Amhr2-Cre-Mediated Global Tspo Knockout.

Although the role of translocator protein (TSPO) in cholesterol transport in steroid-synthesizing cells has been studied extensively, recent studies of TSPO genetic depletion have questioned its role. Amhr2-Cre mice have been used to generate Leydig cell-specific Tspo conditional knockout (cKO) mice. Using the same Cre line, we were unable to generate Tspo cKO mice possibly because of genetic linkage between Tspo and Amhr2 and coexpression of Amhr2-Cre and Tspo in early embryonic development. We found that Amhr2-Cre is expressed during preimplantation stages, resulting in global heterozygous mice (gHE; Amhr2-Cre+/-,Tspo -/+). Two gHE mice were crossed, generating Amhr2-Cre-mediated Tspo global knockout (gKO; Tspo -/-) mice. We found that 33.3% of blastocysts at E3.5 to E4.5 showed normal morphology, whereas 66.7% showed delayed development, which correlates with the expected Mendelian proportions of Tspo +/+ (25%), Tspo -/- (25%), and Tspo +/- (50%) genotypes from crossing 2 Tspo -/+ mice. Adult Tspo gKO mice exhibited disturbances in neutral lipid homeostasis and reduced intratesticular and circulating testosterone levels, but no change in circulating basal corticosterone levels. RNA-sequencing data from mouse adrenal glands and lungs revealed transcriptome changes in response to the loss of TSPO, including changes in several cholesterol-binding and transfer proteins. This study demonstrates that Amhr2-Cre can be used to produce Tspo gKO mice instead of cKO, and can serve as a new global "Cre deleter." Moreover, our results show that Tspo deletion causes delayed preimplantation embryonic development, alters neutral lipid storage and steroidogenesis, and leads to transcriptome changes that may reflect compensatory mechanisms in response to the loss of function of TSPO. © Endocrine Society 2020.

Translocator protein (TSPO) is an outer mitochondrial membrane protein with multiple biological functions, including direct or indirect roles in mitochondrial cholesterol transport and steroid hormone biosynthesis, porphyrin transport and heme synthesis, apoptosis, cell proliferation, and anion transport [1]. TSPO is expressed at high levels in steroid-synthesizing tissues and binds with high affinity to cholesterol and numerous other compounds [2]. The role of TSPO in steroid biosynthesis has been extensively investigated in studies of the function of TSPO in multiple protein complexes involved in mitochondrial cholesterol transport, as well as animal models of disease [3-7]. However, recently published studies of the genetic depletion of TSPO have produced conflicting findings regarding the role of TSPO in steroid biosynthesis, heme synthesis, and the function of the TSPO diagnostic ligand PK 11195 in steroid production [8-11].

To reproduce previous results obtained with Leydig cell-targeted Tspo conditional knockout (cKO) mice, showing a lack of effect on androgen production [8], we started with the same Amhr2-Cre (anti-Müllerian hormone receptor type 2-Cre) mice, but did not succeed in generating mice with the loss of the TSPO protein [12]. However, using a different Cre line (Nr5a1-Cre or Sf1-Cre), we generated steroidogenic cell-targeted Tspo cKO mice that showed a congenital adrenal hyperplasia-like phenotype with lipid accumulation accompanied by the lack of ability to produce corticosterone in response to adrenocorticotropic hormone (ACTH) [12]. This phenotype is similar to that of Star KO mice, which show an accumulation of lipid droplets (LDs) in the adrenal glands [13]. Recently, we used zinc finger nuclease technology to perform Tspo-targeted genome editing in the rat and generated 2 lines, a null mutant lacking TSPO expression and a line expressing a truncated TSPO protein lacking the fifth transmembrane domain, containing the cholesterol recognition amino acid consensus motif [14, 15]. Both Tspo mutations in rat models led to accumulation of esterified cholesterol in all steroidogenic cells examined and attenuated response to ACTH in terms of corticosteroid formation. Basal testosterone production was also reduced in Tspo homozygous (HO) mutant rats. In humans, the presence of the rs6971 polymorphism in the TSPO gene, a nonconservative amino acid substitution Ala147Thr, in the fifth transmembrane domain, results in altered binding affinity of TSPO for specific ligands and cholesterol, as well as reduction in ACTH-induced corticosteroid production [14, 16]. Both studies showed experimental evidence of the role of TSPO in steroidogenesis.

One unsolved question about the Amhr2-Cre–mediated Tspo cKO is that the data presented from the previous 2 reports are in contradiction with each other [8, 12]. One possible explanation is based on the small number of mice (Amhr2-Cre+/–, Tspofl/fl) with a wild-type (WT) phenotype but no conditional target gene deletion [12] because of a suggested strong genetic linkage between Amhr2 and Tspo genes [17]. However, the authors who first reported the Amhr2-Cre–mediated cKO type of Tspo [8] and later suggested the genetic linkage [17] failed to explain why there was no strong genetic linkage in their own animal study using the same Cre line, in which the same mice (Amhr2-Cre+/–, Tspofl/fl) had a WT phenotype, including no dramatic differences in steroid production [8]. We speculate the presence of, at least, some mixed genotypes in the mouse population in the original study [8].

In the present report, we elucidate in detail the discrepancy between the Morohaku et al paper [8] in which the Amhr2-Cre/LoxP system was used to generate a Leydig cell-specific Tspo cKO mouse, and our own work to replicate these findings [12]. We also present a new and fully characterized Amhr2-Cre–mediated Tspo deletion. At the same time we took the opportunity to address criticisms aimed at our work [12] by Selvaraj et al [17], using strong experimental evidence that the early expression of Amhr2 may hinder the generation of cKO mice. Indeed, the early expression of Amhr2 at 2- to 4-cell stages of preimplantation embryos was proven to be functional in Rosa-mT/mG+/– mice, leading to the production of Tspo global KO genotypes (Amhr2-Cre+/–, Tspo+/– and Amhr2-Cre+/–, Tspo–/–). Furthermore, we found that there were adverse effects of the Tspo deletion on preimplantation development; abnormal neutral lipid storage, testosterone formation, and transcriptional changes.

1. Material and Methods

A. Generation of Anti-Müllerian Hormone Type 2 Receptor-Cre–Mediated Global Tspo Knockout Mice

We previously generated Amhr2-Cre–mediated gonadal-specific Tspo cKO mice in which the resulting conditional homozygous (cHO) mice appeared to possess WT phenotypes and to represent an unexpectedly small proportion (4.4%) of the total number of pups examined [8, 12]. This is the same Cre-LoxP KO system used with 2 different genotyping/breeding regimens; we demonstrate in this report that the previously assumed cKO is a global knockout (gKO). To avoid confusion on the nomenclature of the mice used and generated, their corresponding genotypes are presented in Table 1. Considering the genetic linkage between Amhr2 and Tspo, as well as the early expression of Amhr2-Cre at 2- to 4-cell stages, we selected expected conditional heterozygous (cHE) mice with a WT-like genotype (Amhr2-Cre+/–, Tspo–/+) to screen via specific polymerase chain reaction (PCR) for the deleted exon 2/exon3 flanked by the 2 LoxP sites using the primer sets shown in Table 2: gKO-R/gKO-F to give 3818 base pairs (bp) for WT mice, 4092 bp for Tspo-floxed mice, 196 bp for global heterozygous mice (gHE) mice; and exon2-RR/exon2-FF to give 300 bp for WT (Tspo+/+) and global KO (gKO; Tspo+/– and Tspo–/–) mice. The generated global Tspo heterozygous KO (gHE: Amhr2-Cre+/–, Tspo–/+) mice were confirmed by sequencing the amplicons. To produce global Tspo HO KO (global homozygous [gHO]: Tspo–/–) mice, 2 gHE mice were crossed to obtain pups for further investigation. Genotyping the generated pups showed the expected 25% gHO, 25% WT, and 50% gHE ratios. Adult mice were weighed and tissues collected, weighed, and processed for further analysis. This study was carried out in accordance with the recommendations of the Canadian Council on Animal Care. The protocol was approved by the McGill University Animal Care Committee and the University of Southern California Institutional Animal Care and Use Committee.

Table 1.

List of mice used and generated, and their corresponding genotypes

Abbreviation	Full Name	Genotype
Amhr2-Cre	Amhr2-Cre–positive mouse (WT)	Amhr2-Cre+/–, Tspo+/+
WT	C57BL/6J mouse (WT)	Amhr2-Cre–/–, Tspo+/+
Tspo cKO	Amhr2-Cre–mediated Tspo conditional KO	Amhr2-Cre+/–, Tspo+/fl and Amhr2-Cre+/–, Tspofl/fl
WT (mT/mG)	Amhr2-Cre–/– reporter mouse	Amhr2-Cre –/–; Rosa-mmT/G+/–
WT (Amhr2-Cre/mG)	Amhr2-Cre+/– reporter mouse	Amhr2-Cre +/–; Rosa-mG+/–
Tspo gKO	Amhr2-Cre–mediated Tspo global KO	Amhr2-Cre+/–, Tspo+/– and Amhr2-Cre+/–, Tspo–/–
WT (Fl/Fl)	WT mouse with homozygous floxed Tspo alleles	Amhr2-Cre–/–, Tspofl/fl
WT (Fl/+)	WT mouse with heterozygous floxed Tspo alleles	Amhr2-Cre–/–, Tspo+/fl
cHE	Amhr2-Cre–positive, heterozygous floxed Tspo alleles	Amhr2-Cre+/–, Tspo+/fla or Amhr2-Cre+/–, Tspo+/fl
“WT-like” gHE	Amhr2-Cre–positive, genotyped nonfloxed Tspo alleles	Amhr2-Cre+/–, Tspo–/+
cHO	Amhr2-Cre–positive, homozygous floxed Tspo alleles	Amhr2-Cre+/–, Tspofla/fl and Amhr2-Cre+/–, Tspo–/fla
“cHO-like” gHE	Amhr2-Cre–positive, genotyped homozygous floxed Tspo alleles	Amhr2-Cre+/–, Tspo–/fla
Missing gHO	Amhr2-Cre–positive, genotyped cKO (nonreaction)	Amhr2-Cre+/–, Tspo–/–
gHO	Deleted both Tspo alleles	Tspo –/–
gHE	Deleted one Tspo allele	Tspo +/–
Cre reporter	Rosa-mT/mG	Tspo +/+, mT/mG+/–

Abbreviations: Amhr2-Cre, anti-Müllerian hormone type 2 receptor-Cre; cHE, conditional heterozygous; cKO, conditional knockout; Fl, floxed; gHE, global heterozygous; gHO, global homozygous; KO, knockout; WT, wild-type.

The floxed allele seems to be defective while it passes through the embryonic stages without target gene deletion.

Table 2.

Oligonucleotides used in this study

Primers	Sequence	Purpose
gKO-R	ACCCAGAGTTTGCCAATTGC	Global KO
gKO-F	ATCTCATTACGGGTGGTTGC	Global KO
Exon2-RR	TTGTAGAACTGCCCTCACCCCTACC	WT
Exon2-FF	ATTCCAGGGGCAACAGAGCACAGC	WT
WT-F(c)	CTTCCCTCGTGATCTGCAAC	mT/mG
MUT-R	CAATAGGGGGCGTACTTGG	mT/mG
WT-R	CCTCCCATTTTCCTTATTTGC	mT/mG
Abcb6-F	GTTAGCAATGGTGTCGTTGAAG	Real-time PCR
Abcb6-R	CTGGCTGCATCCGAATAGAT	Real-time PCR
Apoo-F	TTTGGGTGCTGCATAGACTC	Real-time PCR
Apoo-R	TGCCAGCGACATGTTCAA	Real-time PCR
Hprt-R	GCGTCGTGATTAGCGATGATGAAC	Reference gene
Hprt-F	GAGCAAGTCTTTCAGTCCTGTCCA	Real-time PCR
Nr4a2-F	CTGCTGGATATGTTGGGTATCA	Real-time PCR
Nr4a2-R	TCGATTCCAATCCGGCAAT	Real-time PCR
Trim5-F	TCTGACTGAACATTCTCCACATC	Real-time PCR
Trim5-R	GGCTGATGGCAGATAAGAAAGA	Real-time PCR
Trp53inp2-F	GGGTAACAAACCAGCTCTCATC	Real-time PCR
Trp53inp2-R	TGGCTCATCATCGACCTACA	Real-time PCR
Ucp1-F	TGGTCCCTAGGACACCTTTAT	Real-time PCR
Ucp1-R	CCTGGCAGATATCATCACCTTC	Real-time PCR
Ucp2-F	AAGCGGACCTTTACCACATC	Real-time PCR
Ucp2-R	GCCTCTACGACTCTGTCAAAC	Real-time PCR

Abbreviations: cHE, conditional heterozygous; cKO, conditional knockout; Fl, floxed; gHE, global heterozygous; gHO, global homozygous; KO, knockout; PCR, polymerase chain reaction; WT, wild-type.

Rosa-mT/mG+/– mice were purchased from the Jackson Laboratory (catalog No. 007576). Genotyping was performed using the following primer sets: WT-F(common): 5’-CTTCCCTCGTGATCTGCAAC-3’ with MUT-R: 5’-CAATAGGGGGCGTACTTGG-3’ (325 bp for positive mT/mG), and with WT-R: 5’-CCTCCCATTTTCCTTATTTGC-3’ (200 bp for negative mT/mG). Mouse blastocysts from Amhr2-Cre+/–; Rosa-mT/mG+/– mice were harvested between E3.5 and cultured overnight to E4.5. The day of vaginal plug detection was considered to be E0.5.

B. Immunofluorescence Staining and Microscopy

Immunofluorescence staining was performed as previously described [12]. Briefly, tissue cryosections or collected blastocysts at E3.5 were fixed in 4% paraformaldehyde, permeabilized with 0.1% triton X, blocked with 1% bovine serum albumin containing 1% donkey serum, and incubated in primary rabbit anti-TSPO monoclonal antibody (EPR5384) (ab109497; Abcam) [18] followed by secondary donkey antirabbit immunoglobulin G (H + L) antibody conjugated with Alexa Fluor 546 (Thermo Fisher) [19]. Nuclei were counterstained using UltraCruz aqueous mounting medium containing 4’,6-diamidino-2-phenylindole (DAPI; Santa Cruz Biotechnology Inc). Microscopy was performed using Olympus FV1000 and Zeiss LSM 780 confocal laser scanning microscopes and an inverted Olympus microscope for epifluorescence imaging. All images were analyzed using ImageJ software (National Institutes of Health).

C. Preimplantation Embryo Collection

After gHE (Tspo–/+) × gHE (Tspo–/+) mouse pairings, the presence of vaginal plugs was checked the next day, and plug-positive females were euthanized on embryonic day E3.5. Blastocysts were collected by flushing the uterus as previously described (http://www.cellmigration.org/resource/komouse/komouse_protocols.shtml). Blastocysts were photographed and cultured in M16 medium (Sigma-Aldrich) for an additional day to observe their development. Abnormal morphology was recorded as previously described [20].

D. Oil Red O Staining

Male and female tissues of interest were embedded in optimum cutting temperature medium, and 6 μm cryosections were prepared at the histology core facility of the Goodman Cancer Research Centre at McGill University. Frozen tissue sections were used for staining as recommended by the manufacturer (NovaUltra Oil Red O stain kit, IHC World). After staining, tissues were mounted with aqueous slide mounting medium for visualization. The staining of LDs were quantified using ImageJ software (NIH) following the procedures described previously [21].

E. Histochemistry and Transmission Electron Microscopy

For histochemistry testes were fixed in 4% paraformaldehyde. Paraffin-embedded testes from WT and Tspo gKO adult mice were prepared and sections stained with hematoxylin (Sigma-Aldrich) and eosin (Thermo Fisher Scientific) as previously described [22].

F. Transmission Electron Microscopy

For ultrastructural analyses, testis blocks were immersed in 5% glutaraldehyde overnight, post-fixed in cacodylate-buffered 1% osmium tetroxide, washed, dehydrated, and embedded in Epon resin [22]. Thereafter, samples were ultrathin-sectioned (~70 nm) and placed on grids. Electron microscopy was performed by the Doheny Eye Institute, using standardized procedures.

G. Blood Plasma Collection and Measurement of Circulating Steroids

We evaluated the activity of testes and adrenal glands in vivo by measuring levels of circulating testosterone and corticosterone, respectively. Blood was collected 6 days before euthanasia via submandibular puncture, and plasma was separated through centrifugation at 2000 × g for 15 minutes and stored at –80°C until used to determine steroid hormone levels. Enzyme-linked immunosorbent assay (ELISA)was used to measure testosterone and corticosterone (Cayman Chemicals) with intra-assay coefficients of variation (CV) of 8.8% and 6.2% respectively, as per the manufacturer’s instructions. Absorbance was read at 420 nm using a VICTOR X5 Multilabel Plate Reader (PerkinElmer Inc).

H. Testicular Interstitial Fluid Collection

Using a 27-gauge needle, the tunica of the testes was punctured 3 to 4 times opposite the rete testis and centrifuged in a 1-mL pipet tip at 500 × g for 10 minutes. Interstitial fluid collected in the bottom of the tube was diluted in ELISA buffer (Cayman Chemicals) for testosterone measurement by ELISA.

I. Interstitial Leydig Cell Preparation

Interstitial cells were isolated from adult WT and Tspo gKO mice ranging in age from 56 to 97 days. Mice were asphyxiated with CO2 and testes were removed. The tunica of the testis was removed, and seminiferous tubules were separated from interstitial cells by digestion (10 minutes, 34°C) in 0.25 mg/mL collagenase (Aldrich-Sigma)/13 mg/mL DNase (Aldrich-Sigma) in M-199 (Thermo Fisher Scientific) followed by unit-gravity sedimentation in M-199 supplemented with 1% bovine serum albumin [23]. Histochemical staining with nitro blue tetrazolium for 3βHSD activity with 0.4 mM etiocholanolone (Steraloids) as the steroid substrate [24] indicated the presence of an enriched Leydig cell population, containing 20% to 25% 3βHSD-positive cells. Cells were incubated at a concentration of 2.33 × 105 in a final volume of 1 mL for 2 hours in a shaking water bath at 34°C. Cells were incubated in buffered Dulbecco MEM:Ham F12 culture medium (Aldrich-Sigma) alone, or with a stimulating dose of 50 ng/mL human chorionic gonadotropin (hCG) (National Hormone Pituitary Program). Testosterone production was assayed by ELISA as described above.

J. Microarray Analysis and Data Mining

To analyze Amhr2 and Tspo gene expression profiles during the preimplantation period of mouse embryo development, we retrieved the relevant raw data sets from gene arrays performed on different preimplantation stages of the mouse embryos (NCBI GEO: GSE18290 and GSE41358) [25, 26]. Data were further processed using FlexArray software (version 1.61; http://genomequebec.mcgill.ca/FlexArray) and normalized by robust multiarray average. The significance analysis of microarrays and analysis of variance (ANOVA) were used to select statistically significant altered genes with a P value equal to or less than .05. Volcano plots of mRNA expression from preimplantation stages were generated using FlexArray data. The data plotted along the x-axis were the means of log2-fold changes and along the y-axis were the negative logarithm of the P values.

K. RNA-Sequencing Data Analysis

Raw RNA-sequencing (RNA-seq) data for 8-cell and morula stages were retrieved from NCBI GEO (GSE44183) [27], and raw RNA-seq data for adrenal glands and lungs were retrieved from the National Center for Biotechnology Information Sequence Read Archive (SRP043599) and Gene Expression Omnibus (GSE84942), respectively [9, 28]. Data analysis was performed using an RNA-seq data analysis pipeline in the Galaxy platform (https://usegalaxy.org/) [29], following the instruction for the RNA-seq analysis pipeline. In brief, the original files with FASTQ format were loaded into the server and then aligned to the mouse genome (NCBI37/mm9) using Bowtie for Illumina, and graphing and display were performed using the UCSC Genome Browser (http://genome.ucsc.edu/) [30]. Data quantification and visualization were performed using SeqMonk built-in visualization tools (http://www.bioinformatics.babraham.ac.uk/projects/seqmonk/), including bp quantitation, and each gene was split into exons and normalized to a larger database from either the WT or HO data sets. At the same time, DESeq2 was used to verify the transcriptome changes after Tspo gene deletion. Mitochondrial proteins were retrieved from MitoCarta [31]. Cholesterol-binding proteins (411 proteins) that were selective for a trans-sterol probe over an N-palmitoylethanolamine structure (N-palmitoylethanolamine-diazirine-alkyne), but not sensitive to cholesterol competition, were retrieved from group III from a previously published proteome-wide mapping study [32]. Protein-protein interaction networks were retrieved from the STRING database (http://string-db.org) to provide information on global transcriptional response under Tspo deletion and were visualized using Cytoscape Network Analyzer (www.cytoscape.org).

L. Real-Time Polymerase Chain Reaction for Validation of RNA-Sequencing Data Analysis

Total RNA from mouse adrenal glands and lungs was extracted using TRIzol Reagent (Thermo Fisher Scientific). The values of 260:280 ratio and 260:230 ratio were 1.8 to 2.0 and 2.0 to 2.2, respectively, determined by a NanoDrop spectrophotometer and deemed acceptable for further analysis. All RNA samples were treated with DNA-free DNA Removal Kit (Thermo Fisher Scientific). RNA was subsequently diluted to 100 ng/μL using DNase/RNase-free water, and first-strand complementary DNA (cDNA) was then synthesized using the Transcriptor First-Strand cDNA Synthesis Kit (Roche Applied Science), according to the manufacturer’s instructions with gene-specific primers (Table 1). The resulting cDNA samples were diluted with nuclease-free water and subjected to real-time PCR using SYBR green dye and a LightCycler 480 system (Roche Applied Science) as previously described [33]. The results reported for each RNA product were normalized to hypoxanthine-guanine phosphoribosyltransferase (Hprt) to correct for differences in the amounts of the template cDNA. The expression levels of each gene were further visualized within the whole RNA-seq data analysis in an MA-plot for reciprocal validation of the expression trend from the generated Tspo gKO mice.

M. Statistical Analysis

Statistical analyses were performed using GraphPad Prism 5.02 and 7.05 software. The significance of the results was determined using the Student t test or one-way ANOVA followed by the Bonferroni post hoc test for multiple comparisons. Chi-square test was performed using GraphPad QuickCalcs (http://graphpad.com/quickcalcs/chisquared1.cfm).

3. Results

A. Genetic Linkage Between Tspo and Anti-Müllerian Hormone Type 2 Receptor-Cre

We previously reported an unexpectedly small number of mice with a conditional homozygous floxed Tspo genotype (cHO: Amhr2-Cre+/–, Tspofl/fl), but with a “WT-like” phenotype, where immunofluorescence staining of the testis showed normal TSPO expression [12]. These findings suggested it would be difficult to generate Tspo cKO mice using the Amhr2-Cre line, which is in contrast to the first report of a Tspo Leydig and Sertoli cell-specific cKO by Morohaku et al. [8]. A likely cause of the inconsistency may be due to the genetic linkage of Tspo and Amhr2-Cre genes on chromosome 15, which could hinder the generation of a cKO animal. Further examination of the issue showed that there is only a small chance of crossover-producing recombination between the Tspo and Amhr2-Cre genes, which depends on the distance between the 2 genes. Based on the genetic distance between Tspo and Amhr2, that is, 18.18 cM (1 cM = 0.01 recombinant rate) from the genetic map of Mouse Genome Informatics (http://www.informatics.jax.org/) [34] and using Haldane’s map function and a controversial corrected Haldane’s map function [35-37], there is approximately only a 7.62% to 9.0% (at least 7.62%) chance of cHO pups resulting from the pairing of a cHE mouse (Amhr2-Cre+/–, Tspo+/fl*) with a floxed Tspo-homozygous WT mouse (cHO: Amhr2-Cre–/–, Tspofl/fl) (Fig. 1A). Because the small percentage of mice (4.4%) with cHO genotypes all have a WT-like phenotype [12], we assumed that either the Cre or the floxed allele(s) has a problem. Reanalysis of previously published microarray data on preimplantation embryos suggested that Amhr2 is expressed as early as the 2-cell stage, suggesting that Tspo could be deleted at preimplantation embryonic stage(s) (Fig. 1B) [12]. Given that the cHE mouse used for the crossing carries a floxed allele (Fl*) of Tspo that passes through those early embryonic stages, but without targeting gene deletion, the Fl* would be a WT allele that could either be deleted in a tissue-specific manner, indicating the cHO genotypes are WT animals, or result in Tspo gHE mice (Amhr2-Cre+/–, Tspo–/fl*) if it carries one Fl* allele even after introducing a new functional floxed allele (Fig. 1A). This is the only way we can interpret why we could not replicate the paper by Morohaku and colleagues [8] and explain why the generated cHO mice exhibit “WT” phenotypes and no tissue-specific deletion of TSPO. All the redrawing of the breeding regimen is based on the fact that Amhr2 is expressed during preimplantation embryo development as shown in microarray data reanalysis [12] and RNA-seq reanalysis (Fig.1B). Therefore, the Amhr2-Cre could not be used to produce a testis-specific Tspo cHO mouse, per se, as proved previously [12].

Figure 1.

Genetic linkage between Tspo and anti-Müllerian hormone type 2 receptor-Cre (Amhr2-Cre) on chromosome 15 and Amhr2-mediated Tspo global knockout (gKO) (Amhr2-Cre+/–, Tspo–/+) and/or (Amhr2-Cre+/–, Tspo–/fl*). A, Breeding scheme originally used in our previous experiments [12], expected chromosomal crossover (8.3%), and the likelihood of Tspo gKO leading to a genotype switch. Some conditional heterozygous (cHE) genotypes switched to a “wild-type (WT)-like” with one allele globally deleted (global heterozygous [gHE]), and some conditional homozygous (cHO) genotypes remained HO genotypes but with one allele globally deleted (“cHO-like” gHE); therefore, the cHO genotype includes both cHO and gHE. Fl* is a “defective” floxed Tspo allele, because it passes by the embryo stage without targeting a deletion, leading to a cHO* that has actually a WT allele. cHE and cHO both contain a WT Tspo allele coming either from Fl* or newly introduced, indicated by a light-color dotted arrow back to the original cHE parents. All numbers indicated as percentages are expected proportions except the blue numbers: 34.2% of cHE, 4.4% cHO, and 2.9% value of “WT-like” gHE, which were observed from our crossing experiments for the conditional knockout (cKO) breeding scheme. Haldane’s map function is shown, and the recombination frequency between Amhr2 and Tspo was estimated to be no more than 7.62%. *Indicates the “defect floxed allele” during cKO genotyping where the Amhr2-Cre does not work at the early embryo development, so the cHO* actually has a WT allele of Tspo all the time with a final mixed genotyping of cHO and gHE. In this report, the 2 gHE mice were crossed to produce a gHO mouse analyzed in Fig. 3. Percentages of the observed ratios are indicated in blue. **The recombination rate was estimated based on Haldane’s and Casares’ map function. Abbreviations: r, recombination rate; d, genetic distance in Morgans. ***Newly introduced “defect floxed allele.” B, Amhr2 is expressed at 8-cell stage of preimplantation embryo. RNA-sequencing data were retrieved from NCBI GEO (GSE44183) [27] and reanalyzed using the Galaxy platform [29], then visualized using SeqMonk built-in visualization tools and in the UCSC Genome Bowser (https://genome.ucsc.edu/) [30]. Each exon is shown as in an MA-plot for the whole set of sequences (right) and also in the Genome Browser sequence alignments (left) from exon(s) 1 to 11 (left).

Figure 3.

Anti-Müllerian hormone type 2 receptor-Cre (Amhr2-Cre)–mediated global Tspo deletion (Tspo–/–) and timing of Amhr2 and Tspo expression during embryonic development. A, Amhr2-Cre–mediated global Tspo knockout (KO) confirmed by polymerase chain reaction genotyping and sequencing of mouse tail snips. Approximately 2.9% of mice used in the experiments had a “wild-type (WT)-like” gHE (Amhr2-Cre+/–, Tspo–/+), indicating that Tspo was deleted globally in Cre-positive mice. Red arrows in the agarose gel used for genotyping: indels, from left to right, show the rare unexpected genotype, regenotyping of the last selected conditional homozygous (cHO) using gene deletion–specific primer to identify global heterozygous (gHE), and global homozygous (gHO) (Tspo–/–) produced by gHE × gHE crossings. The band sizes used for identification of WT and KO are indicated. M: 100 bp ladder. Primers used are indicated in the diagram below sequencing chromatography. B, Volcano plot of expression values between the 2-cell and blastocyst stages, with Tspo and Tspo2 shown as black dots. C, Expression profiles of Amhr2 and Tspo were obtained from 3 probes for each gene during preimplantation embryonic development of in vitro fertilized mouse embryos. Significant changes between stages are highlighted in red (increased) and green (decreased). The dotted orange circle highlights the increased Tspo expression at blastocyst stage. D, Immunofluorescence staining of translocator protein (TSPO) in mouse blastocyst at E3.5. Representative images show that localization of TSPO was in the inner cell mass (ICM) (indicated by arrow) with anti-TSPO antibody (red for confocal imaging, yellow for epifluoresence). Nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI) (blue). Differential interference contrast images were used as controls. Top panels: laser scanning confocal microscopy images; low panels: epifluorescence microscopy images. Scale bar represents 10 µm.

To provide experimental evidence on Amhr2-Cre activity at earlier preimplantation embryos, we investigated mice with a Rosa-mT/mG background (Fig.2A). Rosa-mT/mG mice express a double-fluorescent Cre-reporter in which the original expression of mT would be switched to that of mG with Cre activity [38]. At day 3 (E0.3.5) (or even at earlier stages) of preimplantation embryo development, Amhr2-Cre mediate recombination, where the CRE recombinase deletes a constitutively expressed mT transgene flanked by loxP sites (Fig. 2B). The mT is still switched into mG in certain areas, especially in the inner cell mass, of Amhr2-Cre+/–; Rosa-mT/mG+/– blastocysts (Fig. 2C). In vitro cultured E4.5 embryos continue irreversibly to express the mG in the descendent cells with the recombined genotypes of half trophectoderm, in addition to the inner cell, in the embryos (Fig. 2C). This could be due to the asymmetric cell division for the first cell fate decision during preimplantation embryonic development [39]. The Cre-reporter was actively expressed primarily in the central part of the inner cell mass, indicating that the gene knockout occurs early in the primordial embryo. These conclusions were further confirmed by epifluorescence microscopy (Fig.2E and 2F).

Figure 2.

The specificity and efficiency of anti-Müllerian hormone type 2 receptor-Cre (Amhr2-Cre)–mediated recombination were determined using Rosa-mT/mG(+/–) reporter mice. A, The E3.5 whole blastocysts were from Rosa-mT/mG+/+ mice. The images from 3 channels: red (mT, Tomato red), green (mG, membrane GFP), and differential interference contrast (DIC) were collected and overlaid using an Olympus FLUOVIEW FV1000 confocal laser scanning microscope. B, Diagram of the strategy used to assess the degree of preimplantation embryo-specific recombination. Amhr2-Cre mice were crossed to Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo reporter mice. These reporter mice ubiquitously express mT, A, before Cre activation and C and D, mG on Cre activation, as evaluated by confocal laser scanning microscope. C, Whole blastocysts view of E3.5 from Amhr2-Cre(+/–); Rosa-mT/mG(+/–) mice with dual detection of mT (red) and mG (green) fluorescence. D, View of the same blastocysts at E4.5 Amhr2-Cre+/–; Rosa-mT/mG+/– mice after in vitro culture overnight. Arrow, inner cell mass (ICM) where the recombination occurs. Scale bars = 20 µm. E, Epifluorescence view of whole blastocysts at E4.5 from Rosa-mT/mG+/+ mice after in vitro culture overnight (as shown in A). F, Epifluorescence view of whole blastocysts at E4.5 from Amhr2-Cre+/–; Rosa-mT/mG+/– mice after in vitro culture overnight (as shown in D). Scale bar, 50 µm.

Thus, the Amhr2-Cre+/– mouse serves as a “global deleter” instead of conditional, which is consistent with our findings from bioinformatic analysis of Amhr2 gene expression at preimplantation embryos. This finding suggests it may be necessary to reinterpret any previous data on Amhr2-Cre–conditional KO gene deletion, including the Tspo gene.

B. Anti-Müllerian Hormone Type 2 Receptor-Cre–Mediated Global Tspo Deletion

As shown in Fig. 1, mice of “WT-like” gHE genotype (Amhr2-Cre+/–, Tspo–/+) from the expected cHE (Amhr2-Cre+/–, Tspo+/fl*) were used to cross with WT (C57BL/6J) animals to produce gHE mice (Tspo–/+); crossing of 2 gHE (Tspo–+) will produce gHO (Tspo–/–; gKO) mice (Fig. 3A shows PCR genotyping and sequencing data demonstrating the presence of [left panel] “WT-like” gHE [Amhr2-Cre+/–, Tspo–/+] , cHO mice [Tspo+/+, Tspo–/+] from gHE crossed with [middle panel] WT [C57BL/6J] mice, and [right panel]gHO [gKO] generated from the crossing of 2 gHE (Tspo–/+) mice). Global deletion of the Tspo gene is confirmed by sequencing the targeted region of the Tspo locus and PCR using mouse tail snips and the primer sets: gKO-F/gKO-R in gKO (Tspo–/+ and/or Tspo–/–) mice after exon 2 and exon 3 deletion (196 bp) and exon2-RR/exon2-FF in WT (Tspo+/+) mice (300 bp).

To further examine the reason behind this observation, we reanalyzed previously published microarray data sets (GSE18290) of mouse embryo Tspo expression profiles at 1-cell, 2-cell, 4-cell, 8-cell, morula, and blastocyst stages [12, 26]. We found that Amhr2 was predominately expressed during the 2- to 8-cell stages, whereas Tspo was predominately expressed during morula and blastocyst stages (1-way ANOVA, F(5,12) = 5.588, P < .01, n = 3). Reanalysis of another data set (GSE41358) of mouse preimplantation in vitro fertilized embryos showed that Amhr2 expression decreased during the blastocyst stage, whereas Tspo expression increased during the blastocyst stage (one-way ANOVA, F(4,10) = 25.98, P < 0.001, n = 3; 3 probes per gene; Fig. 3B and 3C) [25]. Further immunofluorescence staining of TSPO in blastocysts at E3.5 indicates that TSPO is strongly expressed at this stage (Fig. 3D). These results provide a theoretical basis to support the Amhr2-Cre–mediated generation of a Tspo gKO but not cKO, and show that TSPO plays a role in normal mouse preimplantation embryo development.

C. Confirmation of Anti-Müllerian Hormone Type 2 Receptor-Cre–Mediated Tspo Global Knockout

TSPO expression in adult WT (Tspo+/+), gHE (Tspo+/–), and gHO (Tspo–/–), mice were investigated by indirect immunofluorescence staining and confocal microscopy. Strong TSPO staining was observed in Leydig cells in WT mice, whereas TSPO staining was weaker in gHE mice and not observed in gHO mice (Fig. 4A-4F). Similar expression patterns were observed using both confocal and epifluorescence microscopy. In adrenal glands, TSPO was strongly expressed in the cortex area of WT mice, whereas it was weakly expressed in the cortex area of gHE mice and not expressed in the cortex area of gHO mice (Fig. 4G-4L). Moreover, strong staining was observed in the fat tissue around WT adrenal glands, but no staining was observed in the fat tissue around gHO adrenal glands.

Figure 4.

Immunofluorescence staining of translocator protein (TSPO) in the testis and adrenal gland of wild-type (WT) (Tspo+/+), global heterozygous (gHE) (Tspo+/–), and global homozygous gHO (Tspo–/–) mice. A to C, Representative confocal fluorescence images of TSPO staining (anti-TSPO antibodies [Ab], red) and nuclei counterstaining (4’,6-diamidino-2-phenylindole [DAPI], blue) and D to F, epifluorescence microscopy images of TSPO staining (anti-TSPO Ab, orange) in the testis. Representative G to I, confocal fluorescence images of TSPO staining (anti-TSPO Ab, red) and nuclei counterstaining (DAPI, blue) and J to L, epifluorescence microscopy images of TSPO staining (anti-TSPO Ab, orange) in the adrenal gland. Abbreviations: C, cortex; M, medulla; Lc, Leydig cell. Scale bar, 200 μm.

Immunofluorescence staining of other tissues showed that TSPO was absent in gHO mice and weakly present in gHE mice compared with WT mice based on both confocal and epifluorescence microscopy. In the WT mouse brain, TSPO immunostaining was seen as sparsely distributed spots (Fig. 5A, 5B, 5D, 5E, 5G, and 5H), which are likely to be localized in microglia and endothelial cells as shown in several previous reports [12, 40-43]. In the gHO brain, there was no expression of TSPO, indicating that the staining in brain is TSPO specific (Fig. 5C, 5F, and 5I). In addition, TSPO was completely absent in the liver of gHO mice, but was weakly expressed in gHE mice and highly expressed in WT mice (Fig. 5J-5O). Taken together, we believe these results suggest that Tspo was globally deleted as we expected from genotyping of the mice, microarray data analysis, and the cell membrane-targeted, 2-color fluorescent Cre-reporter (Rosa mT/mG) analysis.

Figure 5.

Immunofluorescence staining of translocator protein (TSPO) in the brain and liver of wild-type (WT) (Tspo+/+), global heterozygous (gHE) (Tspo+/–), and global homozygous (gHO (Tspo–/–) mice. A to C, Representative confocal fluorescence images of TSPO staining (anti-TSPO antibodies [Ab], red) and nuclei counterstaining (4’,6-diamidino-2-phenylindole [DAPI], blue) within the D to F, highlighted areas and G to I, epifluorescence microscopy images of TSPO staining (anti-TSPO Ab, orange) in the brain. Representative J to L, confocal fluorescence images of TSPO staining (anti-TSPO Ab, red) and nuclei counterstaining (DAPI, blue) and M to O, epifluorescence microscopy images of TSPO staining (anti-TSPO Ab, orange) in the liver. Scale bar, 200 μm.

D. Adverse Effect of Global Tspo Deletion on Preimplantation Embryo Development

We previously reported that there were no morphological changes at E12.5 of “gKO” mice but observed lower than expected proportions of certain genotypes at birth [12]. Thus, we performed a closer examination of preimplantation embryonic stages to detect potential abnormalities in development. Indeed, we found that 33.3% of blastocysts at E3.5 to 4.5 had a normal morphology (Fig. 6A), whereas 66.7% showed delayed development (Fig. 6B), which are close to the expected proportions of 25% of WT (Tspo+/+) and 75% of HE and HO Tspo–/– 25%; Tspo+/– 50% mice. These results are in contrast to 95% to 99% of WT (Tspo+/+) blastocysts obtained from mice produced from the pairing of Tspo+/+; Amhr2-Cre+/– with Tspo+/+; Rosa-mT/mG+/+ mice, which showed normal development in vitro (Fig. 6C) [44]. These results are consistent with a previous report of low TSPO expression in fragmented or apoptotic preimplantation embryos with associated lipid accumulation [20], as well as with our gene expression profiling showing increased Tspo expression during blastocyst development (Fig. 3B, 3C, and 3D). It is clear then that TSPO is needed for normal preimplantation embryo development.

Figure 6.

Morphology of blastocysts from anti-Müllerian hormone type 2 receptor-Cre (Amhr2-Cre)–mediated Tspo global heterozygous (gHE) (Tspo+/–) × gHE (Tspo+/–) pairings. Representative images of preimplantation embryos at E3.5 (in vivo, n = 30 checked in total; upper panel) and E4.5 (in vitro cultured overnight, n = 30 checked in total; lower panel). A, One-third had a normal morphology, whereas B, two-thirds showed delayed development at E0.3.5. C, Morphology of blastocysts from wild-type pairings (Tspo+/+; Amhr2-Cre+/– with Tspo+/+;Rosa-mT/mG+/+). A representative blastocyst is shown with Tspo+/+; Amhr2-Cre+/–; Rosa-mG+/– genotype, where 95% have a normal morphology at E0.3.5 (n = 21). Scale bars = 50 µm.

E. Changes in Neutral Lipid Storage After Global TSPO Depletion

Neutral lipids, including dominant esterified cholesterol in steroidogenic and/or TSPO-rich tissues, are the source of cholesterol for steroid biosynthesis [45]. We evaluated the presence of LDs enriched with esterified cholesterol from these tissues, including the testis, adrenal gland, liver, and ovary. After TSPO depletion, we found decreased neutral lipids in the testis (Fig. 7A-7C) and increased neutral lipids in the adrenal gland (Fig.7D-7F), liver (Fig. 7G-7I), and ovary (Fig. 7J-7L). Similar patterns of lipid accumulation in the testis, adrenal gland, and ovary were previously observed in Nr5a1-Cre–mediated Tspo cKO mice in a tissue-specific manner and were accompanied by a lack of corticosterone production in response to ACTH [12]. The accumulation of neutral lipids is a sign of changes in the use of cholesterol as the substrate of steroid biosynthesis, while in parallel preventing the accumulation of excess toxic free cholesterol, in agreement with previous reports for either Tspo [12, 14] or Star [13].

Figure 7.

Comparison of neutral lipid storage in testis, adrenal gland, liver, and ovary after anti-Müllerian hormone type 2 receptor-Cre (Amhre-Cre)–mediated translocator protein (TSPO) depletion. Representative images of Oil Red O staining showing lipid droplets (LDs) in the WT (Tspo+/+) vs global homozygous (gHO) (Tspo–/–) A to C, testis; D to F, adrenal gland; G to I, liver; and J-L, ovary. C, F, I, and L, quantification of the staining of LDs from 4 to 5 individual animals. Student t test (testis: P = .05; adrenal gland: P = .05; liver: P = .01; ovary: P = .03). Data are presented as means ± SEM; n = 6 animals per group. *P less than .05. Scale bar, 100 µm. White arrow highlights LD-stained areas.

F. Steroid Biosynthesis After Global TSPO Depletion

We measured circulating testosterone and corticosterone levels to assess the steroidogenic capacity of the testes and adrenal glands of Tspo gKO mice. We found that Tspo gHE and Tspo gHO mice showed significantly lower testosterone levels than WT mice (P < .05; Fig. 8A), whereas there was no difference in basal corticosterone levels between WT, gHE, and gHO mice (Fig. 8B), in agreement with previous results on a mouse cKO [12] as well as the recent data on Tspo-mutant rat lines [14]. The constant or changed steroid levels in these steroidogenic tissues may reflect the physical condition(s) of mitochondria under TSPO deficiency, as well as the availability of free cholesterol from different sources. Indeed, indirect evidence of accumulation of lipids in adrenal glands could reflect the upregulated Aup1 gene (Fig. 8C), which is an LD regulating VLDL assembly factor; high expression of this gene results in increased LD clustering [46, 47].

Figure 8.

Circulating A, testosterone and B, corticosterone levels in the plasma of wild-type (WT) (Tspo+/+), global heterozygous (gHE) (Tspo+/–), and global homozygous (gHO) (Tspo–/–) mice. *P less than .05, Mann-Whitney U tests (data presented as means ± SEM; n = 8-12 animals per group). C, Elevated gene expression of ancient ubiquitous protein 1 (Aup1), a lipid droplet regulating VLDL assembly factor, in mouse adrenal glands (WT vs global knockout). Blue and green spots represent each exon of the cholesterol-binding genes. Red spots represent exons of the Aup1 gene.

To further assess the impact of TSPO deficiency on testosterone production, we measured intratesticular testosterone levels in WT and Tspo gKO mice. Data obtained are presented as percentage of WT. Considering that we observed a slight but significant increase in Tspo gKO testis weight (Fig. S1) [48], the data were corrected per gram of testis weight. A significant 53% reduction in intratesticular testosterone levels was seen in Tspo gKO mice (P < .05; Fig. 9 left). These results were consistent with results obtained using enriched interstitial Leydig cell preparations isolated from WT and gKO mice (Fig. 9). Cells isolated from WT animals showed a 3-fold stimulation of testosterone production in response to saturating concentrations of hCG, whereas cells isolated from Tspo gKO mice failed to respond to hCG. No differences were observed in seminal vesicle, epididymis, kidney, and whole-body weights (Fig. S1) [48]. Histochemical analysis of testis sections failed to show any major morphological differences in testes from WT and Tspo gKO mice (Fig. S1) [48]. Transmission electron microscopy studies were undertaken to assess Leydig cell ultrastructure in WT and Tspo gKO mice. No significant differences were seen in Leydig cell morphology between WT and Tspo gKO mice (Fig. 10A-10D), except that Leydig cells from Tspo gKO mice contained autophagosomes that we rarely observed in WT mice (Fig. 10E-10F).

Figure 9.

Left panel. Intratesticular testosterone levels in wild-type (WT) (Tspo+/+) and global homozygous (gHO) (Tspo–/–) mice. Data are presented as means ± SEM; n = 13 to 14 animals per group; *P less than .05. Intratesticular testosterone levels in WT animals were on average 50 ng/g per testis. Right panel. Testosterone formation by isolated Leydig cells in response to 50 ng/mL human chorionic gonadotropin (hCG). Results shown are from n = 4 (WT), n = 7 (global knockout [gKO] control), and n = 5 (gKO hCG) animals. Data shown are the mean ± SEM. **, P less than .01 compared to controls. Abbreviation: ns, nonsignificant.

Figure 10.

Transmission electron microscopy images showing the ultrastructure of A and B, wild-type (WT) (Tspo+/+) and C and D, global knockout (gKO) (global homozygous [gHO], Tspo–/–) Leydig cells. E and F show autophagosomes identified in gKO (gHO, Tspo–/–) Leydig cells. Abbreviations: LDs, lipid droplets; M, mitochondria; N, nucleus; W, swirled variety of smooth endoplasmic reticulum.

G. TSPO Depletion Leads to Global Transcriptional Changes Related to Protein Relocalization, Mitochondrial Ion Transport, and Proton Leak Pathways

Because TSPO is needed for the development of preimplantation embryos and adult steroid biosynthesis, the inconsistent results on the roles of TSPO from previous reports [8, 12, 14, 28] might involve a compensatory mechanism, which is likely to be established in the early preimplantation period of embryogenesis under TSPO deficiency, when survival selection for cell fate decisions has begun via genetic network rewiring. Indeed, 2 previous reports propose that compensation occurs after deleterious mutation via gene or network replacement, but that it does not occur after a gene knockdown [49, 50]. Therefore, we sought to identify the mechanism behind this possible compensation to ultimately increase our understanding of steroidogenesis in animals.

We reanalyzed 2 sets of raw RNA-seq data associated with studies showing that TSPO depletion did not affect gene expression profiles [9, 28]. Tspo-deleted and nondeleted exons of the gene were used as negative and positive controls, respectively, and the relevant graphs were visualized using UCSC Genome Browser under custom tracks. By viewing scatter plots of WT vs Tspo deletion (gHO), we selected genes that were uniquely upregulated (red) or downregulated (green) in the adrenal gland (Fig.11A) and lung (Fig. 11B; Tables S1-S4) [48], with the dashed circle (purple) indicating genes that were specifically downregulated in the lung (Fig. 11B; Table S5) [48]. Corresponding loci for the selected genes, including Tspo, Ucp2, Fdxr, Krt78, Krt5, and Caps4, are shown to indicate the differential sequence alignment(s) (Figs. S3, S4, and S9-S11) [48]. Additionally, the clustering of gene expression profiles and the upregulated and downregulated gene networks likely indicate functional compensation after TSPO depletion (Figs. S5-S7 and S11-S13) [48].

Figure 11.

Whole-genome transcript profiling to identify differentially expressed genes associated with translocator protein (TSPO) depletion in mouse adrenal glands and lungs: reanalysis of 2 previously published sets of RNA-sequencing (RNA-seq) data [9, 28]. Scatter plot of RNA-seq data from A, adrenal glands and B, lungs. Red dots: high expression, green dots: low expression (including Tspo as a positive [exon 4] or negative [exons 2/3] control). Scatter plots for C, adrenal glands and D, lungs showing 4 major players in steroidogenesis: Cyp11a1, Star, Vdac1, and Tspo. Red dotted circle: high expression, green dotted circle: low expression, blue dotted circle: no change in expression. Tspo showed low expression only in the nondeleted exon 4 (positive control), whereas the deleted exons 2/3 are far away from the diagonal line (negative control). Scatter plots of mitochondrial cholesterol-binding protein-encoding genes in E, adrenal glands and F, lungs with 2 value differences of total base pair read counts in log2 scale. Mitochondrial cholesterol-binding proteins were retrieved from MitoCarta and a previous publication [31, 32]. Red: high expression, green, low expression. Two highlighted genes, Apoo and Slc25a13, likely play a role in compensating for TSPO depletion.

Of the 4 major players in steroidogenesis, Cyp11a1, Star, Vdac1, and Tspo, Star had slightly lower expression and Cyp11a1 had higher expression in gHO adrenal glands compared with WT adrenal glands (Fig. 11C), in agreement with previous real-time PCR results [9], whereas there were no noticeable changes in the same genes in the lung (Fig.11D). In cross-comparison with mitochondrial cholesterol–binding proteins, scatter and volcano plots suggest several genes are likely to replace TSPO functionally as an outer mitochondrial membrane cholesterol–binding protein (Figs. 11E and 11F, S7 and S13). To further illustrate the positive changes of mitochondrial cholesterol–binding protein-encoding genes, Ucp2 and Fdxr genes were both retrieved for analysis, instead of exons, as the basic unit (Fig. S4) [48]; genes in a mitochondrial protein interaction network are highlighted (colored) (Fig. 12). Interestingly, one of the upregulated genes, the Apoo gene that encodes 3 isoforms (mitochondrial form Iso1, partial mitochondrial form Iso2, and nonmitochondrial form Iso3), is related to differential pre-messenger RNA (mRNA)-splicing of exon 1, whereas the rest of the genes are involved in ion transport or proton leak associated with mitochondrial membrane potential, which likely reflects the basic role of TSPO conserved across cells, tissues, and species [51]. To cross-compare with the data from previous RNA-seq analyses, we present selected genes from DESeq2 analysis (Fig. S14; Table S6), which is consistent with a previous report [9].

Figure 12.

The compensatory mechanism after Tspo deletion involves upregulation of mitochondrial cholesterol-binding protein-encoding genes. The protein-protein interaction network was produced from the STRING database with medium confidence (0.400) and 20 interactions per each query proteins from the first (color) shell of interaction plus the second (white) shell of interactions and visualized using the Cytoscape plugin AllegroLayout (version 2.2.2; AllegroVivia, Inc) with the option of weak clustering visualization [52]. Red arrows: mitochondrial cholesterol-binding proteins, dark brown arrows: 2 mitochondrial proteins (UCP2 and FDXR) that have dramatic differences in base pair read counts between wild-type (Tspo+/+) and homozygous (Tspo–/–); Green dashed circle: position of translocator protein (TSPO) and red dotted oval: highlight of steroidogenic genes. Exon-based: Apoo, Cox7a2, Cyb5b, Cyc1, Letm1, Mfn1, Mfn2, Ndufa13, Pisd, Sco1, Slc25a10, Slc25a13, Slc25a33, Slc30a6, Timm17b, Timm44, Abcb6, Abcb7, Atad3a, and Opa1; gene-based: Fdxr and Ucp2.

To validate the reanalysis of RNA-seq data, we performed real-time PCR and RNA-seq data visualization in a genome browser (Figs. S3-S13). Several important genes were selected among upregulated and downregulated gene sets to confirm whether their expression levels were increased or decreased in the adrenal glands of the Amhr2-Cre–mediated Tspo gHO mice using real-time PCR (Fig. 13). Real-time PCR confirmed the increased expression levels of Abcb6, Ucp1, Ucp2, and Apoo in Tspo gHO mice but not that of Trp53imp2 (Fig. 13A-13D). Real-time PCR also confirmed the reduced expression of Nr4a2 and Trim5 in the Tspo gHO mice, although there was great variability (Fig. 13E and 13F). Each exon of the investigated genes was mapped to an MA-plot of the RNA-seq data (Fig. 13G). Hprt was used as the reference gene in the real-time PCR and is found around the central line. The upregulated genes Ucp1 and Ucp2 in red are above the line, and the downregulated gene Nr4a2 is below the central line. Each coding exon of Tspo is shown as the control: Exon2 and exon3 are down to the edge of the dot distribution and exon4 is below the central line (Figs. 11G; S3 and S8) [48]. UCP1 and UCP2 are both required for mitochondrial proton leak (Fig. 13H).

Figure 13.

Validation of the RNA-sequencing (RNA-seq) analysis by real-time polymerase chain reaction (PCR). Total RNA was extracted from adrenal glands from wild-type (WT) (Tspo+/+) vs homozygous (HO) (Tspo–/–) mice. The expression of each gene relative to Hprt was determined by real-time-PCR and graphed as dot plots. Results are expressed as fold changes in log2 value, and for each sample the fold change in gene expression was calculated over its mean expression in WT samples. Data are presented as means ± SEM (indicated as cross) from 7 to 9 animals (depicted as dots/squares), each performed in triplicate. A to D, Upregulated genes, Abcb6, Ucp2, Apoo, and Trp53inp2. E and F, Downregulated genes, Nr4a2 and Trim5. *P less than .05, ***, P less than .001, and n.s., nonsignificant, Student t tests (n = 7-9 animals per group). #, P less than .01 and ###, P less than .001, F test to compare variances. G, Highlighted Ucp1, Ucp2, and Nr4a2 genes within the MA plot used to visualize the differential expression analysis of the mouse adrenal RNA-seq data. Hprt, used as reference gene, and Tspo, a negative/positive control, are shown. Points indicate individual exons, x axis indicates the normalized mean, and y axis indicates the log2-fold change. Red/orange, upregulated Ucp1 and Ucp2; Green, downregulated Nr4a2; blue, reference gene Hprt, and purple, negative/positive control gene Tspo. H, Diagram of the main upregulated Ucp1 and Ucp2 genes in relation to the proton leak and mitochondrial membrane potential (Δψ m). Protons (H+) are extruded from the inside (mitochondrial matrix) to the outside of the, inner mitochondrial membrane (IMM), thus forming an electrochemical proton gradient, also referred to quantitatively as the Δψ m. Upregulated UCP1/2 induces a leak of H+ through the IMM and “uncouples” the free energy stored in the electrochemical gradient from adenosine triphosphate (ATP) synthesis. Abbreviations: ANT, adenine nucleotide translocase; ETC, mitochondrial electron transport chain (I-IV, complex I to IV; OMM, outer mitochondrial membrane; Q, ubiquinone form of the lipophilic electron carrier [Coenzyme Q10]; UCP1/2, uncoupling protein 1/2; VDAC, voltage-dependent anion-selective channel.

The transcriptome profiling changes and especially the expression levels of several specific genes, verified from real-time PCR, show that compensation for TSPO deficiency took place. This likely involves mitochondrial membrane potential via changes between inner and outer mitochondrial membrane contact through the mitochondrial contact site complex and cholesterol transport from Golgi vesicles [51, 53, 54].

4. Discussion

Amhr2-Cre was first reported to mediate the targeted disruption of the widely expressed type I bone morphogenetic protein (BMP) receptor Bmpr1a (also known as Alk3) in Amhr2-Cre–knockin mice, where the Cre replaced one allele of the Amhr2 gene [55]. Since then, there have been as many as 90 publications using the Amhr2-Cre–based gene knockout method [17]. The Amhr2-Cre/LoxP system was initially used by Morohaku et al to generate Tspo cKO mice [8]. In this study, the authors used cHO (Tspo-Amhr2; TSPOcΔ/Δ, the nomenclature used by the authors vs Amhr2-Cre+/–, Tspofl/fl, nomenclature used herein) mice to cross back with homozygous Tspo-floxed mice (Tspo) in an effort to generate cHO (TSPOcΔ/Δ) mice. However, we were unable to replicate the Amhr2-Cre–mediated testis-specific cKO mice using the same methodology [12]. Nevertheless, a common observation shared between these 2 reports is that the cKO mice produced a “WT-like” phenotype, but Morohaku et al and we came up with totally different conclusions [8, 12]. Evidence presented in these reports shows a lack of function at early embryonic stages in the presence of Amhr2-Cre, suggesting that one or both of the floxed alleles in TSPOcΔ/Δ mice used was “defective.” Thus, the produced offspring are either all WT or gHE with only an 8.3% chance of being one of the cHO genotypes (Amhr2-Cre+/–, Tspo; and Amhr2-Cre+/–, Tspo) because of the strong genetic linkage between the Tspo and Amhr2 genes. Our data suggest caution should be used when interpreting data generated using the Amhr2-Cre to produce cKO mice because Amhr2-Cre could also serve as a “global deleter” to generate gKO mice. Indeed, our findings provide an explanation of the data published using the Amhr2-Cre to generate an ovarian cKO for β-catenin that led to unexpected recombination in a wide range of tissues [56], as well as the partial depletion of TSPO in the mouse ovary, which actually suggests that these are gHE mice [8]. There is also another possibility that because the cells at this early embryonic developmental stage remain pluripotent, some cells that have been deleted for Tspo might die (if TSPO protein is required at that stage for cell survival), but they might be replaced by the pluripotent “WT” (or Tspo undeleted) cells leading to normal development. Amhr2-Cre would then be reactivated later during development allowing for tissue-specific gene ablation. This possibility is unlikely to occur but remains to be tested further.

During the characterization of the Amhr2-Cre–mediated Tspo gKO, we examined tissues other than the well-known Amhr2-target tissues (gonads). One interesting finding is that Tspo was sparsely but strongly expressed in the brain, which is in agreement with our previous report using immunofluorescence [12]. Numerous studies have shown that Tspo is mainly expressed in brain microglia cells [57, 58]. These observations are also agree with data using radiolabeled PK 11195 and/or other TSPO ligands in positron emission tomography to assess its expression in rodent and human brain [59-61]. However, recent reports suggested TSPO was extremely low in the brain when assessed by immunohistochemistry and immunoblot analyses [28, 62, 63]. This discrepancy raises the question of the availability of sensitive and specific antibodies for immunodetection of TSPO in the brain. In general, any negative reports should provide more experimental controls and be very cautious in their interpretation of the results, which may be due to a flawed experimental system.

Free cholesterol has been shown to be toxic to cells [64], and its detoxification has to be either through cholesterol esterification and storage into LDs or metabolism into hydrosoluble hydroxycholesterols or steroids in steroidogenic tissues. The LDs in steroidogenic tissues are mainly composed of cholesteryl esters, which differ from the LDs in adipose tissues that mainly consist of triglycerides [45, 65]. KO or knockdown of the genes involved in steroid biosynthesis results in LD accumulation, indicating lack of availability of free cholesterol for steroid formation. This is the case for Tspo [12, 14], Star [13, 66], Cyp11a1 [67], Cyp11b2 [68], and Nr5a1 (SF-1) [69]. Indeed, almost all mouse tissues examined herein showed changes in neutral lipid accumulation under the TSPO deficiency, in agreement with a recent study we performed in the rat [14], and also it was positively correlated with elevated Aug1 gene expression [70]. Even during the development of Leydig cells, the appearance or disappearance of cytoplasmic LDs is one of the visible phenotypes used to establish the developmental stage of the cell [71, 72]. Taken together with the fact that TSPO is a high-affinity cholesterol-binding protein extremely abundant in the outer mitochondrial membrane of steroidogenic cells, and the vast literature on the effects of TSPO drug ligands on steroid formation in various tissues and animal models [1, 2, 5, 73], these findings indicate that TSPO must play a regulatory role in steroid formation.

Despite the changes seen in LD accumulation, the effects seen in circulating steroid hormone levels are not as dramatic as one would expect. We noted, however, a significant 50% reduction both in intratesticular and circulating testosterone levels. Further analysis of the ability of Leydig cell preparations isolated from WT and Tspo gKO mice demonstrated the lack of Tspo gKO Leydig cells to respond to hCG in terms of testosterone formation. This result is similar to that reported using steroidogenic cell (Sf1)-targeted Tspo cKO mice that failed to respond to ACTH stimulation [12]. In the present studies, we did not see an effect of basal circulating corticosterone levels in agreement with data reported with Sf1-targeted Tspo cKO mice, suggesting that basal, hormone-independent, steroid production by adrenal and Leydig cells is not affected by the lack of TSPO.

We were surprised by the slight but significant increase in Tspo gKO testis weight that we cannot explain. However, we were not surprised by the lack of changes in seminal vesicle, kidney, and epididymis weights, some of which are androgen-responsive organs. Over the years the mouse, although a great model for genetic manipulation, unlike the rat, has been proven to not be a good model to study the relationship between androgen production and testosterone effects on androgen-targeted tissues. The main difference between the mouse and rat is that the mouse testis contains very low levels of androgen binding protein (ABP), 2% to 4% of what is found in the rat [74]. Without ABP, the majority of testosterone made in the mouse testis is free and thus active. The impact of this finding was shown in studies in which the LH receptor was knocked out in mice but, nonetheless, hormone-independent testosterone formed by the Leydig cells, representing 2% of the total, was sufficient to maintain androgenicity and spermatogenesis [75]. Thus, reduction of androgen production by 50% in Tspo gKO mice would not affect the weight of androgen-responsive organs.

The increased presence of autophagosomes in Tspo gKO mouse Leydig cells, compared to WT, confirms the recently proposed hypothesis that TSPO is an element in the regulation of mitochondrial quality control in autophagy [76]. In this paper the authors elegantly showed that TSPO inhibits autophagy. Thus, in TSPO-deficient cells one should expect increased autophagy. Autophagy in normal Leydig cells has been shown to participate in testosterone production by providing cholesterol for steroidogenesis [77]. In Tspo-deficient Leydig cells, however, lack of TSPO would limit substrate availability and thus testosterone formation.

We believe that there are 2 reasons that might explain the lack of drastic changes in steroid production and mitochondrial structure, to be expected considering that TSPO comprises 2% to 4% of the outer mitochondrial membrane: first, the use of plasma membrane cholesterol, instead of LD cholesterol, for steroidogenesis as previously shown [78, 79] and second, there must be compensatory mechanisms occurring at early stages of embryo development. Previous studies showed that both TSPO and STAR are needed for the development of preimplantation embryos because lower expression of either gene leads to DNA fragmentation of preimplantation embryos, and thus apoptosis, as well as lipid accumulation [20]. In agreement with this report, the delayed development of preimplantation embryos suggests that compensation for the loss of TSPO occurs before embryo implantation. Thus, we speculate that the gHE mice used to generate gHO mice survived after preimplantation selection. This hypothesis is supported by data from another TSPO gKO mouse, the “Guwiyang Wurra” mouse, in which there is no linkage bias away from Mendelian ratios from crossing 2 gHE mice [58]. Additional evidence suggesting that functional adaption may occur when TSPO is absent comes from evolutionary studies showing that TSPO, a 3.5 billion year-old protein, part of the last universal common ancestor, is highly conserved across species, and a human TSPO deletion/mutation has not yet been reported [7, 80-82].

Genetic compensation (gene or gene network replacement) in response to gene KO, where the DNA lesion and presence of a mutated mRNA lead to relevant stress responses, is a widespread phenomenon [49, 50, 83]. Because gene deletion could elicit genetic compensatory mechanisms, gene KO may be a preferable approach to demonstrate the biological function of a gene in vivo [49]. Nevertheless, differences due to the deletion of a gene should be reflected in the transcriptome profiling. Reanalysis of previously published RNA-seq data sets indicated that the loss of TSPO invoked global gene expression changes, in contrast to a previous report [28]. Although the methodology used is state of the art, we are aware that the conclusions from RNA-seq data analysis depend not only on the sequencing coverage, depth, and numbers of reads for the transcriptome, but also on the different data analysis pipelines used in different laboratories, both of which could lead to the discrepancy. Although this is not the first RNA-seq data analysis of a TSPO-depleted tissue, it provides information for the first time regarding transcriptome changes, validated by real-time PCR, that likely, at least in part, provides the biological basis for the compensation to the Tspo deletion. Such transcriptome changes involved in genetic network compensation for a lost gene can provide a reference point for properly interpreting nonvisible phenotypes after gene deletion, which could be traced back to the early stages of preimplantation embryo development, such as that shown in this report [49, 50, 83, 84].

TSPO-deletion–induced compensatory mechanisms could become activated at various steps in the process of steroid hormone formation. Despite previous evidence showing that TSPO is involved in cholesterol import into mitochondria, the rate-limiting step in steroid biosynthesis [1], the complete or partial loss of this protein may change the cell processes involved in the regulation of cholesterol storage, trafficking, and availability in a distinct way. From our analysis of mitochondrial-related gene expression profiles, several genes were identified that showed changes in their expression levels in response to TSPO deletion. Some of these genes encode cholesterol binding proteins and/or are involved in cholesterol metabolism [32, 85-89], and existing literature supports their role in steroid hormone biosynthesis. Indeed, silencing the APOO gene leads to overexpression of UCP2 in HepG2 cells, and APOO expression has been shown to be induced by oleic acid, whereas oleic acid and/or its activated form oleoyl-CoA boosts progesterone biosynthesis in hormone-dependent steroidogenic MA-10 Leydig cells [90, 91]. UCP1 and UCP2 are both required for mitochondrial protein leakage, which is likely involved in the reduced mitochondrial membrane potential [51, 92, 93]. In addition, SLC25A13 catalyzes the calcium-dependent exchange of cytoplasmic glutamate with mitochondrial aspartate and a proton across the inner mitochondrial membrane, and its activation is strongly dependent on mitochondrial membrane potential likely via TSPO [51, 94]. Also, Nr4a2 (or Nurr1), downregulated under TSPO deficiency, is involved in the regulation of adrenal aldosterone production likely via increasing the adrenal principal isoform of 3β-hydroxysteroid dehydrogenase (3β-HSD), HSD3B2 [95, 96].

In conclusion, our data provide strong evidence that the Amhr2-Cre mediated the generation of a Tspo gKO, suggesting that Amhr2-Cre may have wide application as a global “Cre deleter” in the mouse. Our data also show that TSPO deficiency leads to delayed preimplantation embryo development, and disturbed neutral lipid accumulation and steroid biosynthesis, as well as a compensatory gene (genetic) network in response to the TSPO genetic depletion.