The Testicular Form of Hormone-sensitive Lipase HSLtes Confers Rescue of Male Infertility in HSL-deficient Mice*

Inactivation of the hormone-sensitive lipase gene (HSL) confers male sterility with a major defect in spermatogenesis. Several forms of HSL are expressed in testis. HSLtes mRNA and protein are found in early and elongated spermatids, respectively. The other forms are expressed in diploid germ cells and interstitial cells of the testis. To determine whether the absence of the testis-specific form of HSL, HSLtes, was responsible for the infertility in HSL-null mice, we generated transgenic mice expressing HSLtes under the control of its own promoter. The transgenic animals were crossed with HSL-null mice to produce mice deficient in HSL in nongonadal tissues but expressing HSLtes in haploid germ cells. Cholesteryl ester hydrolase activity was almost completely blunted in HSL-deficient testis. Mice with one allele of the transgene showed an increase in enzymatic activity and a small elevation in the production of spermatozoa. The few fertile hemizygous male mice produced litters of very small to small size. The presence of the two alleles led to a doubling in cholesteryl ester hydrolase activity, which represented 25% of the wild type values associated with a qualitatively normal spermatogenesis and a partial restoration of sperm reserves. The fertility of these mice was totally restored with normal litter sizes. In line with the importance of the esterase activity, HSLtes transgene expression reversed the cholesteryl ester accumulation observed in HSL-null mice. Therefore, expression of HSLtes and cognate cholesteryl ester hydrolase activity leads to a rescue of the infertility observed in HSL-deficient male mice.

Inactivation of the hormone-sensitive lipase gene (HSL) confers male sterility with a major defect in spermatogenesis. Several forms of HSL are expressed in testis. HSL tes mRNA and protein are found in early and elongated spermatids, respectively. The other forms are expressed in diploid germ cells and interstitial cells of the testis. To determine whether the absence of the testis-specific form of HSL, HSL tes , was responsible for the infertility in HSL-null mice, we generated transgenic mice expressing HSL tes under the control of its own promoter. The transgenic animals were crossed with HSL-null mice to produce mice deficient in HSL in nongonadal tissues but expressing HSL tes in haploid germ cells. Cholesteryl ester hydrolase activity was almost completely blunted in HSL-deficient testis. Mice with one allele of the transgene showed an increase in enzymatic activity and a small elevation in the production of spermatozoa. The few fertile hemizygous male mice produced litters of very small to small size. The presence of the two alleles led to a doubling in cholesteryl ester hydrolase activity, which represented 25% of the wild type values associated with a qualitatively normal spermatogenesis and a partial restoration of sperm reserves. The fertility of these mice was totally restored with normal litter sizes. In line with the importance of the esterase activity, HSL tes transgene expression reversed the cholesteryl ester accumulation observed in HSL-null mice. Therefore, expression of HSL tes and cognate cholesteryl ester hydrolase activity leads to a rescue of the infertility observed in HSL-deficient male mice.
Hormone-sensitive lipase (HSL) 1 is an intracellular enzyme with a broad substrate specificity. HSL catalyzes the hydrolysis of tri-, di-, and monoacylglycerols, cholesteryl, and retinyl es-ters, as well as other lipid and water-soluble molecules (1). A testicular form of the enzyme HSL tes is expressed in rodent and human testis (2). HSL tes mRNA expression is high in early spermatids (3). Immunolocalization of the protein in human and rodent seminiferous tubules showed that the highest level of expression occurred in elongated spermatids. Other forms of HSL are expressed elsewhere in the testis. HSL-like immunoreactivity is observed in the cytoplasm of type B spermatogonia (human and mouse), primary spermatocytes (human and mouse), and Sertoli cells (human) (3,4). The 3.9-kb human HSL tes mRNA is translated into 1068and 1076-amino acid proteins in rat and humans, respectively (2). HSL tes contains a unique NH 2 -terminal domain in addition to the 775 amino acids common to all forms of HSL (5,6). This additional domain is encoded by a testis-specific exon located 15 kb upstream of the first of the 9 exons common to all known HSL isoforms. The 5Ј-region of HSL tes mRNA is distinct from that of the other human HSL mRNA expressed in testicular diploid cells. The genomic organization of HSL tes suggested, as often seen when a gene is expressed in somatic cells and in haploid germ cells, the use of different promoters to govern tissue-specific expression. Our transgenic studies of the 5Јflanking region of the human HSL tes -specific exon revealed that 95 bp upstream of the transcription start site is sufficient for expression of a reporter gene in mouse testis but not in other HSL-expressing tissues (3,7). Therefore, a short promoter is essential for testis expression of HSL tes .
The role of HSL in testis was revealed by the phenotype of HSL-deficient mice (8,9). Male mice homozygous for the mutant allele (HSL Ϫ/Ϫ ) are sterile. Abnormalities in spermatogenesis result in profound alterations of spermatid maturation and oligospermia. To determine the importance of HSL tes in male sterility, we generated a transgenic line (HSL tes ϩ/? ) expressing the human HSL tes under the control of its own promoter and, through intercross with HSL Ϫ/Ϫ mice, produced animals deficient for HSL in all tissues but haploid germ cells (HSL Ϫ/Ϫ HSL tes ϩ/Ϫ , and HSL Ϫ/Ϫ HSL tes ϩ/ϩ ). Our data show that HSL tes is the sole HSL form responsible for impaired spermatogenesis. A comparison of HSL Ϫ / Ϫ HSL tes ϩ/Ϫ and HSL Ϫ/Ϫ HSL tes ϩ/ϩ animals revealed the ratelimiting role of HSL tes and the importance of cholesteryl ester hydrolase (CEH) activity in the action of HSL on spermatogenesis and more specifically on spermiogenesis.

EXPERIMENTAL PROCEDURES
HSL tes Transgenic and HSL-null Mice-The studies followed the INSERM and Louis Bugnard Institute Animal Facility guidelines and were approved by the local Animal Ethics Committee in Lund, Sweden.
The HSL tes transgene was constructed from a 1.6-kb HindIII/BglII fragment containing 1.4 kb of the 5Ј-flanking region upstream of the testis-specific exon, a 1-kb BglII/XhoI human HSL tes cDNA piece containing the testis-specific exon and part of exon 1, a 2.6-kb XhoI/NotI fragment containing the 3Ј-part of exon 1, intron 1, exon 2, and intron 2, the 5Ј-end of exon 3, and a 2.3-kb NotI/EcoRI human HSL tes cDNA piece containing exons 3-9 and the 3Ј-untranslated region (Fig. 1). The vector pSL301 (Invitrogen, Carlsbad, CA) containing the HSL tes construct was linearized with HindIII and EcoRI. Transgenic mice were generated after microinjection of this construct in fertilized oocytes of female from the B6D2/F1 strain (Elevage Janvier). Screening of the founders was made, after tail genomic DNA extraction, by PCR analysis with the following conditions: sense primer (5Ј-GCAAAGACGGACCA-CTCCA-3Ј) and antisense primer (5Ј-GACGTCTCGGAGTTTCCCCTC-AG-3Ј) with 56°C as annealing temperature. The size of the amplicon was 277 bp. Two transgenic lines (lines 29 and 32) were established. Line 29 was used in this study. HSL-null mice were generated by targeted disruption of the HSL gene in 129SV-derived embryonic stem cells by standard procedures as described elsewhere (10,11). In brief, the cDNA encoding the Aequorea victoria green fluorescent protein was inserted in-frame into exon 5 of the HSL gene, followed by a neomycin resistance gene, thereby disrupting the catalytic domain. Two recombinant embryonic stem cell colonies were used for generation of two independent HSL-null mouse lines. There was no phenotypic difference between the two lines. A single line was used for subsequent studies. We developed a PCR assay with three primers in order to distinguish HSL Ϫ/Ϫ from HSL ϩ/Ϫ or HSL ϩ/ϩ mice. The sense primer, 5Ј-ACTCAA-CAGCCTGGCAAAAT-3Ј, was common to both alleles; the antisense primers used to discriminate the wild type (WT) and targeted alleles were, respectively: 5Ј-AGGTCACAGTGCTTGACAGC-3Ј and 5Ј-GCTG-AACTTGTGGCCGTTTA-3Ј. Primers were used at 500 pM (sense) and 250 pM (antisense) with 2.5 mM MgCl 2 . The denaturation step was at 94°C for 30 s, the annealing step at 52°C for 30 s, and the elongation step at 72°C for 30 s. PCR product sizes were, respectively, 290 and 353 bp.
Intercross of HSL tes Transgenic Mice with HSL-null Mice-First we mated transgenic animals of line 29 to obtain male mice homozygous for the transgene. Two males were selected and mated with HSL-null female mice because male null mice are sterile. All pups were heterozygous both for the testis-specific transgene and for the invalidated gene. To obtain rescued male mice, the double heterozygous mice were either mated together or double heterozygous male mice were mated with female-null mice. We therefore obtained male mice with one or two alleles of the HSL tes transgene that were genotyped using quantitative PCR on tail genomic DNA. From experiments to build standard curves, 2.5 ng of DNA was chosen for the genotyping of animals using a Taqman chemistry-based human HSL tes -specific assay-on-demand and a GeneAmp 7000 sequence detection system (Applied Biosystems, Courtaboeuf, France). The murine hypoxanthine ribosyltransferase gene was amplified as the reference gene using SYBR green chemistry (Applied Biosystems) with the following primers: mHPRT-S (5Ј-TGGC-CATCTGCCTAGTAAAGC-3Ј) and mHPRT-AS (5Ј-GGACGCAGCAAC-TGACATTTC-3Ј).
Mating Trials-Four to 11 male mice from each group were individually housed with two sexually mature (8-week-old) virgin B6D2/F1 females. Vaginal plugs were checked daily, and the development of gestation was monitored. Pregnant females were sacrificed for caesarian section at 8 -19 days post-coitum. The size of the litter was recorded, and the offspring were rapidly examined for the life status and possible growth abnormalities.
Collection of Tissues and Sperm Counting-Mice of each group were weighed and sacrificed. Blood was recovered, and the testes and epididymes were dissected out and weighed. One epididymis and one testis were used for histological analysis. The other epididymis was rapidly frozen in liquid nitrogen and stored at Ϫ80°C until sperm heads (sperm reserves) were counted as follows. The organ was first cut with a scalpel into several fragments and homogenized in 50 ml of 0.15 M NaCl containing 0.005% Triton X-100 (Sigma). After homogenization using three rounds of sonication (12 kHz), an aliquot of the cell suspension was loaded on a Malassez hemocytometer. Sperm heads were counted in duplicate.
Western Blot Analysis of HSL-Testis and adipose tissue were homogenized in 4 volumes of homogenization buffer (0.25 M sucrose, 1 mM EDTA, pH 7.0, 1 mM dithioerythritol, 20 g/ml leupeptin, 20 g/ml antipain) and centrifuged at 110,000 ϫ g at 4°C for 45 min to prepare fat-free cytosolic supernatants. Samples of 50 g of proteins from testis and spleen and 25 g of proteins from adipose tissue were subjected to 7% SDS-PAGE, transferred onto nitrocellulose membrane (Hybond ECL, Amersham Biosciences), and probed with affinity-purified polyclonal antibodies directed against the human 88-kDa HSL or the testisspecific NH 2 -terminal region of human HSL. Immunoreactive proteins were determined by enhanced chemiluminescence reagent (Amersham Biosciences) and visualized by exposure to Fujifilm. Control experiments were performed with preimmune serum. No immunoreactive band was observed (data not shown).
Light Microscopy Analyses-Mice testes and epididymes were stored in aqueous Bouin's solution for 24 h. The fixed tissues were embedded in paraffin wax. Five-m sections were dried overnight at 37°C, deparaffined, and rehydrated through decreasing grades of alcohol. Sections were stained with 0.2% hematoxylin, dehydrated, and mounted in Eukitt (PolyLabo, Strasbourg, France) for microscopic observation.
Enzyme Activity Assays-Tissue samples were homogenized in 4 volumes of homogenization buffer (0.25 M sucrose, 1 mM EDTA, pH 7.0, 1 mM dithioerythritol, 20 g/ml leupeptin, 20 g/ml antipain) and centrifuged at 15,000 ϫ g at 4°C for 20 min. Protein concentrations were determined with the Bio-Rad protein assay using bovine serum albumin as standard. In vitro enzymatic activities were performed as described (12). Briefly, 1-(3)-mono-oleyl-2-O-mono-oleylglycerol and cholesterol oleate were emulsified with phospholipids by sonication. Fat-depleted bovine serum albumin (Sigma) was used as a fatty acid acceptor. The homogenates were incubated for 30 min at 37°C with the different substrates. Hydrolysis was stopped, and released radiolabeled oleic acid was measured using a Tri-Carb 2100TR scintillation counter (PerkinElmer Life Sciences). One unit of hydrolase activity is equivalent to 1 mol of fatty acid released per minute at 37°C.

Measurement of Neutral Lipid Molecular Species by Gas-Liquid
Chromatography-Ethyl acetate, chloroform, and methanol were purchased from Carlo Erba. Diacylglycerol-1,3-dimyristoyl, stigmasterol, cholesteryl heptadecanoate, and triheptadecanoylglycerol were obtained from Sigma. Half of the testis was weighted and crushed using a Potter homogenizer with the following solution: methanol/chloroform/ water containing 5 mM EGTA (1.1/0.5/0.5 ml). Lipids from an equivalent of 10 mg of tissue were extracted with 1 ml of methanol:chloroform: water (2:1:1). The suspension was centrifuged for 10 min at 1500 rpm to pellet the proteins. Supernatant was transferred into an 8-ml Tefloncapped vial containing four internal standards: 3 g of stigmasterol, 2 g of diacylglycerol-1,3-dimyristoyl:14, 2 g of cholesteryl heptadecanoate, and 3 g of triheptadecanoylglycerol. Double phase was generated by adding 2.25 ml of chloroform, 2 ml of methanol, and 1.75 ml of water according to Bligh and Dyer (13). The chloroform phase was filtered over glass wool, evaporated to dryness, and dissolved in 50 l of ethyl acetate. 4 l of lipid extract was analyzed by gas-liquid chromatography on a 4890 Hewlett Packard system using an Ultra1 Hewlett Packard fused silica capillary column (5 m ϫ 0.31 mm i.d.) coated with crosslinked dimethylsiloxane (14). Oven temperature was programmed from 200 to 340°C at a rate of 6°C per min, and the carrier gas was hydrogen (0.5 bar). The injector and the detector were at 315 and 345°C, respectively. Protein pellets were dissolved in 0.5 ml of 0.1 M NaOH overnight and quantified with the Bio-Rad assay.
Quantitative RT-PCR Analysis-Total RNA was isolated using RNA STAT-60 isolation reagent (Tel-Test, Friendswood, TX). Total RNA (1 g) was treated with DNase I (DNase I amplification grade, Invitrogen) and then retrotranscribed using random hexamers and Supercript II reverse transcriptase (Invitrogen). Real time quantitative PCR was performed on GeneAmp 7000 Sequence Detection System using SYBR Green chemistry (Applied Biosystems) as described (15). Human HSL tes mRNA was quantified using an assay-on-demand gene expression assay (Applied Biosystems). 18 S ribosomal RNA was used as control to normalize gene expression using the ribosomal RNA control Taqman assay kit (Applied Biosystems).
Statistical Evaluation-Values are expressed as mean Ϯ S.E. Group means were compared with the control mean using the Mann-Whitney test (SAS software, version 8.0, SAS Institute Inc., Cary, NC).

RESULTS
Creation of Transgenic HSL tes Mice-To produce transgenic mice with expression of human HSL tes in testis, we generated a 7.6-kb fragment containing 1.4 kb of the testis-specific promoter that has been shown in transgenic mice to drive expression of a reporter gene in spermatids (7) (Fig. 1). Only haploid germ cells express significant amounts of human and mouse HSL tes (3,4). To enhance transgene expression, two endogenous introns were added in the construct (16). Two transgenic founders, 29 and 32, were obtained and bred to establish trans-genic lines. Lines 29 and 32 had 5-10 and 2-5 transgene copies, respectively. Western blot analysis was performed on testis protein extracts to determine the expression of human HSL tes . An antibody directed against the testis-specific part of human HSL tes recognized proteins between 110 and 130 kDa in testes of transgenic animals ( Fig. 2A). No reactive band was detected in adipose tissue. Because of the low homology between human and mouse testis-specific sequences (2), the antibody did not recognize the endogenous murine HSL tes . An antibody directed against all forms of HSL detected human and murine testicular forms of HSL and the 82-kDa adipose tissue murine HSL (Fig. 2B).
Body and Testis Weights-The body weights of genetically modified mice with various genotypes were not different from that of the WT mice except that of the HSL Ϫ/Ϫ HSL tes ϩ/ϩ , which was slightly lower (Ϫ11%, p Ͻ 0.05, Table I). The testis weights of HSL Ϫ/Ϫ and HSL Ϫ/Ϫ HSL tes ϩ/Ϫ mice were much lower than that of WT mice (Ϫ40%, p Ͻ 0.001, and Ϫ42%, p Ͻ 0.001, respectively). The weight of this organ was partially restored in HSL Ϫ/Ϫ HSL tes ϩ/ϩ male mice (ϩ18% compared with HSL Ϫ/Ϫ , p Ͻ 0.05, and Ϫ29% compared with WT, p Ͻ 0.01).
Histological Analysis-Histological observations reveal that a low number of early and late spermatids and no spermatozoa were found in the seminiferous epithelium of the HSL Ϫ/Ϫ mice (Fig. 3, B versus A). Furthermore, in these mice, numerous large vacuoles were present in the seminiferous tubules. These vacuoles were absent in the HSL Ϫ/Ϫ HSL tes ϩ/Ϫ mice (Fig. 3C). In these animals, the spermatid pool had notably increased, and rare spermatozoa could be observed. In contrast, numerous spermatozoa were seen in HSL Ϫ/Ϫ HSL tes ϩ/ϩ mice, and spermatogenesis generally appeared qualitatively normal (Fig. 3D). In the different transgenic lines, there were no apparent abnormalities in the interstitium. In accordance with our data showing that spermatogenesis was deficient in HSL-null mice, no spermatozoa was seen in the epididymes, which instead were packed with cell debris and immature round germ cells generally identified as spermatids, a number of which were multinucleated (Fig. 3, insert B versus insert A). The epididy-mis lumen of the HSL Ϫ/Ϫ HSL tes ϩ/Ϫ mice also contained numerous immature germ cells in most instances with a single nucleus and a few spermatozoa (Fig. 3C, insert). By contrast, the epididymis of HSL Ϫ/Ϫ HSL tes ϩ/ϩ mice was filled with numerous spermatozoa. Only a few immature germ cells were present (Fig. 3D, insert).
Sperm Reserves-Sperm reserves in HSL-deficient mice had decreased by more than 99% when compared with WT mice (Fig. 4). Sperm reserves of the HSL Ϫ/Ϫ HSL tes ϩ/Ϫ animals were markedly increased when compared with those of the HSL Ϫ/Ϫ mice (20ϫ, p Ͻ 0.05) but represented only 2.5% of the WT mice values. In HSL Ϫ/Ϫ HSL tes ϩ/ϩ mice, sperm reserves were increased about 10-fold when compared with the reserves of the HSL Ϫ/Ϫ HSL tes ϩ/Ϫ animals to reach 26% of that of the WT. Fertility Studies-The libido (mating behavior) of HSL Ϫ/Ϫ , HSL Ϫ/Ϫ HSL tes ϩ/Ϫ , and HSL Ϫ/Ϫ HSL tes ϩ/ϩ male mice was normal as judged by the time necessary for recovery of all copulatory plugs (10,8, and 3 days, respectively) when compared with WT mice (15 days). The mating trials show that HSL Ϫ/Ϫ male mice were sterile (Figs. 4 and 5). In contrast, the fertile (12%) HSL Ϫ/Ϫ HSL tes ϩ/Ϫ male mice produced litters with a lower number of pups compared with WT male mice. The presence of the two HSL tes alleles totally restored fertility. The average litter size sired by these mice was normal.
Human HSL tes mRNA Levels, Enzymatic Activities, and Neutral Lipid Analysis-We wished to determine whether there was a relationship between the number of transgene alleles and HSL tes mRNA expression and enzymatic activities. Human HSL tes mRNA level was two times higher in the testes of HSL Ϫ/Ϫ HSL tes ϩ/ϩ mice compared with HSL Ϫ/Ϫ HSL tes ϩ/Ϫ mice (69 Ϯ 5, n ϭ 3 versus 28 Ϯ 3, n ϭ 4). HSL possesses both CEH and diacylglycerol lipase activities (1). We found that hydrolysis of cholesteryl esters was extremely low in HSL Ϫ/Ϫ testis, whereas the diacylglycerol lipase activity represented 15% of the WT levels (Fig. 6). The presence of HSL tes on one allele  restored CEH activity to 10% of WT values. This activity was doubled and represented 25% of WT values in homozygous HSL tes transgenic animals. Expression of the transgene provoked an increase in diacylglycerol lipase activity that was less important than the increase in CEH activity. The number of alleles did not significantly influence the capacity to hydrolyze diglycerides. Triglyceride and diglyceride contents were not modified in HSL-null mice (Table II). Diglyceride levels were lower in HSL Ϫ/Ϫ HSL tes ϩ/ϩ mice compared with the other genotypes. Cholesteryl ester levels in testis showed the following rank order between genotypes: WT Ͻ HSL Ϫ/Ϫ HSL tes ϩ/ϩ Ͻ HSL Ϫ/Ϫ HSL tes ϩ/Ϫ Ͻ HSL Ϫ/Ϫ ; there was an Ϸ10-fold difference between WT and HSL Ϫ/Ϫ mice. This rank order was opposite that of CEH activities. DISCUSSION We have shown that expression of HSL tes under the control of its own promoter leads to the rescue of infertility observed in HSL-deficient male mice. The data point to the unique role of HSL tes among HSL forms expressed in testis and its quantitative requirement for normal spermatogenesis.
Multiple forms of HSL have been described in testis. Several protein species with apparent molecular masses ranging from 26 to 130 kDa are expressed (17)(18)(19)(20). These forms may arise from a single mRNA of 3.9 kb in rodents and from two mRNA species of 3.9 and 3.3 kb in humans (2)(3)(4). The larger form encodes a predicted protein of 120 kDa, HSL tes , and possesses a unique NH 2 -terminal domain. The 3.3-kb form encodes a protein of similar molecular mass, 88 kDa, as adipocyte HSL. The 5Ј-regions of the 3.9-and 3.3-kb mRNAs derive from two testis-specific exons located in the 5Ј-region of the HSL gene. The other testicular forms may arise from post-transcriptional modifications. Immunohistochemistry experiments revealed that HSL tes is expressed in haploid germ cells, whereas the short forms are expressed in interstitial and tubular somatic cells as well as in premeiotic germ cells (3,4). Expression of HSL in Leydig cells suggests that the enzyme may play a role in androgen production. However, normal plasma levels of testosterone, follicle-stimulating hormone, and luteinizing hor- 3. Histology of the testis and epididymis from WT, HSL ؊/؊ , HSL ؊/؊ HSL tes ؉/؊ , and HSL ؊/؊ HSL tes ؉/؉ mice. Compared with WT mice (A), HSL Ϫ/Ϫ mice (B) presented an altered spermatogenesis with an increased vacuolization of the seminiferous epithelium (thick arrows) and no spermatozoa in the lumen (arrowheads). C, HSL Ϫ/Ϫ HSL tes ϩ/Ϫ animals presented only a few small vacuoles as well as a few late spermatids and rare spermatozoa (arrowheads). D, in HSL Ϫ/Ϫ HSL tes ϩ/ϩ mice, numerous spermatozoa were observed (arrowheads), and spermatogenesis appeared qualitatively normal. Compared with WT mice (insert in A), the lumen of the caput epididymis of HSL Ϫ/Ϫ (insert in B) and HSL Ϫ/Ϫ HSL tes ϩ/Ϫ (insert in C) mice contained no or very few spermatozoa (arrowheads) but were filled with multinucleated early spermatids (fine arrows) and cell debris. In contrast, the epididymis of HSL Ϫ/Ϫ HSL tes ϩ/ϩ mice (insert in D) were packed with numerous spermatozoa (arrowheads) and less early spermatids, among which many fewer multinucleated cells were present (fine arrows). mone have been reported in HSL-null mice (8). Consistent with a lack of influence on endocrine status, the libido of HSLdeficient male mice was normal, as shown by the production of a normal rate of copulatory plugs. Rescue of the infertility of HSL Ϫ/Ϫ male mice by expression of HSL tes shows that, among the various forms expressed in testis, it plays a predominant role in testicular physiology, as expression of HSL in cell types other than germ cells does not seem critical for a qualitatively normal spermatogenesis.
Studies in rodents and humans have shown that HSL tes mRNA is strongly expressed in early spermatids, whereas the protein accumulates in elongated spermatids (3,7). This lag between mRNA and protein appearance is characteristic of many genes expressed in haploid germ cells. To evaluate the role of HSL tes in these defects, we used the HSL tes promoter to get a spatial and temporal transcription of the transgene similar to the endogenous gene. The sterility of HSL-deficient male mice associated with an alteration of the spermatogenetic process, and more specifically of spermiogenesis (spermatid differentiation) observed here, is in agreement with previous data on HSL-null mice with different genetic backgrounds (8,9). HSL Ϫ/Ϫ mice show a reduction of testis weight, an alteration of the seminiferous epithelium integrity, and almost empty sperm reserves. The presence of numerous immature germ cells (essentially spermatids) in the epididymis of the HSL Ϫ/Ϫ mice is reflective of a massive exfoliation of the seminiferous epithelium. Rescue of spermiogenesis in HSL Ϫ/Ϫ HSL tes ϩ/ϩ mice demonstrates that the aberrant maturation of spermatids and lack of spermatozoa are the consequence of the absence of correct HSL tes transcription. Interestingly, the presence of two alleles was necessary for full restoration of fertility. This may be due to the fact that when two HSL tes alleles are present in the genome of spermatogonia, there is a doubling of normal haploid germ cells.
HSL has been shown in different tissues to be essential for CEH and diacylglycerol lipase activities (1,21,22). In testis, triglyceride hydrolase activity is only slightly reduced in HSL Ϫ/Ϫ mice (23), whereas diacylglycerol lipase activity is markedly decreased. However, there is a substantial residual activity that suggests the existence of other lipases that hydrolyze diglycerides in testis. No relationship was found in the various transgenic lines between diacylglycerol lipase activity and fertility, and there was no difference in triglyceride and diglyceride contents between WT and HSL Ϫ/Ϫ mice. However, a relationship was found between CEH activity, cholesteryl ester level, and fertility. HSL-deficient testes have very low levels of CEH activity and show accumulation of cholesteryl esters as reported previously (8). The doubling of HSL tes mRNA levels and CEH activity observed in HSL Ϫ/Ϫ HSL tes ϩ/ϩ mice compared with HSL Ϫ/Ϫ HSL tes ϩ/Ϫ mice is related to a doubling of cholesteryl ester content in the testis. A comparison of enzymatic activities with fertility data revealed that a threshold of esterase activity around 0.1 milliunit/mg protein is necessary to restore sperm reserves to levels sufficient to totally restore fertility.
The CEH activity mediated by HSL tes in haploid germ cells is therefore essential for spermiogenesis. HSL hydrolyzes various kinds of substrates including triglycerides, diglycerides, cholesteryl esters, and retinyl esters (1,24). The role for retinoic acid has been illustrated in the dysfunction of spermatogenesis observed in vitamin A-deficient mice and mice with a targeted disruption of the retinoic acid nuclear receptor, RAR␣ (25,26). However, the defects observed with impairment of retinoic acid production or signaling occur during the early stages of spermatogenesis. It is therefore likely that other proteins are involved besides HSL tes . The involvement of HSL tes in cholesterol metabolism may be crucial, as shown by CEH activity and testicular cholesteryl ester level data. During development of the guinea pig, there is a positive relationship between HSL expression in seminiferous tubules and the ratio of free to esterified cholesterol (19). Abnormal formation of membranous intercellular bridges between germ cells in HSL-null mice suggests a role for HSL in membrane stabilization and integrity (9). Cholesterol level determines membrane fluidity, and its distribution is highly organized in spermatid membranes (27,28), which probably explains the presence of a great number of multinucleated early spermatids in the epididymis of the HSL Ϫ/Ϫ mice. HSL tes may therefore control the subcellular deposition of cholesterol in specific membrane domains. The anchoring to membranes may be provided by the NH 2 -terminal domain, which is found only in HSL tes . This domain adds 301 amino acids to the protein expressed in adipose tissue (2). Its enrichment in proline may indicate an involvement in proteinprotein interactions but this hypothesis has yet to be tested.
Taken together, our results indicate that male sterility in HSL-deficient mice is due to the lack of expression in spermatids of the testicular form, HSL tes , and suggests that it is mediated by impairment of CEH activities. The dramatic effect of a lack of HSL tes leads to the question of whether mutations in the HSL gene may be the cause of male infertility in humans. Moreover, the synthesis of specific HSL tes  inhibitors may constitute a strategy for the development of a male contraceptive.