Inactivation of peroxisomal β-oxidation in mice, by knocking out multifunctional protein-2 (MFP-2; also called d-bifunctional enzyme), causes male infertility. In the testis, extensive accumulations of neutral lipids were observed in Sertoli cells, beginning in prepubertal mice and evolving in complete testicular atrophy by the age of 4 months. Spermatogenesis was already severely affected at the age of 5 wk, and pre- and postmeiotic germ cells gradually disappeared from the tubuli seminiferi. Based on cytochemical stainings and biochemical analyses, the lipid droplets consisted of cholesteryl esters and neutral glycerolipids. Furthermore, peroxisomal β-oxidation substrates, such as very-long-chain fatty acids and pristanic acid, accumulated in the testis, whereas the concentration of docosapentaenoic acid, a polyunsaturated fatty acid and peroxisomal β-oxidation product, was reduced. The testicular defects were also present in double MFP-2/peroxisome proliferator-activated receptor-α knockout mice, ruling out the possibility that they were mediated through the activation of this nuclear receptor. Immunoreactivity for peroxisomal proteins, including MFP-2, was detected in Sertoli cells as well as in germ cells and Leydig cells. The pivotal role of peroxisomal metabolism in Sertoli cells was also demonstrated by generating mice with a Sertoli cell-selective elimination of peroxisomes through cell type-specific inactivation of the peroxin 5 gene. These mice also developed lipid inclusions and were infertile, and their testes fully degenerated by the age of 4 months. In conclusion, the present data demonstrate that peroxisomal β-oxidation is essential for lipid homeostasis in the testis and for male fertility.
MULTIFUNCTIONAL PROTEIN-2 (MFP-2), also known as d-bifunctional enzyme, is involved in the peroxisomal β-oxidation of 2-methyl branched chain fatty acids (pristanic acid), C27-bile acid intermediates, very-long-chain fatty acids, and the synthesis of certain polyunsaturated fatty acids (PUFA), such as docosahexaenoic acid (C22:6n-3) and docosapentaenoic acid (C22:5n-6) (1–4). Since the first patient with a deficiency of MFP-2 was described (in retrospect) in 1986 (5, 6), more than 70 additional patients have been documented (7). Clinically, MFP-2-deficient patients strongly resemble patients with Zellweger syndrome, a generalized peroxisome biogenesis disorder (8). Patients with these disorders present with severe hypotonia, neuronal migration defects, facial dysmorphisms, and convulsions in the neonatal period and die within their first year of life.
We recently generated MFP-2 knockout mice as a tool to gain more insight into the in vivo role of MFP-2 in lipid metabolism. At birth, mice deficient in MFP-2 were indistinguishable from wild-type littermates (9). During the lactation period, however, MFP-2 knockouts lagged behind in growth, and at least one third died after 2 wk. Females, but not males, were fertile (9). In contrast to most patients with MFP-2 deficiency (7), neuronal migration defects and hypotonia were absent (10). Although the MFP-2 knockout mice are not a good phenocopy of the human disease, they do develop the metabolic abnormalities seen in patients, e.g. accumulations of very-long-chain and branched chain fatty acids (9, 11) and elevated ratios of immature C27/mature C24 bile acids (9, 12).
The early postnatal death of patients with peroxisome biogenesis defects or with MFP-2 deficiency does not allow us to examine maturation or functioning of the testis in these conditions. However, testicular defects were shown in X-linked adrenoleukodystrophy (X-ALD) and X-adrenomyeloneuropathy patients, who lack an ATP-binding cassette (ABC) transporter in the peroxisomal membrane, resulting in the accumulation of saturated very-long-chain fatty acids (13, 14). The lesions consist of lamellar lipid profiles in Leydig cells and some Leydig cell loss. Degenerative changes of seminiferous tubules, including maturation arrest and Sertoli cell abnormalities associated with infertility, were also described in these patients. In one patient who developed adult-onset cerebral X-ALD, severe impairment of spermatogenesis with rapid progression to azoospermia was reported (15). These defects were not recapitulated in X-ALD knockout mouse models (16–18). On the contrary, mouse models with acyl-coenzyme A (acyl-CoA) oxidase deficiency (19) or dihydroxyacetonephosphate acyltransferase deficiency (20), in which, respectively, straight chain fatty acid metabolism and ether phospholipid metabolism are impaired, displayed hypospermatogenesis or complete loss of spermatozoa, leading to male infertility.
In this paper the importance of MFP-2 for male reproductive function was investigated by histological and biochemical analysis of the testis of MFP-2 knockout mice. Because extensive lipid accumulation in Sertoli cells and fatty degeneration of the tubuli seminiferi were found, morphological analysis of the testis was also performed in MFP-2 knockout mice in a peroxisome proliferator-activated receptor α (PPARα)-deficient background and in mice with a Sertoli cell-selective depletion of peroxisomes.
Materials and Methods
Animal housing
Homozygous MFP-2-deficient mice were obtained in the offspring of heterozygous MFP-2+/− breeding pairs, which were in a mixed Swiss [Tac:[Sw]fBR/])/sv129 background. PPARα-knockout mice were obtained from F. Gonzalez (National Institutes of Health, Bethesda, MD).
Peroxin 5 (Pex5)FL/FL mice (21) were bred with anti-Mullerian hormone (AMH)-Cre mice (22), and in the next generation, AMH-Cre Pex5WT/FL mice were mated with Pex5FL/FL mice to obtain AMH-Cre Pex5FL/FL, also denoted SC-Pex5 knockout mice. Genotyping of MFP-2 (9), Pex5-loxP, and Cre mice (23) was performed as described. The PPARα alleles were identified by PCR using the primers 5′-CAGTGGGTGCAGCGCTGCGTCGGACTCGGTC-3′ and 5′-CGCCTTGGCCTTCTAAACATAGGC-3′.
Mice were bred in the animal-housing facility of University of Leuven under conventional conditions. They had unlimited access to standard rodent food chow (Muracon-G, Carfil Quality-Pavan Services, Oud-Turnhout, Belgium) and water and were kept on a 12-h light, 12-h dark cycle. All animal experiments were approved by the institutional animal ethical committee of University of Leuven.
Tissue collection
Mice were anesthetized with Nembutal (Sanofi, Brussels, Belgium). Testicles were dissected, snap-frozen in liquid nitrogen, and stored at −80 C.
Lipid analysis
All solvents were of the highest quality commercially available (Biosolve, Valkenswaard, The Netherlands). Butylated hydroxytoluene (0.05%, wt/vol) was added at all stages of the extraction to minimize autooxidation of PUFA.
Lipids were extracted from tissues, homogenized in 3.8 ml CH3OH/CHCl3/H2O (2:1:0.8) using a Polytron (Brinkmann Instruments, Inc., West Orange, NJ) tissue homogenizer (24), and separated into neutral lipids, fatty acids, and phospholipids by solid phase extraction (500-mg Bond Elut NH2 column; Varian Benelux, Sint-Katelijne-Waver, Belgium) (10, 25).
Cholesterol (26), cholesteryl esters (26), neutral glycerolipids (27), and phosphorus content of the phospholipid fraction (28) were determined as previously described. Phytanic, pristanic acid, C26:0, and C22:5n-6 were quantified by gas chromatography-mass spectrometry analysis in neutral lipids (10). Testosterone levels in plasma were measured using the Testo-RIA-CT kit (BioSource International, Camarillo, CA).
Immunocytochemistry of Sertoli cell cultures
Rat Sertoli cells were prepared essentially as previously described (29). After 3 d in culture, cells were fixed with 4% paraformaldehyde for 20 min and washed in 0.1 m PBS. Nonspecific binding sites were blocked with normal swine serum (20%, 1 h) diluted in 1% BSA in 0.1 m PBS. The latter solution was also used to dilute (1:100) the primary antibodies (all made in rabbits) raised against Pex14 (30), peroxisome membrane protein 70 (PMP70) (31), bovine catalase (Rockland, Gilbertsville, PA), rat acyl-CoA oxidase (32), MFP-2 (33), and 3-ketoacyl-CoA thiolase (34) and to dilute (1:2000) the Cy3-conjugated goat antirabbit IgGs (Fluka-Sigma-Aldrich Corp., Bornem, Belgium).
Morphological analyses
Adult nembutal-anesthetized mice were perfusion fixed with 4% paraformaldehyde in 0.1 m PBS. For each time point at least three different mice were used.
For lipid histochemistry, testes were embedded in increasing concentrations of gelatin (10–30%). Free-floating Vibratome sections (Leica, Nussloch, Germany) (10 μm thick) were stained with the lipid-soluble dye Oil Red O [no. 26125 (BDH Laboratory Supplies, Poole, UK); 0.24% (wt/vol)] in isopropanol/water (3:2) for 18 min. The detection of sterols was performed according to Mallory’s modification of the Schultz method (35). The lipase-lead sulfide technique described by Adams et al. (36) was used to detect neutral glycerolipids. Alternatively, 7-μm-thick frozen sections were used for Oil Red O stainings. For histochemical and immunohistochemical analyses, previously described procedures (37) were applied to paraffin-embedded tissues.
For electron microscopic analysis, mice were deeply anesthetized with Hypnorm (fentanyl/fluanizone and midazolam, Janssen Pharmaceutica, Beerse, Belgium) and transcardially perfused with Ringer’s solution containing 0.1% procain and 5 U/ml heparin (2 min), followed by cacodylate-buffered 2.5% glutaraldehyde (10 min). Specimens were postfixed for another 24 h with glutaraldehyde, osmicated, and embedded in epoxy resin (Embed 812; Electron Microscopy Sciences, Fort Washington, PA). Sections (3 μm) for light microscopy were stained with p-phenylenediamine, and thin sections for electron microscopy were stained with uranyl acetate and lead citrate and investigated in a Philips CM 100 microscope (Philips Electronic Instruments, Eindhoven, The Netherlands).
RNA analyses
Quantitative PCR analyses of the expression of Leydig cell markers were performed exactly as described previously (38). Northern blot analyses was also performed as previously described (39) using probes generated by RT-PCR from mouse liver RNA with the following primers: ABCA1 (5′-gcatccctcccggagagtgctttgg-3′ and 5′-aacggccacatccacaactgtctgg-3′), 3-hydroxy-3-methyl-glutaryl-CoA reductase (5′-tgctgaagcttcaggagttctttcc-3′ and 5′-aagctgccttcttggtgcacgttcc-3′), scavenger receptor B1 (SRB1; 5′-ccggaggcatgcaggtccatgaagc-3′ and 5′-gcccttggcagctggtgacatcagg-3′), and low-density lipoprotein receptor (LDLR; 5′-cattttcagtgccaatcgactcacg-3′ and 5′-tgccacatcgtcctccaggctgacc-3′).
Results
Fatty degeneration of tubuli seminiferi
MFP-2 knockout mice, which were severely growth retarded during the lactation period (9), resumed weight gain after weaning, but they remained smaller than their wild-type littermates. Five different male MFP-2−/− mice, aged 6–20 wk, were mated with wild-type Swiss females; none sired offspring, although vaginal plugs were frequently observed.
The testes of 5-month-old knockout mice were normally descended in the scrotum, but were strongly reduced in size. Upon histological examination of semithin sections, severe atrophy of the seminiferous tubules, filled with osmium tetroxide-reactive material, and the complete absence of germ cells were found (Fig. 1, A and B). To decipher the origin of this lipid atrophy, testes were analyzed at different stages of development. In 10-d-old (Fig. 1, C and D) and 3-wk-old (not shown) mice, lipid droplets could be visualized with Oil Red O in the center of otherwise normal seminiferous tubules, indicative of neutral lipid accumulations. At the age of 5 wk, lipid droplets were present at the periphery of the tubuli (Fig. 1E), compatible with Sertoli cell localization, and spermatogenesis was already severely affected. Some tubuli contained spermatozoa in their lumen, which displayed normal structures based on electron microscopic analysis, but overall they were clearly less numerous than in wild-type mice of the same age. Rounded and elongated spermatids were still present in most tubuli (Fig. 1, F–H). In 7-wk-old mice, the lumen of the tubules was mostly empty or contained fibrous debris. Very few tubules contained elongated spermatids, but multinucleated cells were often found. Most tubuli had lost their typical multilayered appearance (Fig. 1, I and J). The epididymi were devoid of mature spermatozoa, but contained immature germ cells and debris (Fig. 1K). At the age of 12 wk (Fig. 1L) and 16 wk (not shown), the tubuli consisted of only a few cell layers, with big vacuoles appearing between the remaining cells. In some MFP-2 knockout mice, tubular degeneration progressed more slowly, but all mice were infertile.
Histological analyses of testes and epididymi of MFP-2 knockout and control mice. A and B, Semithin sections of testes of 5-month-old mice. Although in control mice (A) only Leydig cells are intensely stained due to their high lipid content, the tubuli seminiferi of MFP-2 knockout mice are completely filled with fat, which is visualized as black deposits due to the osmification procedure. C and D, Light microscopy of frozen sections stained with Oil Red O reveals accumulations of neutral lipids in the center of the tubuli seminiferi of 10-d-old MFP-2 knockout mice (D), compatible with a Sertoli cell localization, but not in control mice (C). E, Oil Red O staining of testis of 5-wk-old MFP-2 knockout mice, showing large lipid droplets in the periphery of the tubuli. F–H, Hematoxylin-eosin staining of tubuli seminiferi of 5-wk-old control (F) and MFP-2 knockout mice (G and H), documenting that spermatogenesis is already affected at this age. In 7-wk-old (I and J) and 12-wk-old (L) MFP-2 knockout mice, the typical cell architecture of the tubuli is gradually lost. Arrowheads in G–J point to white holes in the outer layer of the tubuli in MFP-2 knockout mice, which represent lipid droplets that were dissolved during the embedding procedure. K, The caput epididymi of 7-wk-old MFP-2 knockout mice are not filled with mature spermatozoa, but contain immature germ cells. M, Electron micrograph of testis of 8-wk-old MFP-2 knockout mouse depicting a Sertoli cell. The black dots correspond to lipid droplets due to the reaction of unsaturated fatty acids with osmium tetroxide. Arrowheads point to small lipid droplets that are also present in Sertoli cells of control mice (not shown), whereas the arrows indicate a huge and abnormal lipid droplet present in the basal part of a Sertoli cell. N and O, Vibratome sections (20 μm) of gelatin-embedded testes of 8-wk-old MFP-2 knockout mice were stained using the lipase-lead sulfide-based technique (N) and the Schultz test (O) to detect triglycerides and cholesteryl esters, respectively. Bars: A and B, 50 μm; C–L, N, and O, 100 μm; M, 10 μm.
Characterization of the lipid droplets
Using electron microscopy (Fig. 1M) osmiophilic inclusions with a diameter of several micrometers were found in Sertoli cells of 8-wk-old knockout mice, but not in wild-type mice. These lipid inclusions were globular and not membrane bound, and were surrounded by numerous small osmiophilic lipid droplets. The latter inclusions were obviously normal constituents of Sertoli cells, because they were present in wild-type mice as well (data not shown).
The chemical nature of the droplets was also examined by histological procedures. The inclusions stained positively with Oil Red O at all ages, indicating that they consisted of neutral lipids such as triglycerides and cholesteryl esters (Fig. 1, D and E, and data not shown). This was confirmed by pretreating the sections with anhydrous acetone, which dissolves neutral lipids, and resulted in strongly decreased Oil Red O staining. Both the lipase-lead sulfide-based technique identifying neutral glycerolipids (Fig. 1N) and the Schultz test, which specifically detects sterols (Fig. 1O), revealed positively stained droplets with the same distribution as the Oil Red O-positive droplets.
Identification and quantification of lipids in testis
In testes from 3-month-old mice, an increase in acylglycerides and cholesteryl esters was observed compared with wild-type mice. The relative increase in cholesteryl esters (3-fold) was larger than that in neutral glycerolipids (2-fold), although in absolute values, more neutral glycerolipids were present. No differences in free cholesterol and phospholipid levels could be observed between knockout and wild type mice (Table 1). These data are in accordance with the lipid histochemistry findings.
Levels of different lipid classes in testes of heterozygous and MFP-2 knockout mice
| Genotype | Phospholipids | Neutral glycerolipids | Free cholesterol | Cholesteryl esters |
|---|---|---|---|---|
| +/− | 15.37 ± 0.69 | 4.52 ± 0.68 | 6.25 ± 0.34 | 1.06 ± 0.05 |
| −/− | 14.97 ± 0.12 | 8.31 ± 0.54a | 6.22 ± 0.30 | 3.69 ± 0.63b |
| Genotype | Phospholipids | Neutral glycerolipids | Free cholesterol | Cholesteryl esters |
|---|---|---|---|---|
| +/− | 15.37 ± 0.69 | 4.52 ± 0.68 | 6.25 ± 0.34 | 1.06 ± 0.05 |
| −/− | 14.97 ± 0.12 | 8.31 ± 0.54a | 6.22 ± 0.30 | 3.69 ± 0.63b |
Single testes from four different 3-month-old mice were analyzed. Values are the mean ± sem, expressed as nanomoles per milligram of tissue.
P < 0.005, comparing the same lipid fraction of +/− with −/−.
P < 0.01, comparing the same lipid fraction of +/− with −/−.
Levels of different lipid classes in testes of heterozygous and MFP-2 knockout mice
| Genotype | Phospholipids | Neutral glycerolipids | Free cholesterol | Cholesteryl esters |
|---|---|---|---|---|
| +/− | 15.37 ± 0.69 | 4.52 ± 0.68 | 6.25 ± 0.34 | 1.06 ± 0.05 |
| −/− | 14.97 ± 0.12 | 8.31 ± 0.54a | 6.22 ± 0.30 | 3.69 ± 0.63b |
| Genotype | Phospholipids | Neutral glycerolipids | Free cholesterol | Cholesteryl esters |
|---|---|---|---|---|
| +/− | 15.37 ± 0.69 | 4.52 ± 0.68 | 6.25 ± 0.34 | 1.06 ± 0.05 |
| −/− | 14.97 ± 0.12 | 8.31 ± 0.54a | 6.22 ± 0.30 | 3.69 ± 0.63b |
Single testes from four different 3-month-old mice were analyzed. Values are the mean ± sem, expressed as nanomoles per milligram of tissue.
P < 0.005, comparing the same lipid fraction of +/− with −/−.
P < 0.01, comparing the same lipid fraction of +/− with −/−.
In view of the role of MFP-2 in the degradation of very-long-chain fatty acid and 2-methyl branched chain fatty acids, levels of C26:0 and pristanic acid as well as its 3-methyl precursor, phytanic acid, were analyzed in the neutral lipid fraction. The concentration of C26:0 increased 2- to 3-fold, and pristanic acid levels doubled, but phytanic acid levels were only marginally increased in MFP-2 knockout mice compared with wild-type controls.
In extracts from MFP-2 knockout mice, low amounts of some additional fatty acids appeared that were not present in extracts from heterozygous or wild-type littermates (data not shown). Based on the mass spectra of the additional peaks, these fatty acids are probably very-long-chain PUFAs.
It was previously reported that the synthesis of the PUFAs C22:6n-3 and C22:5n-6 requires a final single peroxisomal β-oxidation cycle after elongation and desaturation steps in the endoplasmic reticulum and that MFP-2 is involved in this process (4, 40). These PUFAs are important constituents of testicular phospholipids in man and rodents, respectively. In line with this biosynthetic pathway, C22:5n-6 levels were severely decreased (0.33 vs. 2.16 nmol/mg tissue; Table 2) in the phospholipids of testes of MFP-2 knockout mice compared with wild-type mice. To exclude that other peroxisomal functions might be affected in the MFP-2 knockout testes, plasmalogen concentrations were measured. No difference in plasmalogen content was observed between MFP-2 knockout and wild-type testis (data not shown).
Levels of branched chain fatty acids and very-long-chain fatty acids in neutral lipids and of C22:5n-6 in phospholipids
| Genotype | Pristanic acid (pmol/mg tissue) | Phytanic acid (pmol/mg tissue) | C26:0 (pmol/mg tissue) | C22:5n-6 (nmol/mg tissue) |
|---|---|---|---|---|
| +/− | 3.2 ± 0.1 (4) | 6.5 ± 0.5 (4) | 131.4 ± 5.4 (4) | 2.2 ± 0.6 (3) |
| −/− | 8.6 ± 0.6 (4)a | 7.5 ± 0.6 (4) | 321.8 ± 7.5 (4)b | 0.3 ± 0.1 (3)a |
| Genotype | Pristanic acid (pmol/mg tissue) | Phytanic acid (pmol/mg tissue) | C26:0 (pmol/mg tissue) | C22:5n-6 (nmol/mg tissue) |
|---|---|---|---|---|
| +/− | 3.2 ± 0.1 (4) | 6.5 ± 0.5 (4) | 131.4 ± 5.4 (4) | 2.2 ± 0.6 (3) |
| −/− | 8.6 ± 0.6 (4)a | 7.5 ± 0.6 (4) | 321.8 ± 7.5 (4)b | 0.3 ± 0.1 (3)a |
Testes from 3-month-old mice were analyzed. Values are the mean ± sem; the number of samples is given in parentheses.
P < 0.01, +/− vs. −/−.
P < 0.001, +/− vs. −/−.
Levels of branched chain fatty acids and very-long-chain fatty acids in neutral lipids and of C22:5n-6 in phospholipids
| Genotype | Pristanic acid (pmol/mg tissue) | Phytanic acid (pmol/mg tissue) | C26:0 (pmol/mg tissue) | C22:5n-6 (nmol/mg tissue) |
|---|---|---|---|---|
| +/− | 3.2 ± 0.1 (4) | 6.5 ± 0.5 (4) | 131.4 ± 5.4 (4) | 2.2 ± 0.6 (3) |
| −/− | 8.6 ± 0.6 (4)a | 7.5 ± 0.6 (4) | 321.8 ± 7.5 (4)b | 0.3 ± 0.1 (3)a |
| Genotype | Pristanic acid (pmol/mg tissue) | Phytanic acid (pmol/mg tissue) | C26:0 (pmol/mg tissue) | C22:5n-6 (nmol/mg tissue) |
|---|---|---|---|---|
| +/− | 3.2 ± 0.1 (4) | 6.5 ± 0.5 (4) | 131.4 ± 5.4 (4) | 2.2 ± 0.6 (3) |
| −/− | 8.6 ± 0.6 (4)a | 7.5 ± 0.6 (4) | 321.8 ± 7.5 (4)b | 0.3 ± 0.1 (3)a |
Testes from 3-month-old mice were analyzed. Values are the mean ± sem; the number of samples is given in parentheses.
P < 0.01, +/− vs. −/−.
P < 0.001, +/− vs. −/−.
Leydig cell function
Upon histological examination, interstitial Leydig cells looked normal until at least the age of 12 wk, although their numbers were difficult to score due to the disintegrating seminiferous epithelium. To test Leydig cell function, plasma testosterone levels were measured under basal conditions and 2 h after the injection of 0.4 IU human chorionic gonadotropin/g body weight in 6-wk-old mice. The surge of testosterone levels was comparable in MFP-2 knockout and control mice (Fig. 2A). In addition, the expression of the Leydig cell-specific genes, 3β-HSD1, P450 cholesterol side-chain cleavage enzyme, and p450c17 (17α-hydroxylase/C17–20 lyase), was assayed by quantitative PCR (Fig. 2, B–D) (38). The transcript levels tended to be lower in MFP-2 knockout testis, although this did not reach statistical significance. Although it cannot be excluded that Leydig cells are affected, potential abnormalities appear to be less dramatic than those observed in the germinative epithelium.
Leydig cell function in MFP-2 knockout mice. A, Testosterone levels in plasma of 6-wk-old mice 2 h before and after injection of human chorionic gonadotropin (hCG). B, The expression of Leydig cell markers was assessed by quantitative PCR. Single testes from four knockout (KO) and four control (Ct) mice were used for each analysis. □, Controls; ▦, MFP-2 knockout mice.
Peroxisomes in Sertoli cells
In view of the lipid inclusions in Sertoli cells, and the fact that Leydig cells were the only testicular cells in which peroxisomes were identified in rodents to date (41), we reinvestigated the localization of peroxisomal markers in testis sections and Sertoli cell cultures. Immunofluorescence analyses of the peroxisomal membrane proteins PMP70 and Pex14p (Fig. 3, A and B) revealed very intense staining in the outer cell layer of the tubuli seminiferi and weaker staining in the center layers and in Leydig cells of 8-wk-old wild-type mice. In contrast, the peroxisomal marker enzyme catalase was more abundantly present in interstitial cells than in the tubuli (Fig. 3C). Primary cultures of Sertoli cells isolated from rats were also immunopositive for PMP70 and Pex14 (Fig. 3, D and E), catalase (Fig. 3F), and other peroxisomal enzymes (MFP-2, acyl-CoA oxidase, and 3-ketoacyl-CoA thiolase; not shown).
Presence of peroxisomal markers in tubuli seminiferi and cultured rat Sertoli cells. The peroxisomal membrane proteins PMP70 (A) and Pex14p (B) were abundantly present in the outer layer of the seminiferous epithelium of wild-type 8-wk-old mice, but lower intensity staining was also present in more central layers and in Leydig cells. Conversely, catalase was more abundantly present in Leydig cells than in tubuli seminiferi (C). The inset in C represents staining after omitting primary antibody. D–F, Immunofluorescent detection of peroxisomal markers in cultured rat Sertoli cells revealed a punctate pattern that is typical for an intraperoxisomal location of these proteins. Individual Sertoli cells are shown. Bars, 100 μm.
Origin of cholesteryl esters in Sertoli cells
Several routes can be considered to explain the excessive accumulation of cholesteryl esters in Sertoli cells. The possibilities of increased cholesterol synthesis, increased uptake via LDLR or SRB1 (42), and impaired export via ABCA1 were tested by conducting Northern blot analyses. However, no clear changes in expression levels of 3-hydroxy-3-methyl-glutaryl-CoA reductase, LDLR, SRB1, or ABCA1 were detected in 10-d,- 3-wk-, and 6-wk-old MFP-2 knockout mice (data not shown).
Elimination of peroxisomes from Sertoli cells
The selective accumulation of lipids in Sertoli cells and the fact that peroxisomal markers were abundantly present in these cells suggested that defective peroxisomal β-oxidation in Sertoli cells might be the primary cause of male infertility and testicular disintegration. This was also investigated by generating a mouse model with selective elimination of peroxisomes from Sertoli cells by inbreeding Pex5-loxP mice (21, 23) with AMH-Cre (22) mice. Pex5p is the receptor that is essential for the import of all peroxisomal matrix proteins (43). AMH-Cre mice were amply shown to direct a complete and Sertoli cell-selective recombination of floxed target genes (22, 44, 45). Large numbers of lipid droplets were detected in the tubuli seminiferi of SC-Pex5 knockout mice at the age of 10 d (Fig. 4, A and B) and 7 wk (Fig. 4, C and D) comparable to the lipid accumulations in MFP-2 knockout mice. However, in 7-wk-old SC-Pex5 knockout mice, the seminiferous epithelium appeared to be less affected than in MFP-2 knockout mice of the same age, because more spermatozoa were present in the lumen and epididymi (Fig. 4, E–H). In the latter tissue, many rounded germ cells were also observed. Nevertheless, upon mating six different SC-Pex5 knockout mice, aged 6–13 wk, with wild-type Swiss mice, no offspring were generated despite the fact that vaginal plugs were found. Progressive degeneration of the tubuli seminiferi was observed in 9- to 11-wk-old SC-Pex5 knockout mice (Fig. 4, I and J), resulting in testicular weights 30–40% those of control littermates at the age of 14 wk, although the onset and progression of testicular degeneration were somewhat variable.
Histological analyses of testes and epididymi of SC-Pex5 knockout mice and controls. Oil Red O staining of frozen sections of 10-d-old (A and B) and 7-wk-old (C and D) mice. Note the extensive lipid accumulations (arrows) in the center (B) and outer layers (D) of the tubuli seminiferi in SC-Pex5 knockout mice. A and C, Littermate controls. Hematoxylin-eosin staining of 7-wk-old testis (E and F) and caput epididymi (G and H) revealed empty vesicles in the outer layer of the seminiferous epithelium (arrowheads), reduced numbers of spermatozoa in the lumen of the tubuli (F), and immature, rounded germ cells in the caput epididymi (H) of SC-Pex5 knockout mice. E and G, Littermate controls. In 9- and 11-wk-old SC-Pex5 knockout mice (I and J), the seminiferous epithelium was degenerating. Bars, 100 μm.
Infertility is not mediated by activation of PPARα
Because male infertility in mice deficient in the first enzyme of the peroxisomal β-oxidation pathway, acyl-CoA-oxidase (Acox1) (46), was fully dependent on PPARα activation, we tested whether this could also be the case for MFP-2 knockout mice. Male MFP2/PPARα double-knockout mice were indistinguishable from MFP-2 knockout mice in that they did not produce offspring, and they exhibited the same kind of lipid accumulation in Sertoli cells as well as testicular degeneration (Fig. 5).
Histological analyses of testes of 7-wk-old MFP-2/PPARα knockout mice. A and B, Oil Red O staining of frozen sections of testes of MFP-2+/+PPARα−/− (A) compared with MFP-2−/−PPARα−/− (B) mice revealed lipid accumulations in the latter mouse strain. C and D, Spermatogenesis was severely affected in the tubuli of MFP-2−/−PPARα−/− mice (D). Bars, 100 μm.
Discussion
Deficiency of the peroxisomal β-oxidation enzyme MFP-2 in humans causes a severe developmental problem, including cortical malformations, severe hypotonia, and early postnatal death (7). These developmental abnormalities were not reproduced in MFP-2-deficient mice (9), but, instead, the surviving adult mice developed pathologies in several organs, including the male reproductive system.
Peroxisomes and Sertoli cells
Male infertility of MFP-2-deficient mice was accompanied by the appearance of large lipid droplets in the tubuli seminiferi, which were already present on postnatal d 10. In young adulthood, some mature spermatozoa could be detected in the lumen of the tubuli, and Leydig cells looked normal in the interstitia. However, the progressive accretion of lipids was followed by complete atrophy of the seminiferous epithelium and fatty degeneration of the testis in 5-month-old mice. The cell type accumulating the lipids was shown to be the Sertoli cell based on their shape, size, and position in the tubuli and on electron microscopic analysis. The reason why Sertoli cells are histopathologically most affected in testes of MFP-2 knockout mice is not clear. In fact, little was known about the presence of peroxisomes or the expression of peroxisomal enzymes in Sertoli cells, whereas Leydig cells of several species were shown to contain peroxisomes, mostly based on the presence of catalase (41, 47–50). Recently, it was shown that α-methylacyl-CoA racemase, a β-oxidation enzyme with a dual location in peroxisomes and mitochondria (30, 51), is expressed in Sertoli cells (52). Until now, the expression of MFP-2 in Sertoli cells was only reported in pigs (53). Using antibodies against several peroxisomal membrane proteins and matrix enzymes, we could clearly demonstrate that peroxisomes are abundantly present in the outer and, to a lesser extent, the inner layers of tubuli seminiferi and in primary cultures of rat Sertoli cells. Together, these data underscore that the local defect of peroxisomal β-oxidation in Sertoli cells can be responsible for the accumulating lipids. The fact that Sertoli cells are selectively affected is most surprising in view of the testicular pathology in two other mouse models with peroxisomal defects. In mice lacking acyl-CoA oxidase, another peroxisomal β-oxidation enzyme, a reduction in Leydig cell populations, hypospermatogenesis, and infertility, but no changes in Sertoli cells, were reported (19). The differences in the pathology seen in the testis of acyl-CoA oxidase and MFP-2 knockout mice may be due to the broader substrate specificity of MFP-2, which is involved in the metabolism not only of straight chain, but also of branched chain compounds (3). Inactivity of the peroxisomal enzyme, dihydroxyacetone phosphate acyltransferase, which is necessary for the synthesis of ether phospholipids, caused an arrest of spermatogenesis before the stage of round spermatids, resulting in the complete absence of spermatozoa, a Sertoli cell-only phenotype in tubuli seminiferi, and Leydig cell proliferation in the intertubular space (20).
Link between lipid accumulations in Sertoli cells and defective spermatogenesis
The presence of large lipid deposits in the basal cytoplasm of Sertoli cells has been described in several pathological conditions associated with defective spermatogenesis (for review, see Ref.54). This was related to the phagocytic activity of Sertoli cells that reabsorb residual bodies and, under pathological conditions, germ cells undergoing degeneration (55). In view of the extensive lipid accumulations in the testes of 10-d-old MFP-2 knockout mice long before the completion of the first spermatogenic cycle, it seems unlikely that the earliest lipid accumulations are due to increased phagocytic activity secondary to germ cell degeneration. Conclusive evidence that the lipid accumulations are inherent to Sertoli cells was provided by the selective inactivation of the peroxisome biogenesis protein Pex5p in these cells using the anti-Mullerian hormone promoter driving Cre expression. Abundant lipid accumulations were already present in 10-d-old SC-Pex5 knockout mice. Together, the data indicate that Sertoli cells require peroxisomal β-oxidation activity for their lipid homeostasis.
In several other recently reported mouse models, a Sertoli cell-confined disturbance of lipid metabolism accompanied male hypo- or infertility. Mice deficient in the nuclear receptors retinoid X receptor β (RXRβ) (56, 57) and liver X receptor β (57, 58); transcriptional intermediary factor 2, a nuclear receptor coactivator (59); and Cnot7, a regulator of RXRβ (60) and mice lacking ABCA1, a transporter that shuttles excess cholesterol and phospholipids out of cells (61), all accumulate abnormally large neutral lipid droplets in Sertoli cells. In comparison with these models, the fatty degeneration of the testes in MFP-2 knockout mice evolved much more quickly; they were already fully atrophic by the age of 5 months. Although defects in the lipid homeostasis of Sertoli cells coincide with spermatogenic arrest in several knockout models, there are a few indications that lipid accumulations in Sertoli cells might not be the only cause of infertility. First, RXRβ knockout mice develop both lipid accumulations and spermatogenic arrest, whereas mice in which only the transcription activation function II of RXRβ was deleted show similar lipid accumulations, but normal fertility until the age of 12 months (57). Secondly, in transcriptional intermediary factor 2-null mutants, lipid droplets were observed in tubules exhibiting signs of degeneration as well as in those with normal germ cell associations (59).
Although some variability in the onset and progression of spermatogenic disruption and disintegration of the tubuli seminiferi was noticed in MFP-2 as well as in SC-Pex5 knockout mice, MFP-2 knockout mice appeared to be affected earlier. This might indicate that peroxisomal β-oxidation is important not only in Sertoli cells, but also in developing germ cells and/or Leydig cells. This is in agreement with our findings and with recently published immunohistochemical stainings (62) documenting the presence of peroxisomes in all testicular cell types.
Origin and nature of lipid accumulations
The oil droplets were identified as acylglycerides and cholesteryl esters by both lipid histochemical and biochemical analyses. Electron micrographs of the droplets in the Sertoli cells of MFP-2 knockout mice suggest that the volume of the huge droplets is 100- to 1000-fold greater than that of the lipid droplets found in wild-type mice. In the neutral lipid fraction, consisting of cholesterol, cholesteryl esters, acylglycerides, and ceramides, the peroxisomal β-oxidation substrates C26:0 and pristanic acid doubled in testes of knockout mice compared with the levels found in wild-type animals. In absolute values, the levels of branched fatty acids in the neutral lipids are about 40 times lower than those of C26:0, which, in turn, constitutes only a minor fraction of the total fatty acid content of the testis. In view of the huge increase in fat content of the testis and the low levels of accumulating substrates of the peroxisomal β-oxidation pathway, we assume that the lipid droplets are primarily composed of the most common fatty acids, i.e. saturated and unsaturated C16 and C18 species.
The more severe phenotype of MFP-2 compared with ABCA1 knockout mice (61) rules out that the inability to export cholesteryl esters from the cell is the single cause of the lipid accumulations. Additional analysis of gene expression by Northern blot analysis confirmed unaltered levels of ABCA1 transcripts and did not support increased cholesterol synthesis or increased uptake via LDLR or SRB1 (42). This leaves unanswered the question of the source of the cholesteryl esters. Some caution should be used, however, because the gene expression data were derived from the whole testis, which could mask changes taking place in Sertoli cells only.
Peroxisome proliferators, which are known testicular toxicants, might exert their effects by interfering with retinoic acid receptor α transcriptional activity (63). Because increased levels of PPARα-regulated proteins were found in liver of MFP-2 knockout mice (9), we tested whether testicular degeneration could be triggered by the activation of this nuclear receptor through endogenously accumulating ligands. However, testicular abnormalities were precisely the same in MFP-2 and double MFP-2/PPARα knockout mice, which is again in sharp contrast with those in acyl-CoA oxidase deficient mice. Indeed, the latter mice regained male fertility in a PPARα-null background.
In conclusion, a block of peroxisomal β-oxidation at the level of MFP-2 in the mouse causes lipid accumulations and degeneration of the testis, proving that this enzyme is essential for the lipid homeostasis in Sertoli cells and the integrity of the seminiferous epithelium. The exact relationship between the absence of MFP-2 and neutral lipid accumulation remains obscure, but it is not impossible that nuclear receptors other than PPARα with a regulatory function in lipid homeostasis become dysregulated by the unmetabolized substrates of peroxisomal fatty acid oxidation.
Acknowledgments
We thank Prof. F. Gonzalez for providing PPARα knockout mice, Prof. M. Fransen for Pex14 antibodies, Prof. J. Billen for testosterone measurements, and Els Meyhi, Lies Pauwels, Benno Das, and Ludo Deboel for excellent technical assistance.
This work was supported by grants from Fonds Wetenschappelijk Onderzoek-Vlaanderen (G.0235.01), Geconcerteerde Onderzoeksacties (99/09 and 2004/08), and the European Union (Mouse Models of Peroxisomal Disorders, QLG1-CT2001-01277, FP5 and Peroxisomes, LSHG-CT-2004-512018, FP6).
S.H., H.S., K.D.G., G.V., F.G., P.P.V.V., and M.B. have nothing to declare.
Abbreviations
- ABC
ATP-binding cassette
- AMH
anti-Mullerian hormone
- CoA
coenzyme A
- 3β-HSD1
3β-hydroxysteroid dehydrogenase
- LDLR
low-density lipoprotein receptor
- MFP-2
multifunctional protein 2
- Pex14
peroxin 14
- PMP70
peroxisome membrane protein 70
- PPAR
peroxisome proliferator-activated receptor
- PUFA
polyunsaturated fatty acid
- RXR
retinoid X receptor
- SR
scavenger receptor
- X-ALD
X-linked adrenoleukodystrophy.
