Linked Immunosorbent Assay (ELISA) kit (Beckman Coulter, Ref A79765). The hormone assay values are all expressed as fold change compared to the mean of their respective control, after a normalization of hormone production with D0 values for each sample.
Immunohistochemistry and cell counting
Immunohistochemistry was performed on Bouin solution-fixed, paraffin-embedded, 5 µm-thick sections of testis explants. Each fifth section was used for immunohistochemical analysis with cell-specific labeling, as previously described [
24,
25,
28]. Following dewaxing and rehydration, antigen retrieval was performed for all immunostaining by treating the sections with 10 mM citrate buffer (pH 6.0) at 80 °C for 45 min or Tris-Ethylenediaminetetraacetic acid (EDTA buffer (pH 9.0) at 80 °C for 30 min before cooling them at room temperature (RT). Non-specific sites were then blocked with a 10% BSA or chicken serum treatment for 1 h at RT and sections were incubated overnight at 4 °C with the primary antibody diluted in Dako antibody diluent (Dako, Ref S3022) or PBS-Tween 0.01%.
Leydig cells were labelled with a rabbit primary antibody directed against the cytochrome P450, family 11, subfamily A, polypeptide 1 (CYP11A1) (1:250, Sigma-Aldrich, Ref HPA016436). Germ cells were stained using LIN28 rabbit primary antibody (1:800, Abcam, Ref Ab462020). Sertoli cells were labelled using AMH goat primary antibody (1:200, Santa-Cruz, Ref sc6886). Apoptotic and proliferating cells were stained with a polyclonal rabbit primary antibody directed against cleaved caspase-3 (1:100; Cell Signaling Technology, Ref 9661), and a monoclonal mouse primary antibody directed against KI-67 (1:100; Dako, Ref M7240), respectively. The appropriate biotinylated goat anti-mouse (Dako, Ref E0433) and anti-rabbit (Vector, Ref BA-1000), and rabbit anti-goat (Dako, Ref E0466) antibodies were used at 1:500 as previously described [
29]. Signal corresponding to the secondary antibody was amplified by an incubation with an avidin and biotinylated enzyme complex (ABC Vectastain Elite Kit, Vector Laboratories). Sections were developed using 3,3’-diaminobenzidine tetrahydrochloride (DAB, Sigma-Aldrich) and then counterstained with Masson’s Hemalun and Lithum Carbonate before dehydration. Between all steps (except between blocking of non-specific sites and the first antibody incubation), sections were washed using phosphate buffer saline (PBS) with 0,1% of Tween 20 detergent (Sigma-Aldrich). The number of LIN28-positive germ cells was determined by counting the stained cells in 8–12 randomly selected histological explant sections of each condition, using a light microscope (Bh2 Olympus Microscope) coupled to Mercator Expert Software (Explora NOVA). For the 3-day cultures, apoptotic cleaved caspase 3-positive and proliferative KI67-positive cells were counted in the testis cords (including germ and Sertoli cells) and interstitial tissue (including Leydig cells) in 8–12 randomly selected histological explant sections of each condition, giving so-called “intracordonal” and “extracordonal” counting values. The surface area of each section was measured and data presented as cell number per surface area unit.
In long-term cultures, relative Leydig and Sertoli cells, and apoptotic cell area were estimated by relating the surface occupied by the CYP11A1-positive Leydig cells, AMH-positive Sertoli cells, and cleaved caspase-3-positive cells respectively onto the total surface of 8–12 randomly selected histological explant sections. For this purpose, we measured the surface occupied by each investigated cell type labelled in dark brown by DAB and related it to the surface of the corresponding explant using ImageJ software (US National Institutes of Health, Bethesda, MD, USA).
In situ hybridization (RNAscope)
RNAscope was performed on 5 μm formalin-fixed, paraffin-embedded tissue sections from 6 different donors of 7–14 DW using RNAscope 2.5 HD-Red kit (Advanced Cell Diagnostics; ACD) according to the manufacturer’s instructions with some modifications. Briefly, the slides were dried 1 h at 60 °C, deparaffinized, and pretreated for 10 min at RT with H2O2. For target retrieval, sections were boiled with the manufacturer’s target retrieval buffer (ACD Bio-techne) for 15 min at 90–100 °C and then washed twice in distilled water and once then 100% EtOH. To increase target accessibility, protease digestion was then carried out by incubating sections with protease plus solution (ACD Bio-techne) diluted 1:5 in PBS-DEPC at 40 °C for 30 min (HybEZ Oven, ACD Bio-techne). Sections were finally washed with distilled water before proceeding to probe hybridization. Sections were incubated with the CNR1 (Bio-techne, ref 591521) and CNR2 (Bio-techne, ref 596021) gene’s probes for 2 h at 40 °C. Positive (Polr2A; Bio-techne, ref 310451) and negative (DapB; Bio-techne, ref 310043) controls, developed by ACD Bio-techne, were also tested on our tissue sections to validate tissue quality and technique specificity. The slides were washed 2 times during 2 min in 1X wash buffer at RT, and the hybridized signals were then amplified by six consecutive signal amplification steps (Hybridize Amp 1–6, ACD Bio-techne) according to the manufacturer’s instructions, with the following modification: sections were incubated with Hybridize Amp 5 for 60 min at RT and washed twice 2 min each in 1X wash buffer (ACD Bio-techne) at RT after each signal amplification step. For signal detection, sections were incubated for 10 min at RT in red solution (RNAScope 2.5 HD Assay–RED, ACD Bio-techne) and rinsed in distilled water for 2 min.
Immunofluorescence
Sections stained for RNAscope were blocked with PBS-BSA at 10% for 30 min at RT and incubated overnight at 4 °C with primary antibody. Leydig cells were labelled with a rabbit primary antibody directed against CYP11A1 (1:100, Sigma-Aldrich, Ref HPA016436). Germ cells were stained with a LIN28 rabbit primary antibody (1:500, Abcam, Ref Ab46020). Sertoli cells were labelled with an AMH goat primary antibody (1/100, Santa-Cruz, Ref sc6886). Serial frozen sections from MALDI imaging were rehydrated in PBS, submitted to antigen retrieval with citrate buffer as described above, and processed for fluorescent CYP11A1 and AMH double immunostaining. The appropriate AF488 chicken anti-rabbit (Invitrogen, Ref A21441) or AF488 Donkey anti-goat secondary antibody (Invitrogen, Ref A11055) were used at 1:500 for 2 h at RT. PBS with 0.1% of Tween 20 detergent (Sigma-Aldrich) was used for washing steps. Slides were then mounted using Invitrogen Prolong Gold Antifade Mountant with DAPI (Invitrogen, P36935).
RT-qPCR
Total RNA was extracted using RNA/DNA extraction kit (Qiagen, Germany). After extraction, RNA was precipitated, washed, and diluted in RNA-free water. RNA quantity was assessed using a NanoDrop™ 8000 Spectrophotometer (Thermo Fisher Scientific), and RNA quality using a 2100 Bioanalyzer Instrument (Agilent Technologies, CA, USA). RT-qPCR was performed as described using iTaq Universal SYBR Green Supermix (Bio-Rad) and cDNA template in a CFX384 Touch Real-Time PCR Detection System (Bio-Rad). RPLP0 mRNAs were used as internal controls for normalization. Primers are listed in Table
1. Results calculated using the ΔΔCT method are presented as
n-fold differences in target gene expression relative to reference gene and calibration sample.
Table 1
Primers used for qPCR experiments to characterize isoforms of endocannabinoid receptors
CNR1 (CB1) | F: TCAGTACGAAGACATCAAAGGTG | 85pb | 60 °C |
R: CTTCCCCTAAAGGAAGTTAAAGG |
CB1A | F: AGACATCAAAGGAGAATGAGGAG | 113pb | 60 °C |
F: AGACATCAAAGGAGAATGAGGAG |
CB1B | F: AGACATCAAAGGAGAATGAGGAG | 84pb | 64 °C |
R: AATGTTCACCTGGTCTGCTG |
CB2A | F: GATTATGCCAGCCAGATGC | 77pb | 64 °C |
R: GCTCGGTGAGTGAGAGGTG |
Statistical analyses
The effect on testosterone secretion of CBD and/or THC treatments at various concentrations versus control was assessed using Freidman tests followed by the two-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli. For cell counting, AMH, Inhibin B, and INSL3 measurements, the control conditions, and CBD and/or THC 10−5 M treatments were compared two by two using Wilcoxon tests. For RNA-seq data analysis, mRNA expression between two ages was assessed using Kruskal–Wallis tests followed by the two-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli. For long culture treatment, Friedman test comparisons was used to measure difference between control and treated conditions, followed by two-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli. Statistical significance threshold was set at 0.05. Data are expressed as mean ± SEM.
RNA-seq experiments and data analysis
RNA-seq raw data were mined from our previously published dataset [
30] to study the dynamic of gene expression of endocannabinoids synthetizing and degrading enzymes.
BRB-seq experiments and data analysis
Preparation of the sample
Total RNA was extracted from 8 to 10 (n = 10) and 10 to 12 DW fetal testis explants (n = 10) after 72 h exposure to CBD, THC, or CBD/THC at 10−5 M and their respective controls, as described above.
Library preparation and sequencing
3’BRB-seq experiments were performed as previously described [
31‐
33]. Briefly, RNAs were distributed onto two 96-well plates. A first step of reverse transcription and template switching reactions was performed using 4 µL total RNA at 2.5 ng/µL and sample-specific barcoded oligo-dT. Then, cDNAs from each plate were pooled together and purified and double-strand (ds) cDNAs were generated by PCR. The two corresponding sequencing libraries were next built by tagmentation using 50 ng of ds cDNA with the Illumina Nextera XT Kit (Illumina, #FC-131–1024) following the manufacturer’s recommendations. The resulting library was finally sequenced on a NovaSeq sequencer as Paired-End 100 base reads by the IntegraGen Company (
https://integragen.com/fr/). Image analysis and base calling were performed using RTA 2.7.7 and bcl2fastq 2.17.1.14. Adapter dimer reads were removed using DimerRemover (
https://sourceforge.net/projects/dimerremover/).
Data preprocessing and normalization
A phred quality score higher than 10 was required for the first reads (R1, 16 bases). Among these, the first 6 bases correspond to the unique sample-specific barcode needed to demultiplex the sequencing data, while the following 10 bases correspond to a unique molecular identifier (UMI) used for quantification purposes. The second reads (R2) were aligned to the human reference transcriptome from the UCSC website using BWA version 0.7.4.4 with the parameter “ − l 24”. Reads mapping to several positions in the genome were filtered out from the analysis. After quality control and data preprocessing, a gene count matrix was generated by counting the number of unique UMIs associated with each gene in lines for each sample in columns. The UMI matrix was further normalized with the regularized log (rlog) transformation package implemented in the DeSeq2 package [
34]. Raw and preprocessed data will be deposit at the GEO repository. The GEO accession number is GSE223827 [
35].
Differential gene expression analysis
Principal component analysis (PCA) and Uniform
Manifold Approximation and Projection (UMAP) were performed with the FactoMineR [
36] and umap packages implemented in R v4.1.1. Differentially expressed genes (DEGs) were identified by comparing each compound (CBD, THC, CBD + THC) to their corresponding control samples at each developmental stage (10–12GW and 12–14GW). For each comparison, different filtration steps were applied: (i) the median gene expression value (= 0.0) of all the samples was used as a background cut-off; (ii) a foldchange cut-off of at least 1.3; and (iii) a statistical filtration with a paired samples
t-test and a
p-value cut-off of 0.05 adjusted with the Benjamini & Hochberg method [
37]. The resulting transcriptomic signatures was deposited at the TOXsIgN repository (
https://toxsign.genouest.org/) [
38].
Clustering and functional analysis
The resulting lists of DEGs were clustered into distinct gene expression patterns with the mclust algorithm [
39]. A Gene Ontology term enrichment analysis was performed with the Annotation Mapping Expression and Network (AMEN) [
40]. A specific annotation term was considered significantly enriched in a given gene expression pattern when the false discovery rate (FDR)-adjusted
p-value (Fisher’s exact probability) was ≤ 0.05 and the number of associated genes was ≥ 3.
Discussion
This study is the first to demonstrate the presence of the ECS components in the human fetal testis and to reveal an impact of phytocannabinoids on the fetal testis physiology ex vivo. The dynamic regulation of the expression of specific endocannabinoid receptors, endogenous ligands and their respective synthesizing and degrading enzymes during the first months of testis development suggests that ECS signaling plays a role in this process.
HPLC dosages evidenced the presence of the two best-described endogenous ligands of the ECS in 7–12 DW fetal testis, with a 1000 fold higher levels of 2-AG compared with AEA. Similar elevated levels of 2-AG
versus AEA (up to 1000 times) have been described in the mouse brain [
41]. The respective roles of 2-AG and AEA in the human fetal testis are currently unknown. However, the especially high level of 2-AG in the fetal testis suggests a fundamental role in its development. 2-AG being an important metabolite in prostaglandin synthesizing pathway [
42], it could also have an ECS-independent mode of action (e.g., inflammation regulation). Both 2-AG and AEA intratesticular levels increased between 7 and 12 DW, albeit only significantly for AEA, likely because of the high inter-individual variability for 2-AG levels. 2-AG in situ detection by MALDI imaging in 11 DW human fetal testis demonstrated its localization primarily within CYP11A1-positive Leydig cell-enriched areas in the interstitial tissue, which contrasts with its localization within the seminiferous tubules of the adult human testis [
43]. Similarly to adult testis, we failed to detect AEA in fetal testes in situ, which is probably due to its very low expression level [
43].
In agreement with the detection of 2-AG and to a lower level of AEA, our analysis of human fetal testis bulk RNA-seq data demonstrated that 6–17 DW testes express the synthesizing and degrading enzymes of these molecules in variable quantities. 2-AG synthesis is first mediated by PLC enzymes [
44,
45], which isoforms
PLCG1 and
PLCD1 were detected between 7 and 12 DW. Likewise, we demonstrated the expression of 2-AG main synthesizing (
DAGLA and
B) and degrading enzymes (
MAGL). Levels of 2-AG increased between 7 and 12 DW but did not reach significance, consistent with the stable expression of
PLCG1, PLCD1, DAGLA, DAGLB, and
MAGL enzymes. The expression of
ABDH12 [
46], another 2-AG degrading enzyme, was also stable from 9 to 12 DW. ABHD2 is an alternative degrading enzyme of 2-AG [
47], expressed by adult testicular germ cell, in which it may regulate meiosis entry [
43]. The increase of its expression between 7 and 12 DW suggest that this enzyme is not primarily involved in 2-AG degradation in fetal testis at this time. AEA main synthesizing enzymes
NAT10 and
NAPE-PLD were also present in the fetal testis, as well as its main degrading enzyme,
FAAH. However, the decreased expression of
NAT10 concomitantly to the increased expression of
FAAH between 6 and 12 DW, a time window during which AEA levels significantly increased, suggest that AEA synthesis and degradation occurs through alternative pathways in the first trimester testes. As a matter of fact, we detected the stable expression of two AEA alternative synthesizing enzymes (
ABHD4 and
GDE1) [
48,
49] and two AEA alternative degrading enzymes (
FAAH2 and
CYPXN1) [
50‐
52] in testes from 6 to 12 DW.
The analysis of RNA-seq data additionally highlighted the expression in the human fetal testis of the genes
CNR1 and
CNR2, which encode the two main cannabinoid receptors, CB1 and CB2.
CNR1 expression was variable between 7 and 17 DW and overall higher than
CNR2, as confirmed by RT-qPCR. In 10 to 12 DW testis, in situ hybridization using RNAscope, a recognized highly specific and sensitive technique, demonstrated the expression of
CNR1 mainly in gonocytes and Leydig cells, and to a lower level in Sertoli cells. Small dots (each dot represent the detection of an RNA copy) corresponding to
CNR2 were occasionally observed in these three cell types. Biggest dots were observed in the case of
CNR1 meaning the amount of
CNR1 RNA is more important. CB1 and CB2 detection in the fetal testis by immunohistochemisty was not successful, which could be due to their relatively low expression levels combined with the high background staining generated by the available antibodies. Similarly, in the adult testis, immunohistochemistry of CB2 generated high background staining and only post-meiotic germ cells appeared strongly labelled [
43]. In our study, the localization of
CNR1 in Leydig cells and gonocytes indicate both these cell types could be sensitive to 2-AG, a highly effective agonist of CB1 [
53] mainly synthesized by fetal Leydig cells. This uncovers a potential new autocrine and paracrine signaling pathways between these two cell types. In addition, we detected several cannabinoid alternative receptors by RNA-seq analysis, which could also be involved in this signaling pathway (data not shown).
In pregnant women, THC, CBD, and their metabolites quickly cross the placenta barrier and are found in the umbilical cord [
5], cord blood, or infant urine and hair after delivery [
54]. The use of cannabis during pregnancy and breastfeeding can induce negative birth outcomes such as an increased risk of prematurity, a reduced birth weight and cognitive deficits in children [
7,
55,
56]. To date, its effect on the genital tract and on the secretion of reproductive hormones during development, required for the differentiation of male genital organs and ducts and for brain masculinization, has never been investigated [
8]. Despite the deleterious outcomes reported in the fetus, there are currently no data available regarding the exact concentrations of cannabis compounds that the human fetus may be exposed to. Moreover, the peak of plasma concentrations (Cmax) differs depending on the mode of administration of cannabis and the content of THC. THC Cmax after smoking a cannabis cigarette is around 10
−7 M to 7.4 × 10
−7 M for cigarettes containing between 19 and 69.4 mg of THC (Ohlsson et al. 1980; Kauert et al. 2007; Hunault et al. 2008), while CBD Cmax after smoking a cannabis cigarette are higher, ranging from 1.8 × 10
−5 M to 2.9 × 10
−5 M for 1.5 mg CBD ingested [
57]. The placenta transfer of cannabinoid components is considered to be very high [
58]. Thus studies performed on animal models showed that the concentration of THC can be equal between fetal and maternal plasma 3 h after THC administration and that CBD transfer to the fetus began 15 min after CBD administration with a rate of 66.9% [
11,
59]. Based on these data, we speculate that the concentrations of 10
−7 to 10
−5 M used in this study could reflect the range of concentrations of cannabis compounds that the fetus may be exposed to upon maternal consumption, with 10
−5 M at the high end.
Using our original and well characterized ex vivo model of human fetal testis [
24,
25,
28,
29], we evaluated the effects of the main components of cannabis, CBD, and THC on 8–12 DW testes. This model has already enabled us to identify the toxic effects of environmental compounds visible at the macroscopic level in short-term culture [
25,
60]. Here, we provide the first evidence that 72 h exposure of testis explants to phytocannabinoids induce a range of deleterious effect. While the morphology of the tissue was not affected by the treatment, even at the highest concentrations, and no massive toxicity was observed, the number of apoptotic cells in intracordonal compartments increased, fetal testicular cell proliferation decreased, and both testosterone secretion by Leydig cells and AMH secretion by Sertoli cells diminished. BRB-seq transcriptomic analysis of phytocannabinoid-exposed testis explants revealed 187 differentially expressed genes (DEGs) upon treatment, of which key genes for steroid synthesis and toxic substance response. The observed effects varied depending on the molecules (CBD or THC) and age of the tissue, probably in relation with the complex interactions between ligands and cannabinoid receptors [
14]. THC is described as a partial agonist with high affinity for both CB1 and CB2, while CBD is thought to be a negative allosteric modulator for CB1. CBD displays a weak inverse agonism for CB2 and may also act as an “indirect” CB1/CB2 agonist by increasing the level of 2-AG or by weakly inhibiting AEA enzymatic hydrolysis. CB1 and CB2 are both coupled to G proteins and activate a number of intracellular signaling pathways [
14].
As showed by our data, alteration of testosterone secretion was not due to a modification of Leydig cell density after 72 h, as it remained stable. Moreover, CBD alone did not induce a significant decrease of proliferating cell density in extracordonal tissue of 8–10 DW testes, nor did THC in 10–12 DW testes, although these compounds affected testosterone levels at these respective ages. On the one hand, the accentuation of the decrease in testosterone production as a function of the concentration at 3 days of exposure, and over time in the presence of CBD, without modification of the relative area of the CYP11A1-positive Leydig cells suggests that the steroidogenic function more than the nature and/or differentiation of these cells is altered by CBD. This is consistent with transcriptomic data that indicate an alteration of central lipid metabolism in CBD-exposed cells. The absence of major alteration of the Sertoli endocrine function suggests a specific action of CBD on lipid metabolism. The anti-androgenic effects of THC on the production of testosterone are more tenuous, depend on age, and are only visible in the 10–12 DW-exposed testes after 2 weeks of exposure. The fact that it is accompanied by a relative increase in the surface occupied by CYP11A1-positive Leydig cells without the surface of the cords being altered suggests that the proportion of Leydig cells within the interstitial tissue is modified. It is possible that Leydig cells proliferate or differentiate in compensation for the decrease in their ability to synthesize testosterone. These data suggest that the individual effect of CBD and THC on testosterone secretion occurred at the molecular level. Nevertheless, the increased effect of the THC/CBD mix on testosterone might be due to a combination of both molecular and cellular effect, since a decreased proliferation was observed upon exposure to CBD/THC in 8–10 and 10–12 DW testes. The alteration in testicular cell viability visible from 3 days of exposure to the mix at 10
−5 M culminates in an overall shrinkage of the tissue at 14 days. Total cord depletion after 14 days of exposure, targeting both Sertoli and germ cells, materializes from 3 days of exposure onwards with an increase in intracordonal apoptosis. The concentration-dependent decrease in endocrine function of testosterone at 3 days is concomitant to a decrease in the proliferation of interstitial cells. This suggests that the effects of CBD on cell metabolism, and of THC on the cell differentiation/function balance are additive to the detriment of tissue viability. In line with the hypothesis of a specific impact of cannabinoids on testosterone production by the fetal testis, anti-androgenic effect of CBD and THC alone have been demonstrated in embryonic rodent in vivo [
61‐
63]. Here we reveal that direct exposure of human fetal testis to CBD and/or THC is harmful to testosterone secretion, likely as a result of the alteration of steroid biosynthesis by Leydig cells. This hypothesis is supported by our transcriptomic analysis of testis explants after phytocannabinoid treatments, which evidenced a downregulated expression of numerous genes involved in steroids and cholesterol synthesis after a CBD/THC exposure, as it has been previously shown after phthalate exposure of human fetal testis [
64]. For instance, the expression of cytochrome P450 family 17 subfamily A member 1 (CYP17A1), a key testosterone biosynthesis enzyme, was significantly decreased after exposure of 10–12 DW testicular explants to CBD/THC mixture, while expression of enzymes like
HSD3B1/2 or
CYP11A1 were not altered. Our results are coherent with other studies that show the effect of cannabinoid components on steroid synthesis. In the rat, in vitro experiments showed that cannabinoids affect the secretion of steroids such as testosterone [
65‐
67], progesterone, or estradiol [
67]. In vivo, THC administration given during the third week of gestation in rat significantly blocked the surge of testosterone occurring in male fetus [
68].
In parallel, the expression of genes involved in the response to toxic substance, detoxification and cellular stress significantly increased. Cell metabolism alterations might be responsible for cell toxicity in longer-term culture, as suggested by the massive apoptosis observed in fetal testis exposed to CBD/THC for 14 days. As the majority of effects are observed in 10
−5 M CBD/THC condition, a non-specific toxic effect on testicular cords cannot be excluded. Nevertheless, the presence of ECS in human fetal testis, the absence of macroscopic signs of tissue toxicity, the absence of obvious changes in the number of testis cell types, and the deregulation of steroid synthesis pathway suggest that the observed effects could be due to a specific interaction of phytocannabinoids with the ECS. Of note, the overall absence of gene expression alteration with THC alone exposure could be due to the chosen BRB-seq fold change threshold impairing the detection of more discrete gene expression modification, or, alternatively, to post-translational mechanisms. Surprisingly also, while the mix of THC and CBD at 10
−5 M for 72 h impacted a similar number of genes in 8–10 DW and 10–12 DW testicular explants, CBD or THC alone led to very few variations of gene expression in the older testes. This might reflect an exacerbated window of sensitivity in the 8–10 DW testis (as observed for CBD specifically impacting testosterone production at this early age for instance), the targeting of distinct cell populations or their under-representation in older testes, and/or differences in cell metabolism between these two developmental ages. The exacerbated effects of the mixture when compared with the individual molecules may be explained by the interaction of cannabis compounds with cannabinoid receptors and the activation through CB1 and CB2 of signaling pathways involved in the regulation of apoptosis and cell proliferation and by an additive effect of CBD and THC [
14].
In addition to testosterone decrease, our results demonstrate an effect of phytocannabinoids on AMH secretion in short-term cultures, without any alteration of Sertoli cell density assessed by AMH staining. Of note, cell markers can still be detectable in apoptotic cells [
60]. Since the number of apoptotic cells increased in intracordonal tissue, we believe that the cannabinoid treatments affected the microenvironment necessary to the homeostasis of germ cell and Sertoli cell populations. The increased cell death in intracordonal tissue is consistent with the alteration of human adult Sertoli cell viability reported in another study after an in vitro exposure to CBD from 7.10
−6 M [
69]. While THC alone had no negative impact on Sertoli cell activity in another study, the highest dose tested (6.4 0.10
−7 M) was lower than the 10
−5 M used here [
70]. We investigated the expression of genes involved in AMH synthesizing pathways in testicular explants exposed with CBD and/or THC at 10
−5 M for 72 h by BRB-seq, but failed to observe any significant modification of gene expression. This may suggest a post-translational effect of cannabinoids on AMH secretion linked to a globally altered Sertoli cell metabolism, since an alteration of genes involved in cell metabolism was observed in BRB-seq experiments on whole testis explants. The alteration of Sertoli cells by CBD/THC mixture was confirmed after 14 days of treatment in our model through the disappearance of this somatic cell type and the significant decrease of AMH secretion. Globally, major deleterious effects were observed on cell viability and growth inside the cords, and on Leydig cell function outside the cords, consistent with the localization of CB1 receptors. The depletion of germ cells in long-term cultures could therefore be due to an indirect effect of the drugs on Sertoli cells and on testosterone production by Leydig cells and/or to a direct effect on this cell population.
The development of TGCT, which partly originates from a dysregulation of fetal germ cell differentiation [
71], has been associated with cannabis consumption in men [
19‐
21] but the mechanisms at play are unknown. The elevated incidence of TGCT these last decades is suspected to be linked to exposure of the fetal testis to a range of chemicals through maternal impregnation [
72], with subsequent environmental exposures during teenage or adulthood acting in some cases as a secondary trigger [
73]. Indeed, the differentiation of germ cells is mainly influenced by the hormonal secretions of Leydig and Sertoli cells, which were affected in our study by cannabis component exposure. Animal models and data in men point at a role of Leydig cell dysfunction and subsequent decreased testosterone level production in the etiology of TGCT [
74,
75]. The imbalance of the cellular microenvironment could potentially result in an impaired gonadal development, an arrest of gonocyte differentiation and the formation of germ cell neoplasia in situ (GNIS) that can promote neoplastic transformation later in life [
72]. Altogether, these elements suggest that phytocannabinoids might have a deleterious effect on germ cells during fetal life and represent secondary triggers for cancer when used in men after puberty. Further studies, for instance using single-cell RNA seq, are warranted to better apprehend the mechanisms underlying the effects of cannabis component on the genome of rare population of pluripotent germ cells in fetal human testis.