Arylsulfatase A deficiency causes seminolipid accumulation and a lysosomal storage disorder in Sertoli cells.

Sulfogalactosylglycerolipid (SGG) is the major sulfoglycolipid of male germ cells. During spermatogenesis, apoptosis occurs in >50% of total germ cells. Sertoli cells phagocytose these apoptotic germ cells and degrade their components using lysosomal enzymes. Here we demonstrated that SGG was a physiological substrate of Sertoli lysosomal arylsulfatase A (ARSA). SGG accumulated in Sertoli cells of Arsa(-/-) mice, and at 8 months of age, this buildup led to lysosomal swelling and other cellular abnormalities typical of a lysosomal storage disorder. This disorder likely compromised Sertoli cell functions, manifesting as impaired spermatogenesis and production of sperm with near-zero fertilizing ability in vitro. Fecundity of Arsa(-/-) males was thus reduced when they were older than 5 months. Sperm SGG is known for its roles in fertilization. Therefore, the minimal sperm fertilizing ability of 8-month-old Arsa(-/-) males may be explained by the 50% reduction of their sperm SGG levels, a result that was also observed in testicular germ cells. These unexpected decreases in SGG levels might be partly due to depletion of the backbone lipid palmitylpalmitoylglycerol that is generated from the SGG degradation pathway in Sertoli cells and normally recycled to new generations of primary spermatocytes for SGG synthesis.

birth to two litters. Afterwards, the female was replaced with another 8-week-old wild-type female. The replaced females were caged individually for an additional month to ensure that no live births were missed. All the females used were proven fertile by their ability to deliver pups following mating with wild-type males after this 1-month waiting period.

Sperm-ZP binding and in vitro fertilization
General protocols were as previously described ( 21 ). Motile sperm were collected from the swim-up fraction, and an equal number of wild-type and Arsa Ϫ / Ϫ sperm were added to medium droplets, each containing 20-30 ovulated eggs retrieved from superovulated wild-type females.

Histology and transmission electron microscopy
Testis and epididymis sections from wild-type and Arsa Ϫ / Ϫ mice were prepared for light and electron microscopy analysis, and profi le areas of their seminiferous tubules were quantifi ed and statistically analyzed as previously described ( 22 ).

Detection of apoptotic cells
The terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay was performed on testis sections of wild-type and Arsa Ϫ / Ϫ mice. Paraffi n sections were deparaffi nized and rehydrated serially in decreasing concentrations of ethanol and were permeabilized in 0.1% Triton X-100 in 0.1% sodium citrate [8 min, room temperature (RT)]. After two rinses with PBS, the sections were treated with 1% hydrogen peroxide in 70% ethanol for 10 min to quench endogenous peroxidase activity. Following two washes in PBS, TUNEL staining was carried out using the TUNEL Enzyme and TUNEL Label Mix (Roche Applied Science, Indianapolis, IN) according to the supplier's instructions, and the sections were viewed under an epifl uorescence microscope. The fl uorescent signals were subsequently converted into colorimetric signals by incubating the sections with anti-fl uorescein Fab-peroxidase (POD) complex (Roche) (30 min, 37°C), using a 3,3 ′ -diaminobenzidine (DAB) substrate kit (Vectorlabs, Burlington, ON, Canada), which gave brown staining to positive cells.

Quantifi cation of SGG from whole testis and sperm lipid extracts
Lipids were extracted from decapsulated testes and PBSwashed epididymal and vas deferens sperm, and then subjected to liquid chromatography ESI-MS/MS and MRM for SGG quantifi cation using deuterated SGG as an internal standard, as previously described ( 23 ).

Testis lipid extracts from wild-type and Arsa
Ϫ / Ϫ mice at both 5 and 8 months of age were subjected to high-performance thin layer chromatography (HPTLC) followed by orcinol staining, as previously described ( 24 ). Glycolipids were stained purple, whereas phospholipids turned brown. In Arsa Ϫ / Ϫ mice, an extra glycolipid band chromatographed between SGG and phosphatidylcholine was observed; the area of the HPTLC plate around this band, starting underneath the SGG band, was scraped for lipid extraction by the Bligh/Dyer method ( 24 ). Extracted lipids were subjected to ESI-MS/MS as previously described ( 23,25 ).

Imaging mass spectrometry
Imaging MS was performed on frozen sections of 8-month-old wild-type and Arsa Ϫ / Ϫ testes, which were thaw-mounted onto indium-tinoxide-coated glass slides (Bruker Daltonics, Leipzig, SGG is a sulfoglycolipid present selectively in mammalian male germ cells at high levels (10 mol% of total cell lipids). The main molecular species, accounting for more than 90% of total SGG in male germ cells, contains C16:0 alkyl and C16:0 acyl chains ( 8 ). SGG is synthesized in primary spermatocytes through sequential steps, including galactosylation of alkylacylglycerol (palmitylpalmitoylglycerol, PPG for the main SGG species) by ceramide galactosyltransferase (CGT) to form galactosylglycerolipid (GG), which is then sulfated by cerebroside sulfotransferase (CST) to form SGG ( 9,10 ). The utilization of these two enzymes in the biosynthesis of SGG is the same as that of SGC in the brain ( 8 ). Once synthesized, SGG is targeted to the plasma membrane where it remains throughout the developing steps of male germ cells and fi nally sperm ( 8 ). Signifi cantly, SGG is essential for the continuation of spermatogenesis. Mice genetically null for Cgt or Cst are infertile due to an arrest of spermatogenesis during pachytene spermatocyte development ( 11,12 ).
Although the biosynthetic pathway of SGG has been delineated, the degradation process of SGG in the testis has not been well defi ned. Degradation of SGG would likely occur on membrane remnants of apoptotic germ cells after phagocytosis by Sertoli cells. It is well known that during spermatogenesis, testicular germ cells (TGC), especially spermatogonia and spermatocytes, undergo apoptosis at a signifi cant rate (>50% of total TGCs) ( 13 ) and that Sertoli cells are responsible for phagocytosis and degradation of apoptotic germ cell components ( 14,15 ), including SGG. Like SGC and other glycolipids, elevated intracellular levels of SGG are likely cytotoxic ( 16 ). In fact, intracellular buildups of SGC and several other glycolipids are known to cause lysosomal storage disorders ( 17,18 ). Therefore, Sertoli cells would have to degrade SGG into a deglycosylated neutral nontoxic lipid, such as PPG. Based on the known degradation pathway of SGC ( 8 ), this would require ARSA to desulfate SGG to form GG, and then galactosylceramidase (GALC) to degalactosylate GG to form PPG ( 17 ). This SGG processing would allow Sertoli cells to continue their roles in supporting the ongoing development of TGCs ( 13,19 ) and, through a recycling process, make available the essential substrate(s) for SGG synthesis by the new generation of testicular germ cells. In this study, we describe the signifi cance of Sertoli cell lysosomal ARSA in these processes using Arsa Ϫ / Ϫ mice.

Details of Arsa
Ϫ / Ϫ mice with congenic C57BL/6J background were as described ( 20 ). Both wild-type and Arsa Ϫ / Ϫ mice were bred and housed in our animal facilities, following the protocols approved by the Ottawa Hospital Research Institute Animal Care Committee.

Mating study
Mating was done with four wild-type and eight Arsa Ϫ / Ϫ male mice over a 6-month period starting when the males were ‫ف‬ 2 months old. Each male was individually caged with an 8-week-old wild-type female (C57BL/6J) for a duration that the female gave 5% goat serum in PBS, the fi xed and permeabilized Sertoli cells were treated with O4 anti-SGG IgM antibody (5 µg/ml, 2 h, RT) followed by two washes in PBS and then incubation with Alexa-488 conjugated goat anti-mouse IgM (Invitrogen, 1:400 dilution, 1 h, RT). Fifteen minutes prior to the completion of secondary antibody incubation, 10 µg/ml of Hoechst 33342 was added to the cell culture to stain the nucleus. The cells were then viewed for SGG staining under an epifl uorescence microscope.

ARSA enzymatic activity assay
Homogenates of Sertoli cells and TGCs in 250 mM acetate buffer, pH 5, were used to assess the desulfation activity of p -nitrocatecholsulfate as previously described ( 28 ).

Statistical analyses
Student t -test and ANOVA were used to determine signifi cant differences between two data sets.

Decreased fecundity and spermatogenesis in older
Natural mating experiments revealed that Arsa Ϫ / Ϫ males from 2 to 5 months of age produced the same accumulated numbers of pups per mating pair as age-matched wild-type males. However, at older ages, these parameters were lower for Arsa Ϫ / Ϫ males, so that at 8 months of age (equivalent to 40 years in men), it was 65% of the wild-type value ( Fig. 1A ). Between 2 and 8 months of age, the average litter size that Arsa Ϫ / Ϫ males produced from 4.6 litters was 6.2 ± 2.4 pups, as compared with 7.9 ± 1.3 pups from 5.5 litters that were produced by wild-type males ( P < 0.05). The ability of epididymal and vas deferens sperm from 8-month-old Arsa Ϫ / Ϫ males to bind to the ZP and to fertilize eggs in vitro was also minimal compared with sperm from age-matched wild-type males ( Fig. 1B ). In contrast, sperm-ZP binding and sperm fertilizing ability were the same for Arsa Ϫ / Ϫ and wild-type males at 5 months of age ( Fig. 1B ). Of note was the high number of sperm with abnormal morphology retrieved from 8-month-old Arsa Ϫ / Ϫ mice. Even in the swim-up suspension, ‫ف‬ 30-40% of these motile sperm had abnormal morphology ( Fig. 1C ), and this may contribute partly to the decreased fertilizing ability. However, the decreased fecundity was not from the inability of Arsa Ϫ / Ϫ males to copulate, as the rate of vaginal plug formation was not noticeably different in females caged with wild-type males versus Arsa Ϫ / Ϫ males. Although mild ataxia and ocular problems were observed in some Arsa Ϫ / Ϫ male mice at 8 months of age, their body weights (supplemental Table I) and general appearance were similar to those of age-matched wild-type males, and there were no visible indications that their sexual activities were impaired in any way. Testis histology revealed that spermatogenesis in 8-month-old Arsa Ϫ / Ϫ males was compromised with high rates of apoptotic germ cells ( Fig. 2B ), disorganization of germ cell layers ( Fig. 2A ) and smaller profi le areas of seminiferous tubules ( Fig. 2C ) compared with age-matched wild-type tubules ( Fig. 2A-C ). As a result, testis weights and the numbers of sperm produced by 8-month-old Germany), using a MALDI-time-of-fl ight(TOF)/TOF-type instrument, Ultrafl ex 2 TOF/TOF (Bruker Daltonics), in the negative mode with 2,5-dihydroxybenzoic acid as a matrix, as previously described ( 26 ). The spatial resolution of the obtained images was 25 µm.

Flow cytometry of SGG of TGCs
A testicular germ cell mixture was prepared by sequential enzymatic digestion of seminiferous tubules with collagenase and trypsin. Decapsulated testes were fi rst digested with 0.5 mg/ml collagenase in Dulbecco's Modifi ed Eagle Medium (DMEM, Invitrogen, Burlington, ON, Canada) at 33°C for 15 min with constant shaking at 100 cycles per minute. At the end of the incubation, the dissociated tubules were allowed to sediment by unit gravity and then washed three times with DMEM. The tubules were further digested with 0.5 mg/ml trypsin in DMEM supplemented with 1 µg/ml of DNase I at 33°C for 15 min with constant shaking at 130 cycles per minute. During the collagenase and trypsin treatments, the tubules were continually fl ushed with 5% CO 2 . As a last step, soybean trypsin inhibitor (SBTI) was added to the suspension to a fi nal concentration of 0.5 mg/ml. The suspension was gently processed through a large bore plastic transfer pipet to dissociate small cell clumps and then fi ltered through a 70 µm nylon screen. The fi ltered cell suspension was washed three times with washing buffer (0.5% BSA (BSA), 0.25 mg/ml SBTI, 0.5 µg/ml DNase I in DMEM) by centrifugation (450 g , 5 min). The resulting cell suspension was mainly a mixed population of TGCs, although a certain number of Sertoli cells were unavoidably coisolated.
The isolated testicular germ cells were subjected to fl ow cytometry. They were fi rst stained with DRAQ5 to differentiate TGC types based on their DNA amounts and/or compactness. Approximately 2 × 10 6 of TGCs in 1 ml of DMEM were incubated (15 min, RT) with 20 µM DRAQ5 (Biostatus, Shepshed, Leicestershire, UK). After washing once in PBS by low-speed centrifugation, the cells were fi xed with 4% paraformaldehyde in PBS (15 min, RT). Following two washes with PBS supplemented with 0.3% BSA (PBS-BSA), the cells were blocked with 5% normal goat serum and then incubated (45 min, RT) with 5 µg/ml O4 mouse monoclonal anti-SGG IgM antibody (Neuromics, Edina, MN) or normal mouse IgM diluted in PBS-BSA. The cells were then washed in the same medium and incubated (30 min, RT) with goat anti-mouse IgM conjugated with Alexa 488 fl uorochrome (Invitrogen; 1:400 dilution). After washing and resuspending in PBS-BSA, the cells were analyzed on a Coulter EPICS XL fl ow cytometer (Beckman Coulter Ltd, Mississauga, ON, Canada) equipped with a 15 mW argon laser with excitation at 488 nm. Ten thousand cells from each sample were analyzed at a fl ow rate of 200-400 cells/second. Fluorescence emission of DRAQ5 (emission max 681 nm/697 nm) was quantifi ed in FL4 after passage through a 675 nm bandpass fi lter (660-700 nm), and SGG fl uorescence was refl ected by a 550 nm dichroic longpass fi lter and quantifi ed after transmission through a 525 ± 20 nm bandpass fi lter.

Immunofl uorescence of intracellular SGG in Sertoli cell cultures
Sertoli cells were isolated from denuded testes of 8-month-old Arsa Ϫ / Ϫ and wild-type mice by sequential enzymatic treatments as described ( 27 ) and allowed to attach to BD Falcon CultureSlide (BD Biosciences, Mississauga, ON, Canada). Some TGCs were still attached to Sertoli cells even after the culture was treated twice with a hypotonic buffer. Following fi ve days in culture, adherent Sertoli cells were washed in PBS, fi xed with 4% paraformaldehyde in PBS (15 min, RT), and treated with 0.1% Triton X-100 in PBS (10 min, RT). After washing twice in PBS and blocking with epididymis of 8-month-old Arsa Ϫ / Ϫ mice ( Fig. 2A , bottom row, far right panel). However, the epididymal epithelium appeared to be normal ( Fig. 2A ), and the epididymis and seminal vesicle weights were unchanged from those of wild-type mice (supplemental Table II), results suggesting that the observed abnormalities in spermatogenesis were likely not due to alterations of testosterone levels. In 5-month-old Arsa Ϫ / Ϫ males, spermatogenesis and other related parameters were normal ( Fig. 2A-E ).

Changes in SGG levels in Arsa
Ϫ / Ϫ m ales: SGG accumulation in Sertoli cells lead to a lysosomal storage disorder As ARSA has been shown to desulfate SGG in vitro ( 28 ), SGG levels should be increased in Arsa Ϫ / Ϫ mice. As expected, SGG levels, quantifi ed by ESI-MS/MS with multiple reaction monitoring (MRM), were higher in the testes of Arsa Ϫ / Ϫ mice at both 5 and 8 months of age compared with age-matched wild-type mice ( Fig. 3A ). However, levels of epididymal and vas deferens sperm SGG were unchanged in 5 month-old Arsa Ϫ / Ϫ males, and they even decreased to 50% of wild-type levels in 8-month-old Arsa Ϫ / Ϫ animals ( Fig. 3B ). SGG levels in Arsa Ϫ / Ϫ TGCs at both 5 and 8 months of age were also compared with those of wild-type TGCs by fl ow cytometry. Primary spermatocytes, round spermatids, and highly condensed elongated spermatids have differential amounts or compactness of DNA. Therefore, by prestaining the TGCs with DRAQ5, a quantitative DNA dye, these cell fractions could be differentiated from each other for analysis of SGG staining intensity (supplemental Fig. I). In 5-month-old mice, SGG levels in primary spermatocytes (SC), round spermatids (RS), and highly condensed elongated spermatids (HC) were similar in Arsa Ϫ / Ϫ and wild-type males. However, SGG levels in all three TGC types were decreased compared with the wildtype cells in 8-month-old males, and the decreases in primary spermatocytes and condensing elongated spermatids were statistically signifi cant ( P < 0.05) ( Fig. 3C ). Round spermatids, consisting of a number of development steps (steps 1-8 in mice), may contain various amounts of SGG. Although the decreases in SGG levels may have indeed occurred in various round spermatid steps in Arsa Ϫ / Ϫ males, the inability to separate the different spermatid steps from each other did not allow for an indication of a signifi cant difference of SGG levels from the wild-type values. Interestingly, C18:0/C16:0 SGG ( m/z 823) and C18:1/C16:0 SGG ( m/z 821), as well as C16:0 SGC (palmitoylsulfatide, m/z 778), were uniquely present in Arsa Ϫ / Ϫ testes in addition to three other SGG species that were normally present in wild-type testes [i.e., the main species, C16:0/C16:0 ( m/z 795), and two minor SGG species, C16:0/C14:0 ( m/z 767) and C17:0/C16:0 ( m/z 809)] ( Figs. 4 and 5 ). All these SGG species and C16:0 SGC utilize the same biosynthesis pathway, indicating that CGT and CST were still active in Arsa Ϫ / Ϫ males. In fact, Cgt and Cst were expressed at similar levels in TGCs of 8-month-old Arsa Ϫ / Ϫ and age-matched wild-type mice (supplemental Fig. II).
The increased levels of SGG in 8-month-old Arsa Ϫ / Ϫ seminiferous tubules were confi rmed by imaging mass Arsa Ϫ / Ϫ males were only 60% and 50%, respectively, of wild-type males of the same age ( Fig. 2D, E ). Immature round germ cells were also present in the lumen of the of SGG would be in Sertoli cells ( Fig. 3D ). Immunolocalization of SGG in Triton X-100-treated Sertoli cells isolated from 8-month-old Arsa Ϫ / Ϫ and wild-type mice verifi ed this ( Fig. 6 ). SGG staining was at a higher intensity and in larger clumps in Arsa Ϫ / Ϫ Sertoli cells than it was in wildtype cells ( Fig. 6 ). These Arsa Ϫ / Ϫ Sertoli cells also showed abnormal morphology, including dislocalization of the nucleus and smaller sizes with fewer cellular processes. Electron microscopy of testis sections revealed further details in the abnormality of Sertoli cells of 8-month-old Arsa Ϫ / Ϫ mice. In Sertoli cells of 8-month-old wild-type animals, lysosomes were small dense structures, which varied spectrometry ( Fig. 4 ). In the testis, ARSA is present in two cell types, TGCs and Sertoli cells ( 3 ). In TGCs, ARSA is synthesized in primary spermatocytes ( 3,29 ) and fi nally traffi cked to the acrosome of spermatids ( 3 ). ARSA in Sertoli cells is localized to lysosomes and late residual bodies ( 3 ). Notably, the activity of Sertoli cell ARSA was much higher than that of TGCs, as assayed using an artifi cial substrate, p -nitrocatecholsulfate (62.65 ± 8.30 mU/10 6 cells for Sertoli cells versus 2.67 ± 1.12 mU/10 6 cells for TGCs). Sertoli cell ARSA is likely important for desulfation of SGG in membrane remnants of apoptotic cells and residual bodies that Sertoli cells phagocytose. By deduction, accumulation However, these parameters were the same in wild-type and Arsa Ϫ / Ϫ mice at 5 months of age. Data in A, B, and C are representative of three replicate experiments, whereas data in D and E are mean ± SD of values from fi ve wild-type and fi ve Arsa Ϫ / Ϫ mice. *Signifi cant difference between Arsa Ϫ / Ϫ and Arsa +/+ samples from 8-month-old animals; P < 0.005 for B, and P < 0.05 for C, D, and E. tubal lumen, and the heads of elongated spermatids are positioned vertically to the basement membrane ( 19 ). In the tubules of 8-month-old Arsa Ϫ / Ϫ mice, several elongated spermatid heads were disoriented, lying horizontal to the basement membrane ( Fig. 7B , upper panel). The phagocytic function of Sertoli cells for "cleaning up" apoptotic germ cells was also impaired in 8-month-old Arsa Ϫ / Ϫ mice. Apoptotic germ cells were frequently present in Arsa Ϫ / Ϫ tubules ( Fig. 7B , lower panel), whereas they were not abundant in wild-type testes.

DISCUSSION
The ability of ARSA to desulfate SGC and SGG in the presence of saposin B has been well characterized in vitro ( 1 ), and for SGC, it has been confi rmed in vivo ( 2,30 ). In this study, we revealed that SGG accumulation occurred in Sertoli cells of Arsa Ϫ / Ϫ mice, indicating that SGG was also a physiological substrate of enzymatically active ARSA in Sertoli cell lysosomes. Presumably, saposin B, which is essential for delivering "solubilized" SGG into the active site pocket of ARSA, is also produced by Sertoli cells, as prosaposin is de novo synthesized in these cells ( 31 ) and can be cleaved to saposin B by cathepsin D ( 32 ), which is also present in Sertoli cell lysosomes ( 33 ). As the major glycolipid of male germ cells, the degradation pathway for SGG in Sertoli cells may be essential for the process of spermatogenesis, as apoptosis occurs at >50% of total germ cell population in normal mice ( 13 ). These apoptotic germ cells are usually phagocytosed by Sertoli cells ( 34 ), and their subcellular components, especially those that are cytotoxic, need to be processed after phagocytosis. Once phagocytosed within the Sertoli cell, apoptotic germ cell fragments presumably fuse with the Sertoli cell lysosomes for the degradation reactions ( 14 ). Such events in normal mice and rats must be rapid and effi cient, as examples of this phenomenon are rarely seen ( 34,35 ). A similar process also occurs with residual bodies ( 36 ). However, the latter are numerous, rendering evidence of phagocytosis and the involvement of Sertoli lysosomes in this degradation process ( 36 ).
Phagocytic activity of Sertoli cells is likely enhanced by preferential internalization of acidic lipids (e.g., phosphatidylserine ( 37 )) present on the surface of apoptotic germ cells. Negatively charged SGG is also preferentially internalized into Sertoli cells, even at a higher capacity than phosphatidylserine (supplemental Fig. III). This "eat me" signal of SGG may refl ect the need for SGG degradation in Sertoli cells. Like its analog SGC, SGG may be cytotoxic when present intracellularly at aberrantly high levels ( 16 ). Because Sertoli cells and male germ cells contain lipases (38)(39)(40), lyso-SGG may be produced following the intracellular accumulation of SGG in Arsa Ϫ / Ϫ testes. Lysoglycolipids are believed to be more cytotoxic than the parental lipids ( 18 ). Psychosine (galactosylsphingosine) is one such lyso-glycolipid, which is accumulated in the brain of galactosylceramidase-depleted Krabbe patients and twitcher mice, resulting in severe dys/demyelination ( 18,41,42 ). HPTLC of lipid extracts from both testis and in number according to the stages of the spermatogenic cycle ( Fig. 7A , panel a), whereas in Arsa Ϫ / Ϫ mice, some Sertoli cells had numerous lysosomes with irregular shapes and a heterogeneous content at all stages of the cycle ( Fig. 7A , panel b). In other seminiferous tubules, lipid droplets were also apparent in Arsa Ϫ / Ϫ Sertoli cells and were especially abundant at Stages I-VII of the cycle ( Fig. 7A ,  panel c). Such droplets were not found in wild-type Sertoli cells. In some tubules, lysosomes of Sertoli cells of Arsa Ϫ / Ϫ mice were very large, more or less spherical, and numerous, contrasting the smaller uniform sizes of Sertoli lysosomes of wild-type mice ( Fig. 7A , panel d). These abnormal appearances in Arsa Ϫ / Ϫ Sertoli cells were typical of a lysosomal storage disorder ( 17,18 ), which likely contributed to Sertoli cell dysfunction. In normal testes, Sertoli cells provide spatial support to developing TGCs: they cytoorganize the more advanced TGCs to localize closer to the  In sperm, lyso-SGG was undetected in both wild-type and Arsa Ϫ / Ϫ mice. All these results argued strongly that the impairment of spermatogenesis and sperm fertilizing ability was a consequence of the changes of SGG levels, and not its lyso form, in Sertoli cells and male germ cells.
sperm homogenates failed to reveal the lyso-SGG band. Furthermore, ESI-MS analyses revealed that lyso-SGG ( m/z 556.6) was present in the testis homogenate of Arsa Ϫ / Ϫ mice at very low levels (from undetected to 4% of SGG), and consistently, the wild-type testis homogenates contained undetectable amounts of lyso-SGG (supplemental  delicate balance. With a drastic increase in apoptotic cell numbers in 8-month-old Arsa Ϫ / Ϫ mice, their Sertoli cells were likely unable to maintain their normal functions in preserving the integrity of the seminiferous tubules. Nonetheless, all these abnormalities associated with spermatogenesis were unlikely to result from altered levels of circulating and luminal androgens, as the morphology of the epididymal epithelium, a target tissue of androgens, as well as epididymal and seminal vesicle weights, were normal in 8-month-old Arsa Ϫ / Ϫ mice (supplemental Table II).
The lower levels of SGG in TGCs and in sperm of 8-month-old Arsa Ϫ / Ϫ mice may be a factor related to reduced spermatogenesis and marked decrease in sperm fertilizing ability, respectively. SGG on the testicular germ cell surface is essential for spermatogenesis ( 11,12 ). For sperm surface SGG, our accumulated fi ndings indicate that it has direct affi nity for the ZP and is involved in the binding between acrosome intact sperm and the ZP ( 7,8 ). In addition, SGG on the acrosome-reacted sperm head plays roles in sperm-egg plasma membrane interaction ( 44 ). SGG is also an integral component of sperm head In Arsa Ϫ / Ϫ mice, SGG accumulation over time in Sertoli cells may be a direct and/or indirect cause for lysosomal swelling observed in these cells, a common feature of a lysosomal storage disorder. This disorder may lead to functional defects, including the spatial and nutritional support that Sertoli cells provide to developing germ cells during spermatogenesis ( 13,19 ). In 8-month-old Arsa Ϫ / Ϫ mice, the orientation of spermatid heads was often abnormal; they were positioned horizontally instead of vertically to the basement membrane. Abnormal morphology of the sperm heads was also observed at high rates in these older Arsa Ϫ / Ϫ mice. These abnormalities may be associated with defective formation of the apical ectoplasmic specialization in Sertoli cells, a device for anchoring and shaping of the elongated spermatid heads ( 13 ). Furthermore, the phagocytic function of Sertoli cells may be compromised in 8-month-old Arsa Ϫ / Ϫ mice due to increased numbers of apoptotic germ cells, as well as the inability of Sertoli cells to effectively deal with this increased load. It is known that Sertoli cells can only support a fi nite number of germ cells ( 13,43 ), and an increased workload could disturb this Ϫ / Ϫ males, relative to wild-type males, at 8 months of age. Sertoli cells (S) were isolated from 8-month-old wild-type and Arsa Ϫ / Ϫ mice. Some germ cells (G) were coisolated. Cells were aldehyde-fi xed and treated with 0.1% Triton X-100 followed by indirect immunofl uorescence for SGG detection. Note the intracellular staining of SGG in Sertoli cells of both wildtype and Arsa Ϫ / Ϫ mice; however, the staining was more intense and appeared as clumped patterns in the knockout mice. SGG on the surface of germ cells was absent due to the Triton X-100 extraction. SGG levels in germ cells could arise from decreased activities of CGT and CST in the knockout mice to synthesize SGG, although both enzymes were still expressed at a similar transcriptional level in 8-month-old Arsa Ϫ / Ϫ and wildtype testes (supplemental Fig. II). Alternatively, the reduced SGG levels in TGCs of 8-month-old Arsa Ϫ / Ϫ mice may be due to limited supplies of the initial substrate, palmitylpalmitoylglycerol (PPG), required for SGG biosynthesis. In normal testes, PPG may be de novo synthesized within TGCs as well as transported from Sertoli cells following SGG degradation by ARSA to GG, which in turn is processed to PPG by GALC, which is also present in Sertoli cells (supplemental Fig. V). In Arsa Ϫ / Ϫ mice, this PPG recycling pool will not be available, although at a younger age, TGCs may compensate for this missing pool of PPG by increasing de novo synthesis. This compensation may not be possible in 8-month-old Arsa Ϫ / Ϫ mice, leading to decreased SGG levels in germ cells.
Our results on Arsa Ϫ / Ϫ mice have revealed the significance of SGG homeostasis for male fertility, and a model explaining the link between the SGG biosynthesis and degradation processes is shown ( Fig. 8 ). While the biosynthesis lipid rafts, which contain a number of ZP binding proteins, including ARSA, and are thus platforms for ZP binding ( 5,6 ). Our results revealed normal in vitro fertilizing ability and SGG levels in sperm from 5-month-old Arsa Ϫ / Ϫ mice, but near-zero fertilizing ability and a 50% decrease in sperm SGG levels in 8-month-old Arsa Ϫ / Ϫ animals. This indicated that sperm ARSA was dispensable for sperm fertilizing ability, whereas normal levels of sperm SGG was essential for in vitro fertilization. However, 8-month-old Arsa Ϫ / Ϫ males still sired offspring, although at a reduced rate compared with age-matched wild-type males. This result argued that the Arsa Ϫ / Ϫ sperm may be partly rescued in the female reproductive tract, as in the case of sperm genetically null for a surface serine protease, Prss21 ( 45 ). Nonetheless, the reduced rate of spermatogenesis and fecundity together with altered levels of SGG in the testis and sperm in 8-month-old Arsa Ϫ / Ϫ mice strongly suggests that homeostasis of SGG is important for male fertility, especially at advancing ages.
The decreases in SGG levels in TGCs and sperm in 8-month-old Arsa Ϫ / Ϫ mice were unexpected, considering that ARSA is a SGG degradation enzyme. The reduced pathway has been well documented ( 8 ), this study describes for the fi rst time the initial part of the SGG degradation pathway regulated by Sertoli cell lysosomal ARSA. The degradation of SGG from apoptotic germ cells may serve dual purposes: to prevent SGG-induced cytotoxicity and to generate a neutral lipid intermediate, PPG, for SGG synthesis in new generations of primary spermatocytes.