Peripheral Benzodiazepine Receptor/Translocator Protein Global Knock-out Mice Are Viable with No Effects on Steroid Hormone Biosynthesis^,

Background: Translocator protein (TSPO) has been considered a mitochondrial cholesterol transporter critical for steroid hormone production. TSPO knock-out mice were reported to be embryonic lethal.

Results: TSPO knock-out mice are viable with no effects on steroidogenesis.

Conclusion: TSPO is not essential for steroidogenesis and is not necessary for sustaining life.

Significance: This study rectifies a serious inaccuracy in the current understanding that is critical for treating steroid hormone disorders.

Abstract

Translocator protein (TSPO), previously known as the peripheral benzodiazepine receptor, is a mitochondrial outer membrane protein implicated as essential for cholesterol import to the inner mitochondrial membrane, the rate-limiting step in steroid hormone biosynthesis. Previous research on TSPO was based entirely on in vitro experiments, and its critical role was reinforced by an early report that claimed TSPO knock-out mice were embryonic lethal. In a previous publication, we examined Leydig cell-specific TSPO conditional knock-out mice that suggested TSPO was not required for testosterone production in vivo. This raised controversy and several questions regarding TSPO function. To examine the definitive role of TSPO in steroidogenesis and embryo development, we generated global TSPO null (Tspo^−/−) mice. Contrary to the early report, Tspo^−/− mice survived with no apparent phenotypic abnormalities and were fertile. Examination of adrenal and gonadal steroidogenesis showed no defects in Tspo^−/− mice. Adrenal transcriptome comparison of gene expression profiles showed that genes involved in steroid hormone biosynthesis (Star, Cyp11a1, and Hsd3b1) were unchanged in Tspo^−/− mice. Adrenocortical ultrastructure illustrated no morphological alterations in Tspo^−/− mice. In an attempt to correlate our in vivo findings to previously used in vitro models, we also determined that siRNA knockdown or the absence of TSPO in different mouse and human steroidogenic cell lines had no effect on steroidogenesis. These findings directly refute the dogma that TSPO is indispensable for steroid hormone biosynthesis and viability. By amending the current model, this study advances our understanding of steroidogenesis with broad implications in biology and medicine.

Introduction

The peripheral benzodiazepine receptor (PBR)² was first identified as a protein with distinct pharmacology and high binding affinity to benzodiazepines in “peripheral” sites, as opposed to “central” sites found in brain tissue (1–4). The central benzodiazepine receptor is the γ-aminobutyric acid type A (GABA_A) receptor that has specific function in inhibitory neurotransmission (5, 6). However, the function for the PBR has remained a topic of debate for decades (7, 8), despite its prominence as a therapeutic target in a variety of diseases/disorders (9–12).

Genetic sequence comparisons show that the PBR gene is highly conserved from archaea to metazoans (13). In mammals, highly prominent PBR protein expression has been identified in steroidogenic tissues (14–16). Binding sites for PBR ligands have also been identified in other tissues including heart, brain, adrenal, kidney, salivary gland, platelets, brown adipose tissue, skin, and liver (7). In subcellular fractions, binding sites for the PBR were identified to be present in the outer mitochondrial membrane (OMM) (14, 17).

The transport of cholesterol from intracellular stores and the OMM to the inner mitochondrial membrane (IMM) forms the first and rate-limiting step in steroid hormone biosynthesis (18, 19). Cholesterol is then converted to the first steroid, pregnenolone, by CYP11A1 (cytochrome P450 side chain cleavage), an enzyme located on the matrix side of the IMM (20, 21). The initial link between PBR and steroid hormone biosynthesis emerged by studying the effect of a PBR-binding chemical PK11195 that triggered steroidogenesis in the Y1 mouse adrenal tumor cell line (22). In agreement, targeted disruption of PBR in the R2C rat Leydig tumor cell line inhibited steroidogenesis (23).

The structure of murine PBR showed a five α-helix fold, each forming a transmembrane segment (24), and the possibility of a channel-like structure with an interior surface for potential substrate translocation (25). Cholesterol binding to PBR was defined by the identification of a cholesterol recognition amino acid consensus sequence at the C-terminal region (26). These structural features combined with functional considerations led to the modeling of PBR as a membrane-associated cholesterol transport protein (9, 27). Based on these findings, to more accurately represent its subcellular role, PBR was renamed as translocator protein (TSPO) (28).

Independent of studies on PBR/TSPO, another mitochondrial protein that is expressed in rapid response to steroidogenic stimuli, the steroidogenic acute regulatory protein (STAR), was identified in the MA-10 Leydig tumor cell line (29). Induced expression of STAR resulted in increased steroid hormone production even in the absence of hormonal stimulation (29). Anchoring STAR in a mitochondrion-affixed TOM20-STAR fusion protein in MA-10 cells resulted in the constitutive production of steroid hormone (30). In humans, STAR mutations cause lipoid congenital adrenal hyperplasia that ranges from an almost complete inability to synthesize steroids (31) to less severe forms that retain partial STAR protein activity (32). Also in support, Star gene-deleted mice showed an almost complete inability to synthesize steroid hormones underscoring its critical role in steroidogenesis (33).

Two key experiments appeared to cement a functional link between STAR and PBR/TSPO. First, inhibition of protein synthesis that blocks STAR production and steroidogenesis (34) did not affect the ability of PK11195 to trigger steroidogenesis (35). Second, knockdown of PBR/TSPO in mitochondrion-affixed TOM20-STAR fusion protein expressing MA-10 cells resulted in a reduction in their steroidogenic capacity (36). Thus, the PBR function was linked to transport of cholesterol from the OMM to the IMM (37).

In a recent study, we questioned this model showing that PBR/TSPO expression is not required for in vivo testicular steroidogenesis (38). This result has indeed raised a lot of questions and controversy regarding the current understanding of the precise PBR/TSPO-mediated functions in steroid hormone production (39). Moving forward, there remained an absolute need to examine specific deficits in TSPO global knock-out mice to gain additional functional understanding of this protein (40).

In a review article published in 1997, it was claimed that efforts to generate a Pbr/Tspo gene-deleted model failed as a result of early embryonic mortality in PBR/TSPO-deficient mice (41). This led to the conclusion that in addition to steroidogenesis, TSPO is involved in basic functions that are critically necessary for embryonic development. However, this review article (41) failed to explain experimental methods covering the recombination strategy used and the exact stage of embryonic mortality; these specific methods/results were never published. Given the current gap in understanding PBR/TSPO function in vivo, we decided to revisit this global knock-out question.

With the primary objective of studying the role of PBR/TSPO in steroidogenesis and other vital functions (if any) that may be critical for embryonic development, we used a germ cell specific knock-out approach to generate Pbr/Tspo global gene-deleted mice. Our results are unexpected and contrary to the prevailing view in that we find PBR/TSPO knock-out mice are viable and fertile with no effects on steroid hormone biosynthesis.

EXPERIMENTAL PROCEDURES

Generation of Tspo^−/− Mice

Mice with floxed alleles for Pbr/Tspo with loxP sites flanking exons 2 and 3 (Tspo^fl/fl) were generated and validated as described previously (38). In this study, Tspo^fl/fl female mice were crossed with Ddx4-cre+ (FVB-Tg^{(Ddx4-cre)1Dcas/J} (42)) males to generate a germ cell-specific deletion of PBR/TSPO (TspocΔ/+) in both male and female mice. TspocΔ/+ Ddx4cre+ positive offspring were then backcrossed to generate TspocΔ/Δ mice of both sexes. These sperm- and oocyte-specific knock-outs were bred to generate global PBR/TSPO null (Tspo^−/−) offspring in the C57BL/6 background. We also observed that oocyte-specific TspocΔ/+ Ddx4-cre+ mice induced global recombination directly as described previously for the Ddx4-cre+ strain of mice (42). Cre was subsequently bred out from Tspo^−/− mice (Fig. 1). Breeding colonies for Tspo^−/− and Tspo^fl/fl mice were maintained separately, and offspring were genotyped for all experiments (Fig. 1) using primers that have been previously validated (38). ROSA26-tdTomato (R26-tdTom) reporter females (B6.Cg-Gt(ROSA)26Sor^{tm14(CAG-tdTomato)Hze} (43)) were used to confirm recombination and bred to generate Tspo^fl/fl-R26-tdTom homozygous mice to directly mark PBR/TSPO deletion. Animals were maintained in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals. The Institutional Animal Care and Use Committee of Cornell University approved all experiments in this work.

FIGURE 1.

Generation of TSPO knock-out mice. A, schematic showing genomic Tspo locus in wild type (Tspo^+/+), intact floxed (Tspo^fl/fl), and knock-out (Tspo^−/−) mice. Tspo^fl/fl mice contained exons 2 and 3 flanked with LoxP sites (open arrowheads). Tspo^fl/fl female mice were bred with Ddx4-cre male mice to generate germ cell-specific deletion of TSPO as heterozygotes. These sperm and oocyte-specific knock-out mice were bred to generate Tspo^−/− offspring. Ddx4-cre transgene was subsequently bred out from this colony. B, panel showing genotyping for floxed alleles and cre transgene in Ddx4-cre mice, Tspo^fl/fl mice, Tspo^+/− Ddx4-cre mice, Tspo^−/− Ddx4-cre mice, and Tspo^−/− mice. C, genomic DNA PCR for recombination at the Tspo locus detects deletion of exons 2 and 3 in Tspo^−/− mice (product sizes: Flox-2697 bp, Null-872 bp, and Con-161 bp).

Phenotypic Analysis

Growth rate of Tspo^fl/fl and Tspo^−/− mice were examined from postnatal day 0 to week 5. Litter sizes from homozygous crosses in Tspo^fl/fl and Tspo^−/− mouse colonies were examined in at least eight unique pairings. Recombination was examined in different tissues by both PCR of genomic DNA using specific primers, as validated previously (38), and tdTomato reporter expression indicating cre-mediated recombination, as described previously (38).

Histology

Tissues were processed for hematoxylin and eosin staining as described (38). Immunohistochemistry for PBR/TSPO was performed using a rabbit monoclonal anti-TSPO antibody (Abcam) that specifically recognizes amino acids 156–169 at the C terminus of exon 4 as described (16, 38). For examining neutral lipid deposits, 9-μm frozen sections were prepared, and lipid staining was performed using Oil Red O (Matheson Coleman & Bell) in 60% isopropyl alcohol for 30 min. Fluorescent tdTomato was also visualized in 9-μm frozen sections in multiple tissues. Images were acquired using either a DFC365FX or an ICC50HD camera (Leica) or a Meta 510 confocal scope (Zeiss). Density of Oil Red O staining density was quantified using ImageJ (44).

Immunoblots

Proteins were separated by SDS-PAGE and immunoblotted using rabbit monoclonal primary antibodies against TSPO (Abcam), isocitrate dehydrogenase 2 (Abcam), and a rabbit polyclonal STAR antibody (45). Each primary antibody was multiplexed with the loading control ACTB (Li-Cor). Simultaneous detection was performed using IRDye 700 and 800 under a laser fluorescence scanner (Li-Cor) and quantified as described previously (38). For immunoblots performed in knockdown experiments, TSPO polyclonal (Cell Signaling) and ACTB (Invitrogen) primary antibodies were used. Specific band intensities in knock-down experiments were calculated by image analysis using Quantity One software (Bio-Rad).

Hormone Assays

Plasma was collected from 8- to 10-week-old Tspo^fl/fl and Tspo^−/− mice. Levels of different steroid hormones were measured using radioimmunoassay as follows: corticosterone, aldosterone, and testosterone (Siemens), estradiol and progesterone (Serono Maia), as described previously (38, 46). For adrenal/corticosterone stimulation test, 500 ng/g body weight of adrenocorticotropin (ACTH) fragment 1–24 (Sigma) was administered subcutaneously, and plasma was collected after 1 h. For the testis/testosterone stimulation test, 10 IU of human chorionic gonadotropin (EMD Biosciences) was administered intraperitoneally to male mice, and plasma was collected after 1 h.

Transmission Electron Microscopy

Adrenal glands were fixed in Karnovsky's fixative (2.5% glutaraldehyde, 2% paraformaldehyde in 0.08 m phosphate buffer) overnight. After fixation, tissues were processed for transmission electron microscopy as described previously (12). Stained sections were examined in a Tecnai^TM T12 Spirit electron microscope (FEI), at 120 kV. High magnification images were acquired using a high resolution CCD camera Megaview III (Olympus Soft Imaging System) to examine cellular and mitochondrial morphology within regions in the adrenal cortex.

RT-PCR and Quantitative PCR

Total RNA was extracted from adrenal glands of 8–10-week-old Tspo^fl/fl and Tspo^−/− mice using RNeasy lipid tissue mini kit (Qiagen). For human A172 and H295R cells, total RNA was extracted using TRIzol (Invitrogen). Reverse transcription was carried out using Multiscribe^TM reverse transcriptase (Invitrogen). Validated TaqMan gene expression assays were used for quantitative PCR estimation of murine Tspo (Mm00437828_m1), human TSPO (Hs00559362_m1), Tspo2 (Mm01281420_m1), Star (Mm00441556), Cyp11a1 (Mm00490735_m1), and Hsd3b1 (Mm00476184_g1). For all other genes, expressions were quantified by SYBR® Green detection method using validated primer sequences obtained from PrimerBank database (47). The following primers were used: Vdac1, 5′ AAGTGAACAACTCCAGCC TGA 3′ and 5′ CACCAGCATTGACGTTCTTG 3′; Vdac2, 5′ CTCCACCCTATGCTGACCTC3′ and 5′ CCCGCTAACTTTACCAGTGTCT 3′; Vdac3, 5′ CAAAGGGTATGGGTTTGGCAT 3′ and 5′ TTGGTCTCTAGGTTGCCTGAT 3′; Ant1, 5′ CTGGTGTCCTACCCCTTTGA 3′ and 5′ TGGCTCCTTCGTCTTTTGCA 3′; Ant2, 5′ CAAGACAGCGGTAGCACCC 3′ and 5′ CGCAGTCTATGATGCCCTTGTA 3′; Hxk2, 5′ GGGCATGAAGGGCGTGTCCC 3′ and 5′ TCTTCACCCTCGCAGCCGGA 3′; Acbp, 5′ TTTCGGCATCCGTATCACCT3′ and 5′ TTTGTCAAATTCAGCCTGAGACA 3′; and Pap7, 5′ GAGGAGCTTTACGGCCTGG 3′ and 5′ CTTATGCAGTGCCACGAACTT 3′. All expressions were normalized to an internal control gene, Gapdh. Relative quantification of fold-change was performed using the 2^−ΔΔCt method (48).

Adrenal Transcriptome Shotgun Sequencing

Adrenal gland total RNA from Tspo^fl/fl and Tspo^−/− mice were extracted, and poly(A) RNA were isolated using oligo(dT)₂₅ Dynabeads (Invitrogen). Fragmentation and strand-specific RNA-seq library preparation was performed as described (49). Each sample with unique bar codes was pooled to one lane for obtaining short single-reads in a HiSeq 2000 sequencer (Illumina). RNA-seq reads were first aligned to the ribosomal RNA gene database (50) using Bowtie (51), and the mapped reads were discarded. The resulting filtered reads were aligned to the mouse genome (Ensembl) using TopHat (52). Subsequent to alignments, raw counts for each mouse gene were derived and normalized to reads per kb of exon model per million mapped reads. Differentially expressed genes were identified using the DESeq package (53). Raw p values of multiple tests were corrected using false discovery rate (54).

TSPO Knockdown in Different Steroidogenic Cell Lines

Specific small interfering RNAs (siRNAs) were used to knock down endogenous TSPO expression in different steroidogenic cell lines as described previously for other genes (55). Tspo siRNAs and the Silencer® negative control (scrambled) were obtained as annealed oligonucleotides (Invitrogen). Sense strand sequences for TSPO siRNAs utilized were as follows: 1) 5′ GGAAAGAGCUGGGAGGUUUtt 3′; 2) 5′ GCUCCUACAUAGUCUGGAAtt 3′; and 3) 5′ UGGGCCUGCUAGUCUGUCAtt 3′. Cell lines MA-10, Y-1, H295R, R2C, and MLTC cells (ATCC) were transfected with 100 nm Tspo siRNA1–3 or control using Lipofectamine 2000 (Invitrogen). Forty eight hours after transfection, cells were examined for TSPO protein expression and used to estimate steroidogenic potential with 0.5 mm dibutyryl cyclic AMP (Bt₂cAMP) as described previously (55).

RESULTS

Tspo^−/− Mice Are Viable and Fertile

The Tspo^fl/fl locus allows cre-mediated deletion of exons 2 and 3 of the Tspo gene (Fig. 1A). Exon 1 does not contain any start codons. Removal of exons 2 and 3 eliminates the start codon, and exon 4 does not have any in-frame start codons. Therefore, deletion of exons 2 and 3 will completely eliminate TSPO protein expression (38). Germ cell-specific Ddx4-cre mice were used to induce germ cell Tspo deletion in F1 and global recombination in F2 generations. Litters from all crosses were genotyped, and the Ddx4-cre transgene was subsequently bred out from this colony (Fig. 1B). Using a Tspo^fl/fl-R26-tdTom reporter line, we confirmed that cre-mediated recombination occurred in all tissues globally (Fig. 2, A and C). Recombination at the Tspo^−/−genomic locus was also confirmed by PCR in different tissues (Fig. 1C). The complete absence of TSPO protein expression in different tissues in Tspo^−/− mice was confirmed by both Western blots (Fig. 2B) and immunohistochemistry (data not shown), using a monoclonal antibody that recognizes protein coded by exon 4. This proved that no partial TSPO peptide was produced in mice deleted for Tspo exons 2 and 3 as shown previously (38). All Tspo^−/− mice were viable and showed no apparent phenotypic abnormalities. Litter sizes from Tspo^−/− pair matings were not different from Tspo^fl/fl pair litters (Table 1). Growth rate in Tspo^−/− mice measured up to 5 weeks of age were quite similar to Tspo^fl/fl cohorts. However, when analyzed based on sex, Tspo^−/− female mice showed a very modest (∼1 g) but significantly higher body weight than Tspo^fl/fl mice at 1–5 weeks of age; males body weights at this same age interval were not different between the two genotypes (data not shown).

FIGURE 2.

Validation of Tspo^−/− genotype. A, Tspo^fl/fl-ROSA26-tdTomato reporter mice were used to examine cre-mediated recombination induced in offspring from germ cell-specific deletions. Global recombination was clear in R26-tdTom-Tspo^−/− P3 pups. B, TSPO monoclonal antibody specific for peptide corresponding to TSPO exon 4 detected a protein of expected size (18 kDa) in Tspo^fl/fl tissues but not in Tspo^−/− tissues. ACTB and mitochondrial isocitrate dehydrogenase 2 were used as controls. Scale bar, 1 cm. C, representative images of different tissues from an R26-tdTom-Tspo^fl/fl mouse with no tdTomato fluorescence in contrast to tissues from an R26-tdTom-Tspo^−/− mouse showing tdTomato fluorescence (red) indicating recombination. Nuclei were counterstained with DAPI (blue). Scale bar, 50 μm.

TABLE 1.

Litter sizes from Tspo^fl/fl and Tspo^−/− male and female mice

Mating pairs	n	Litter sizea ± S.E.
Tspofl/fl female × Tspofl/fl male	17	8.11 ± 0.44
Tspo−/− female × Tspo−/− male	8	8.00 ± 0.53

^a No significant differences in litter size were detected between the two crosses. Sex ratios in these litters were also not different.

Gonadal Steroid Hormone Production Is Not Affected in Tspo^−/− Mice

The lack of TSPO did not affect the functional morphology of the testis (Fig. 3A) or the ovary (Fig. 3B) in Tspo^−/− mice compared with Tspo^fl/fl controls. In male mice, examination of plasma testosterone concentration showed that base-line levels were not different between Tspo^−/− and Tspo^fl/fl mice (Fig. 3C). The increase in plasma testosterone in response to human chorionic gonadotropin in vivo was not different between Tspo^−/− and Tspo^fl/fl mice (Fig. 3D). Cauda epididymal sperm counts were not different between Tspo^−/− and Tspo^fl/fl mice (data not shown). In female mice, examination of plasma estradiol and progesterone concentration showed that TSPO deletion did not decrease base-line values (Fig. 3, E and F). In actuality, Tspo^−/− mice showed modest but significantly higher estradiol levels compared to Tspo^fl/fl cohorts. We did not observe any abnormalities in gonad size, reproductive behavior, and fecundity in Tspo^−/− mice (data not shown).

FIGURE 3.

TSPO deletion does not affect gonadal steroidogenesis. A, testes sections showing TSPO localization in Leydig and Sertoli cells in Tspo^fl/fl but not in Tspo^−/− testis. Functional morphology of seminiferous tubules was not affected in Tspo^−/− testis. B, ovary sections showing TSPO localization in interstitial cells and surface epithelium, weak in theca and granulosa cells in Tspo^fl/fl ovary; no staining was observed in Tspo^−/− ovary. Functional morphology was not affected in Tspo^−/− ovary. C, base-line plasma testosterone levels were not different between Tspo^fl/fl and Tspo^−/− male mice (n = 24–25/group). D, increase in plasma testosterone levels after human chorionic gonadotropin stimulation was similar between Tspo^fl/fl and Tspo^−/− male mice (n = 6–7/group). E, base-line plasma estradiol levels were significantly higher in Tspo^−/− compared with Tspo^fl/fl mice (p < 0.05), but the difference was modest (n = 12–15/group). F, base-line plasma progesterone levels were not different between Tspo^fl/fl and Tspo^−/− mice (n = 12–15/group). Scale bar, 200 μm.

Adrenal Steroid Hormone Production Is Not Affected in Tspo^−/− Mice

The lack of TSPO expression did not affect the functional morphology of the adrenal cortex (Fig. 4A) in Tspo^−/− mice. Quantitation of neutral lipid staining in the adrenal cortex did not indicate any major steroid transport defects in Tspo^−/− mice (Fig. 4, B and C). Base-line levels of aldosterone (Fig. 5A) and corticosterone (Fig. 5B), two major hormones produced by the adrenal cortex, were not different between Tspo^−/− and Tspo^fl/fl mice. An increase in plasma corticosterone in response to ACTH in vivo was not different between Tspo^−/− and Tspo^fl/fl mice (Fig. 5C). There was also no evidence for compensatory increase in STAR expression in adrenals of Tspo^−/− mice compared with Tspo^fl/fl mice after ACTH treatment (Fig. 5, D–F).

FIGURE 4.

TSPO deletion has no effect on morphology and lipid deposits in the adrenal cortex. A, adrenal sections from Tspo^fl/fl and Tspo^−/− mice showing TSPO localization in adrenocortical cells with a higher density in the zona glomerulosa; no staining was observed in Tspo^−/− adrenal. No difference in adrenocortical morphology was apparent between two genotypes. B, Oil Red O staining of adrenal glands showed no difference in neutral lipid deposits between Tspo^fl/fl and Tspo^−/− adrenal glands. C, quantification of Oil Red O (ORO) labeling density was similar in both Tspo^fl/fl and Tspo^−/− adrenal cortex (n = 5/group; mean ± S.E.). Scale bar, 200 μm.

FIGURE 5.

TSPO deletion has no effect on adrenal steroid hormone production. A, plasma aldosterone base-line levels were not significantly different between Tspo^fl/fl and Tspo^−/− mice (n = 7/group). B, plasma corticosterone base-line levels were not significantly different between Tspo^fl/fl and Tspo^−/− mice (n = 24–30/group). C, increase in plasma corticosterone in response to ACTH stimulation was similar between Tspo^fl/fl and Tspo^−/− mice (n = 10–14/group). D, representative Western blot showing no compensatory increase in STAR and TSPO protein expression after ACTH stimulation in both Tspo^fl/fl and Tspo^−/− mice. Quantification of STAR (E) and TSPO (F) protein levels also showed no significant increase at 1 h after ACTH stimulation (n = 3). A.U., arbitrary units. Data are represented as mean ± S.E.

Loss of TSPO Does Not Affect Steroidogenic Gene Expression in Adrenal Glands

Comparison of gene expression profiles by transcriptome shotgun sequencing of Tspo^−/− and Tspo^fl/fl adrenals revealed only a short list of differentially expressed genes (Table 2), none of which were directly associated with steroidogenesis. Mining reads representing Tspo mRNA confirmed deletion of exon 2 and exon 3. A majority of the differentially expressed genes was involved in immune activity. Validated expression of genes involved in the steroid hormone biosynthetic pathway, Star, Cyp11a1, and Hsd3b1 (Fig. 6, A–C), and genes encoding TSPO-interacting proteins as follows: Vdac, Ant, Hxk2, Pap-7, and Acbp, were not affected in Tspo^−/− adrenals (Fig. 6, E–L). Complete dataset is available through NCBI-SRA SRP043599. To evaluate possible functional redundancy, we also examined expression levels for Tspo2, a Tspo paralog (56). We did not detect Tspo2 expression in both Tspo^fl/fl and Tspo^−/− adrenal glands (Fig. 6D).

TABLE 2.

Genes differentially expressed between Tspo^fl/fl and Tspo^−/− adrenal glands

Genesa	Gene name/description	Tspofl/fl (mean)	Tspo−/− (mean)	Ratio	p value
BC018473	Unclassified noncoding RNA gene	0.01	1.86	186.33	1.77E-13
Trim12a	Tripartite motif-containing 12A	1.97	0.01	0.01	1.19E-12
Gm12196	Predicted gene 12196	0.26	40.59	156.12	5.94E-12
Pydc4	Pyrin domain containing 4	1.63	0.02	0.01	7.36E-10
Gm4955	Predicted gene 4955	3.25	0.07	0.02	9.45E-09
Gm16510	Predicted pseudogene 16510	1.19	23.5	19.81	5.15E-06
Gm5540	Predicted pseudogene 5540	0.71	19.88	28.14	0.0018
Prrc2a	Proline-rich coiled-coil 2A	1.16	20.69	17.78	0.0084
RP23–81C12.1	Known lincRNA	0.93	22.69	24.49	0.0084
Atxn2l	Ataxin 2-like	1.1	16.44	14.99	0.0147
Abhd1	Anhydrolase domain containing 1	1.34	10.53	7.84	0.0157
Ankrd52	Ankyrin repeat domain 52	0.35	5.62	15.91	0.0166
Nxpe2	Neurexophilin and PC-esterase domain family, member 2	0.8	0.12	0.15	0.0173
Tmem178	Transmembrane protein 178	16.36	3.04	0.19	0.0188
Epn1	Epsin 1	0.8	9.97	12.47	0.0320
Zbtb7b	Zinc finger and BTB domain containing 7B	0.38	5.51	14.64	0.0346
Osgin1	Oxidative stress induced growth inhibitor 1	2.46	19.15	7.79	0.0346
Cyp21a2-ps	Cytochrome P450, family 21, subfamily a, polypeptide 2 pseudogene	159.04	1352.39	8.5	0.0346
Abca2	ATP-binding cassette, sub-familyA (ABC1), member 2	0.41	3.38	8.24	0.0414

^a Data are presented in the order of significance based on p value.

FIGURE 6.

TSPO deletion does not alter expression levels of steroidogenic genes and TSPO-interacting proteins in the adrenal glands. Expression levels of steroidogenic genes Star (A), Cyp11a1 (B), and Hsd3b1 (C) were similar between Tspo^fl/fl and Tspo^−/− mice. Tspo2 expression (D) was not detected in both Tspo^fl/fl and Tspo^−/− adrenal glands. Genes encoding TSPO-interacting proteins Vdac1 (E), Vdac2 (F), Vdac3 (G), Ant1 (H), Ant2 (I), Hxk2 (J), Pap7 (K), and Acbp (L) showed similar expression levels between Tspo^fl/fl and Tspo^−/− mice. Tspo expression (M) was not detected in Tspo^−/− adrenal glands (n = 6/group). Data are represented as mean ± S.E. ACBP, acyl-CoA-binding protein; ND, not detected.

Ultrastructure of Adrenal Cortex Remains the Same in Tspo^−/− Mice

Examination of steroidogenic cells of the adrenal cortex using transmission electron microscopy did not show any aberrations in cellular ultrastructure in Tspo^−/− mice (Fig. 7). Despite its presence in the OMM, TSPO deficiency did not affect mitochondrial morphology. Lipid droplet distribution was also identical between regions within adrenal cortex (zona glomerulosa, zona fasciculata, and zona reticularis) in both Tspo^fl/fl and Tspo^−/− mice. TSPO deficiency also did not affect mitochondrial volume in cells, as assessed in Tspo^fl/fl and Tspo^−/− primary fibroblasts that showed no quantitative differences between the two genotypes (data not shown).

FIGURE 7.

Ultrastructure of cells in the adrenal cortex is not affected by TSPO deletion. Transmission electron micrographs showing cell morphology in the zona glomerulosa (A) and zona reticularis (B) of the adrenal cortex. Subcellular organelle morphology, including mitochondria and lipid droplets, is identical between both Tspo^fl/fl and Tspo^−/− mice.

TSPO Knockdown in Steroidogenic Cell Lines Does Not Affect Steroidogenesis

Progesterone production after Bt₂cAMP treatment by the different murine (MA-10, Y1, and MLTC) and human (H295R) steroidogenic cell lines was not adversely affected by knockdown of TSPO expression (Fig. 8, A–D). TSPO protein expression in murine cell lines decreased by ∼85, 75, and 82%, respectively; surprisingly, base-line TSPO protein expression was not detectable in human H295R cells (confirmed by extremely low levels of Tspo mRNA (Fig. 8, F and G)). This observation showed that H295R cells are capable of making steroid hormones without TSPO. In all conditions, TSPO knockdown did not have an effect on STAR expression. In the case of MA-10 cells, progesterone production showed a significant increase after Bt2cAMP treatment in TSPO knockdown cells compared with scrambled controls. This is in direct contrast with another study using the same MA-10 cells that showed significantly decreased progesterone production after just 60% knockdown of TSPO protein expression (36). However, such increases were not observed in Y1 and MLTC cells after TSPO knockdown under identical conditions. In R2C cells that abnormally display constitutively high levels of steroidogenesis, TSPO knockdown showed a modest but significant decrease in progesterone production in conditions with and without Bt₂cAMP treatment compared with scrambled controls (Fig. 8E). Mono-allelic Tspo gene deletion in R2C cells was reported to abolish TSPO expression and inhibit steroidogenesis (23). We find this hard to reconcile, as we do not find any evidence for allele-specific TSPO expression in our in vivo studies (Fig. 9). Moreover, R2C cells are pathologically constitutive in steroid hormone production, the mechanism of regulation of which is not physiologically relevant. Data from TSPO knockdown experiments in three different murine cell lines together with the discovery that TSPO expression is below the detection levels in H295R cells confirm that TSPO is not essential for steroidogenesis in vitro.

FIGURE 8.

TSPO knockdown in steroidogenic cells does not affect steroid hormone biosynthesis. TSPO knockdown (1 and 2) resulted in substantial decrease in TSPO protein compared with controls (scrambled) in MA-10 (A), Y1 (B), MLTC (D), and R2C cells (E). No base-line TSPO was detected in H295R cells (C). Bt₂cAMP treatment resulted in a similar increase in STAR protein expression in MA-10, Y1, MLTC, and H295R cells and higher progesterone production. TSPO knockdown MA-10 cells showed significantly higher progesterone production (p < 0.05) compared with scrambled controls. Progesterone production in TSPO knockdown Y1 and MLTC cells was similar to the scrambled controls. H295R cells did not show TSPO expression but still produced progesterone upon Bt₂cAMP treatment. Careful re-examination confirmed that H295R cells do not express both TSPO protein and mRNA (F and G). Abnormal R2C cells constitutively expressed high STAR protein levels and constantly produced high levels of progesterone. TSPO knockdown in R2C resulted in a modest but statistically significant decrease in progesterone production (p < 0.05), both with and without Bt₂cAMP treatment.

FIGURE 9.

TSPO expression is not allele-specific. A, TSPO monoclonal antibody detected similar expression levels in Tspo^fl/fl maternally transmitted heterozygous Tspo^+/− and paternally transmitted heterozygous Tspo^+/−; no TSPO was detected in Tspo^−/− adrenal glands. B, there was no significant difference in Tspo mRNA expression levels between Tspo^fl/fl and maternal heterozygous Tspo^+/− adrenal glands; no Tspo mRNA was detected in Tspo^−/− adrenals (n = 4/group; ND*, not detected).

DISCUSSION

According to existing literature, TSPO function is absolutely essential for steroid hormone biosynthesis (23), and global Tspo gene deletion results in early embryonic lethality in mice (41). In direct contrast to those observations, our experimental findings disprove these functional arguments and definitively demonstrate that TSPO is not essential for steroidogenesis and that Tspo^−/− mice are viable and fertile. These concrete results form a key step toward rectifying the prevailing model of steroidogenesis and underscore that the precise role of TSPO in mammalian physiology has yet to be uncovered.

In a previous study, we demonstrated that conditional Tspo gene deletion in testicular Leydig cells did not affect testosterone production (38). At that time, the finding was met with skepticism that Amhr2 cre-mediated recombination may not have complete penetrance in Leydig cells and criticism that it may not be the case for all steroid-synthesizing cells (39). From observations in this study in which Tspo has been deleted globally, it is clear that TSPO is not involved in physiologically relevant steroid hormone biosynthesis. This calls into question the previously suggested functional cooperation between STAR and TSPO for mitochondrial cholesterol import (57).

It could be argued that Tspo^−/− mice respond to the loss of TSPO with some unknown compensatory mechanism. If this is the case, it is surprising that redundant mechanisms were not evident in earlier studies that claimed an indispensable role for TSPO in the steroidogenic machinery (23). TSPO2, a paralog of TSPO (35% homology) and a likely candidate for functional compensation, is restricted to the adult bone marrow and is not expressed in steroidogenic tissues (56), negating such a possibility. Moreover, there is no experimental evidence that multiple mechanisms are involved in mitochondrial cholesterol import. Based on observations in STAR knock-out mice, it appears unmistakable that there is one rate-limiting mechanism of mitochondrial cholesterol import that is governed by STAR (33). Therefore, a model for TSPO in mitochondrial cholesterol transport is highly unlikely.

Given these results, it is unclear how the global TSPO knock-out reported in an earlier study by Papadopoulos et al. (41) could be embryonic lethal. As exact experimental details for that study were not published, we can only speculate that an error occurred in genome manipulation during this previous attempt. Similarly, a study from this same research group reported that more than 70% TSPO knockdown in cell lines resulted in cell death (58). This is certainly not the case in both our in vivo and in vitro experimental observations and indicates that experimental conditions used in these earlier studies may have not been optimal. This conclusion is supported by another recent study that independently generated conditional liver- and heart-specific Tspo^−/− mice and did not report any incidence of cell death after abolishing TSPO expression; TSPO was found not to be involved in mitochondrial permeability transition and cell death (59).

At base-line physiology, gene expression differences between Tspo^fl/fl and Tspo^−/− adrenals were surprisingly few. A majority of these genes are involved in immune modulation as follows: Tmem178 is a negative regulator of inflammatory cytokine production (60); Prrc2a is a gene within the major histocompatibility complex class III region and is involved in the inflammatory process (61); Osgin1 encodes an oxidative stress response protein that regulates inflammation (62); Zbtb7b is a regulator of T cell-mediated tumor-induced immunity (63); Pydc4 is part of a PYHIN domain family of proteins involved in innate immunity (64); and Trim12a is a putative antiviral gene (65). Two genes were involved in vesicular transport and lipid sequestration as follows: Epn1 is an endocytic adaptor involved in clathrin-mediated endocytosis (66), and Abca2 encodes an endolysosomal protein with a role in intracellular lipid trafficking (67). Three genes expressed were of unknown function: Ankrd52, Abhd1, and Atxn2l. The remainders were two noncoding RNA genes, two predicted genes, and three pseudogenes. Careful consideration of these differentially expressed genes and the phenotype observed in Tspo^−/− mice indicates that TSPO is not essential for sustaining physiological function.

TSPO binding chemicals have been shown to increase steroid hormone production in isolated mitochondria from steroidogenic cells (35), MA-10 cells (37), and even live rats (68). Given our current findings, we can offer two possible explanations for these pharmacological effects. With ligand binding, TSPO may undergo a conformational change (as suggested previously (24, 69)), resulting in the release of bound cholesterol that subsequently becomes available to mitochondrial transport systems and steroidogenesis. Alternatively, the effect of pharmacological agents could be due to drug-membrane interactions resulting in TSPO-independent biological effects (70). It has also been discovered that TSPO ligands like PK11195 can insert themselves into lipid bilayers affecting membrane properties (71), something that can modulate cholesterol availability to the steroidogenic machinery and also the nature of TSPO-ligand interactions. Nevertheless, as TSPO is considered a therapeutic target for numerous inflammatory conditions (9–12), these properties of TSPO ligands may be important to assess specificity and side effects.

The sexual disparity in body weight differences in that only female Tspo^−/− mice showed subtle but higher growth weights than their Tspo^fl/fl cohorts suggests a hormone-dependent metabolic mechanism (72). But the overall lack of phenotype in Tspo^−/− mice appears to suggest that TSPO function could be more relevant only in pathological disease states. We have now observed Tspo^−/− mice for up to 16 months of age with no apparent abnormalities/problems. Comparison of gene expression profiles after TSPO deletion also suggests potential changes to immune system function associated with TSPO deletion. We intend to investigate this further in future studies to gain insight into TSPO function.

In summary, this study provides irrefutable evidence that PBR/TSPO is not essential for steroid hormone biosynthesis and that Tspo^−/− mice are viable and fertile. These findings form a critical step toward understanding TSPO function and will help rebuild a defined model for steroid hormone biosynthesis. In addition, this understanding also forces reevaluation of interpretations made for TSPO function across multiple fields, including neuroscience, immunology, and oncology.

Note Added in Proof

In the original version of the manuscript that was published in JBC Papers in Press, the ACTB subpanels used to demonstrate equal protein loading in panels A, B, and E of Fig. 8 were incorrect. The correct ACTB images from the blots used to generate the original TSPO and STAR data in these panels are now provided in the revised Fig. 8. This correction does not change the interpretation of the results or the conclusions.