Quantification of seminolipid by LC-ESI-MS/MS-multiple reaction monitoring: compensatory levels in Cgt(+/⁻) mice.

Seminolipid, also known as sulfogalactosylglycerolipid (SGG), plays important roles in male reproduction. Therefore, an accurate and sensitive method for SGG quantification in testes and sperm is needed. Here we compare SGG quantitation by the traditional colorimetric Azure A assay with LC-ESI-MS/MS using multiple reaction monitoring (MRM). Inclusion of deuterated SGG as the internal standard endowed accuracy to the MRM method. The results showed reasonable agreement between the two procedures for purified samples, but for crude lipid extracts, the colorimetric assay significantly overestimated the SGG content. Using ESI-MS/MS MRM, C16:0-alkyl/C16:0-acyl SGG of Cgt+/− mice was quantified to be 406.06 ± 23.63 μg/g testis and 0.13 ± 0.02 μg/million sperm, corresponding to 78% and 87% of the wild-type values, respectively. CGT (ceramide galactosyltransferase) is a critical enzyme in the SGG biosynthesis pathway. Cgt−/− males depleted of SGG are infertile due to spermatogenesis arrest. However, Cgt+/− males sire offspring. The higher than 50% expression level of SGG in Cgt+/− animals, compared with the wild-type expression, might be partly due to compensatory translation of the active CGT enzyme. The results also indicated that 78% of SGG levels in Cgt+/− mice were sufficient for normal spermatogenesis.

with specifi c measurement of the major C16:0/C16:0 isoform using 2 H 3 -C16:0-alkyl/C16:0-acyl SGG, synthesized as previously described ( 24 ) as the internal standard. Further, the MRM method was used to quantify C16:0/C16:0 SGG in testes and sperm of fertile Cgt heterozygous male mice. The results allowed a correlation between the testicular and sperm SGG levels and the fertility status of Cgt +/ Ϫ mice. This information is relevant to future studies, including a determination of the minimal level of SGG required to maintain male fertility.

Materials
Silica gel HPTLC plates (60 Å, 200 µm thickness, 10 × 10 cm) were purchased from Whatman (Kent, UK). Acetonitrile (Optima® grade) was from Fisher Scientifi c (Pittsburgh, PA). Chloroform and methanol (HPLC grade) were from EMD chemicals, purchased through VWR (Mississauga, ON, Canada). Phosphatidy lethanolamine (PE), phosphatidylcholine (PC), and sphingomyelin (SM) were from Doosan Serdary Research Laboratories (Eaglewood Cliffs, NJ). Cholesterol (Chol) and SGC were from Sigma-Aldrich (St. Louis, MO). SGG, 2 H 3 -SGG, and galactosylglycerolipid (GG) were prepared in-house as described below. Azure A and Coomassie Brilliant Blue G250 were purchased from Sigma-Aldrich (Oakville, ON, Canada). Ultrapure water (16)(17)(18) Ϫ 1 ) was produced by an ELGA water purifi cation system (Wycombe, UK). All other reagents and solvents were of analytical grade or better. Rabbit polyclonal IgG antibody directed against a CGT peptide (RNGKYKGNGRVKHEKKVR: residues 524-541) was generated by AnaSpec Inc. (San Jose, CA). used ( 13,14 ). Azure A is a cationic dye that reacts with anionic lipids, including sulfolipids ( 15,16 ). Azure A is also useful for staining sulfolipids on thin layer chromatography (TLC) plates. The limit of detection (LOD) of the colorimetric assay for sulfolipids is around 10 µg in the original method and, despite modifi cations and improvements, is still around 0.3 µg ( 13,17 ). The LOD of Azure A staining of chromatographed lipids on high-performance (HP)TLC plates is even higher: at least 1 µg of a sulfolipid is required for detection. While the Azure A assay is convenient and simple, it has drawbacks, including lack of sensitivity and specifi city, as it reacts with all sulfolipids and certain anionic lipids, such as cardiolipin ( 15,16 ). HPTLC can be used to separate sulfolipids from other lipid classes; therefore, HPTLC coupled with Azure A staining remains a useful procedure for obtaining a global picture of sulfolipids in a lipid extract. More precise quantifi cation of SGG in a lipid extract can be made by scraping the SGG band from the HPTLC plate. SGG extracted from the silica powder can then react with Azure A in solution. However, SGG molecular species that differ by subtle changes in the acyl/alkyl chains are usually not resolved from one another by HPTLC. Therefore, individual minor SGG molecular species in a sample would be overlooked by the HPTLC-Azure A-based method; further, the amount of the major C16:0/C16:0 isoform would be overestimated by Azure A reaction, if the collective amount of the minor SGG molecular species is signifi cant. A method precisely targeted to specifi c SGG molecular species is required for accurate quantifi cation.
Liquid chromatography (LC)-tandem mass spectrometry (MS/MS) with multiple reaction monitoring would be an ideal quantifi cation method for SGG molecular species. It has been used for quantifi cation of sulfogalactosylceramide (SGC) (also known as sulfatide and SM4s), a sulfolipid analog of SGG ( 18 ). Both SGG and SGC carry the same galactosyl sulfate head group, but unlike SGG with a glycerol backbone, SGC has a sphingosine skeleton ( Fig. 1C ). SGC is the better studied sulfolipid, as its accumulation in the brain and other tissues due to a genetically inherited defi ciency of arylsulfatase A (ASA) or saposin B (SGC carrier protein) results in metachromatic leukodystrophy, a fatal neurological disease ( 19,20 ). With a suitable internal standard, LC-MS/MS-MRM quantifi cation can be supremely accurate and sensitive, as it can quantify individual sulfolipid isoforms that differ by relatively subtle structural changes in the hydrocarbon chains ( 18,(21)(22)(23). The best internal standards are stable isotopelabeled analogs ( 2 H-, 13 C-, and 15 N-labeled forms) because their physical and chemical properties, including extraction effi ciency, chromatographic behavior, and ionization effi ciency, are virtually identical to those of the natural, nonlabeled analyte. Although LC-MS/MS-MRM analysis has recently been used to determine the amounts of seminolipid in saposin A-and prosaposin-defi cient mice ( 23 ), the analyses were performed without an internal standard and relied on absolute signal intensity uncorrected for sample losses during extraction. In this study, we compared SGG quantifi cation on the same samples using the colorimetric assay and the LC-MS/MS-MRM procedure assay and ESI-MS/MS-MRM analyses, as well as analytical HPTLC, was performed on CD-1 testis lipids (with and without further purifi cation for SGG), and Cgt +/ Ϫ and Cgt +/+ total testis and sperm lipids.

Preparation of SGG and GG
SGG from pig testis. Partially purifi ed SGG starting material was previously prepared in bulk amount from pig testis total lipid extracts via Biosil-A column chromatography followed by preparative TLC ( 29,30 ). It contained a prominent band with an R f at 0.33 and other components when examined by HPTLC with Coomassie Blue staining ( Fig. 2 , top panel, lane a). This material, referred to as "partially purifi ed SGG," was stored as powder at Ϫ 20°C. Then 30 µg was dissolved in chloroform/ methanol (1/1, v/v) and divided into two equal fractions, one of which was further purifi ed by preparative HPTLC ( 29 ). HPTLC plates were predeveloped in chloroform/methanol (1/1, v/v), air dried, and heat activated at 100°C for 1 h prior to use. Lipids were then loaded as a broad band (8 cm), with SGG standard (highly purifi ed from pig testis in our lab and characterized by ESI-MS/MS ( 24 )) loaded as a separate narrow band (0.8 cm) on the side. The plates were developed with chloroform/methanol/water (65/25/4, v/v/v), air dried, and placed in a closed jar of iodine vapor, which transiently stained the lipids yellow. The band corresponding to SGG (based on the R f value, and the same position as the cochromatographed SGG standard) was scraped and extracted three times with chloroform/methanol (1/1, v/v). The pooled extracts were dried in a stream of nitrogen and then subjected to Bligh-Dyer partitioning. This fi nal product, referred to as "highly purifi ed SGG," appeared as a single band with an R f of 0.33, as visualized by analytical HPTLC/Coomassie Blue staining (see methodology below) ( Fig. 2 , top panel, lane b). Both the partially purifi ed and highly purifi ed SGG samples revealed similarity of the m/z 97 parent ion ESI tandem mass spectra ( Fig. 2 , bottom panel), and both possessed the expected major signal at m/z 795.4 corresponding to the C16:0/ C16:0 SGG anion (calculated 795.52982 Da for C 41 H 79 O 12 S). However, both samples still contained other copurifi ed sulfated components ( Fig. 2 , bottom panel), highlighting the diffi culty of preparing a molecularly homogeneous standard from natural sources. Therefore, the highly purifi ed SGG sample was still inadequate for use as a quantitative standard for MRM experiments, although it could still be used as a standard in analytical HPTLC. Both the partially purifi ed (the unprocessed 15 µg fraction) and the highly purifi ed SGG samples were used for the initial comparison of the two SGG quantification methods-the Azure A colorimetric assay and LC-MS/ MS MRM analysis ( Table 1 ).
SGG from CD-1 mouse testes. Half of the total lipids extracted from both testes of the CD-1 male mouse was used to prepare partially purifi ed mouse testis SGG following the same preparative HPTLC method as described for the pig testis SGG. Total mouse testis lipids and partially purifi ed SGG samples were subjected to both quantifi cation methods ( Table 1 ), as well as analytical HPTLC and parent (sulfate) ion ESI tandem mass spectrometry.
Galactosylalkylacylglycerol. GG was prepared from partially purifi ed pig testis SGG via mild acid hydrolysis ( 31 ). Following HPTLC with Coomassie Blue staining, the product showed a single band with an R f of 0.659. During positive ion ESI in the presence of lithium, the fi nal product gave an intense ion at m/z 723.3 ( 24 ) corresponding to the lithiated adduct (calculated as 723.5953 for C 41 H 80 O 9 Li). This sample was quantifi ed by weighing and used as the HPTLC reference standard.

Animals
CD-1 male and female mice (4-to 6-weeks-old) were purchased from Charles River Canada (St-Constant, QC, Canada). One wildtype male and two Cgt +/ Ϫ female mice (both having C57BL/6J genetic background), previously generated by Dr. Brian Popko (formerly at the University of North Carolina at Chapel Hill, Chapel Hill, North Carolina) ( 25 ), were used to start the colony that was maintained by continuing breeding of heterozygous males with heterozygous females. DNA isolated from tail tissues was used for genotyping by PCR following the previously described protocol ( 26 ) with two sets of primers for DNA amplifi cation: Cgt (forward), 5 ′ -CTC TCA GAA GGC AGA GAC ATT GCC-3 ′ and Cgt (reverse), 5 ′ -CAT CCA TAG GCT GGA CCC ATG AAC-3 ′ (this set gave a wild-type gene product of ‫ف‬ 500 bp); Neo (forward), 5 ′ -GGA GAG GCA ATT CGG CTA TGA C -3 ′ and Neo (reverse), 5 ′ -CGC ATT GCA TCA GCC ATG ATG G -3 ′ (this set gave a ‫ف‬ 300 bp product of the inserted neomycin gene in the exon 2 of the Cgt sequence in Cgt Ϫ / Ϫ mice) ( 25 ). All mice were kept in a temperature-controlled room (22°C) with a 12:12 dark/light cycle. The boarding and handling of these mice, as well as all subsequent experimental procedures with them, were approved by Animal Care Committee, Ottawa Hospital Research Institute (OHRI).

Mouse testis and sperm preparation
Mice were sacrifi ced by cervical dislocation. Testes were removed, decapsulated, weighed ( ‫ف‬ 100-120 mg/testis) and individually hand-homogenized in 1 ml of phosphate buffered saline (PBS), which was then used for lipid extraction. Sperm were retrieved from the caudal epididymis and vas deferens from each animal into 1 ml of Hepes-buffered Krebs Ringer solution (KRB-Hepes; 119.4 mM NaCl, 4.8 mM KCl, 1.7 mM CaCl 2 , 1.2 mM KH 2 PO 4 , 1.2 mM Mg 2 SO 4 , 4 mM NaHCO 3 , 1 mM sodium pyruvate, 25 mM sodium lactate, 5.6 mM glucose, 1 U/ml of penicillin G, 1 µg/ml of streptomycin sulfate, 21 mM Hepes, and 28 µM phenol red at pH 7.4) ( 14 ). For lipid extraction, the sperm suspension was centrifuged (430 g , 10 min, 25°C), then pellet was washed twice with PBS and fi nally resuspended in 0.5 ml PBS. For the in vitro fertilization experiments, the sperm suspension was subjected to Percoll gradient centrifugation to select for the motile population as previously described ( 27 ). Sperm numbers were counted under a microscope in a 10 µl aliquot of the sperm suspension using a hemocytometer.

Lipid extraction from mouse testes and sperm
Lipids were extracted from testes of a CD-1 male, and three sets of Cgt +/ Ϫ males and their wild-type littermates, as well as from caudal epididymal and vas deferens sperm retrieved from the individual Cgt heterozygous and wild-type mice. Following a modifi ed Bligh-Dyer method ( 14,(28)(29)(30), methanol (2.5 ml) and chloroform (1.25 ml) was added to the 1 ml PBS testis homogenate. After thorough mixing and 1 h incubation at room temperature, the sample was diluted further with chloroform and water (1.25 ml each), then vigorously mixed and centrifuged (800 g , 5 min, room temperature). The lower phase was transferred to a clean glass tube, and the upper phase was reextracted twice with chloroform (2.5 ml each time), again using centrifugation to separate the phases. The lower phases containing extracted total lipids from two testes of the same animal were then pooled and dried in a stream of nitrogen for storage at Ϫ 20°C until further use. Lipids were also extracted from the sperm suspension from 1 animal (18-25 million sperm) using the same procedure as described for the testis.
Testis lipids from the CD-1 male were divided into two portions. Partially purifi ed SGG was prepared from one portion by preparative HPTLC (see below). SGG quantifi cation by Azure A

High-performance thin-layer chromatography
Conditions for analytical HPTLC plate preparation and development were the same as described for preparative HPTLC. Lipids dissolved in chloroform/methanol (20 µl, 1/1, v/v) were loaded as a 0.8 cm band onto the preactivated HPTLC plate. Samples loaded included (A) partially purifi ed pig testis SGG (3.75 µg); (B) highly purifi ed pig testis SGG [3.75 µg prepared from a parallel sample of (A)]; (C) total mouse testis lipids extracted from 5 mg of tissue; and (D) partially purifi ed mouse testis SGG prepared from a parallel sample of (C). After plate development, lipids were detected by various dyes. These included Coomassie Brilliant Blue G-250 (0.03% in 30% methanol/100 mM NaCl at room temperature), the excess of which was removed with 30% methanol in 100 mM NaCl ( 14,32 ); and Azure A (0.016% Azure in 1 mM H 2 SO 4 ), the excess of which was removed with 50 mM H 2 SO 4 /methanol (3/1, v/v) ( 7,14 ). Coomassie Blue stained all lipids blue, whereas Azure A stained sulfolipids blue ( 29 ). Lipid standards run on each plate included SGG, GG, PE, PC, Chol, and SM. Digital images were recorded after staining using an Epson scanner (Epson Canada, ON, Canada).

SGG quantifi cation
Azure A assay. The method described by Kean ( 15 ) and as modifi ed by Weerachatyanukul et al. ( 13 ) was used. Dried lipid extracts were redissolved in chloroform/methanol (1/1, v/v), and aliquots corresponding to 7 million sperm or 1 mg wet weight of testes were transferred to screw-capped glass tubes and dried under N 2 . CHCl 3 /MeOH (1.5 ml, 1/1, v/v), aqueous H 2 SO 4 (1.5 ml, 50 mM), and Azure A dye solution (0.12 mg in 0.3 ml of 2.5 mM H 2 SO 4 ) were added to each tube. After thorough mixing and centrifugation (300 g , 10 min, 4°C), the chloroform phase containing the blue Azure A-lipid complex was transferred to a new glass tube, and the absorbance at 635 nm was recorded (Ultraspec 2000 UV/Visible Spectrophotometer, Pharmacia Biotech, Piscataway, NJ). Sperm samples were assayed in duplicate, and testis samples were assayed in triplicate. Sulfated lipid concentrations were inferred from interpolation from a standard curve constructed with SGC (0-10 µg) as standard. The LOD of this assay was about 0.3 µg of SGC.

RNA purifi cation and analysis
Total RNA was extracted from Cgt +/ Ϫ and wild-type testes using an RNeasy kit (Qiagen, Mississauga, ON, Canada). The Cgt Ϫ / Ϫ samples were used as a negative control. The RNA was fi rst quantifi ed by A260 nm, and the same amount of RNA from each sample was used for reverse transcription using a Retroscript Kit (Ambion, Foster City, CA). The cDNA products were purifi ed from other components in the reactions using a QIAquick PCR Purifi cation Kit (Qiagen). To prepare Cgt cDNA for the quantitative real-time PCR, an aliquot of the cDNA products from reverse transcription of the wild-type sample was further subjected to PCR using the specifi c primer sets for Cgt, forward: 5 ′ -CTC TCA GAA GGC AGA GAC ATT GCC-3 ′ and reverse: 5 ′ -CAT CCA TAG GCT GGA CCC ATG AAC-3 ′ . The amplifi ed Cgt PCR product was purifi ed as described above and subjected to agarose gel electrophoresis and eithidium bromide staining ( 26 ). The stained Cgt PCR product was quantifi ed by densitometric analyses by comparison to the intensity of DNA markers using ImageQuant software (Molecular Dynamics, Sunnyvale, CA

Immunoblotting of CGT
Testes collected from Cgt +/ Ϫ and wild-type littermates were decapsulated and individually homogenized in 1 ml of homogenization buffer (250 mM sucrose containing 1 mM EDTA and 10 mM Hepes, pH 7.2). Post-nuclear supernatants (PNS) were obtained by centrifugation (375 g , 15 min) ( 35 ). Aliquots of PNS were quantifi ed for total proteins using the Bio-Rad Bradford assay with various concentrations of BSA to build a standard curve. recorded using instrument manufacturer-supplied software (MacSpec version 3.3). Calculations of the amount of C16:0/ C16:0 SGG in each sample were made from the amount of internal standard added to each sample adjusted by the relative intensity of the two sets of signals, assuming 100% purity of the internal standard synthesized as the sodium salt. The average of the data obtained from the two transitions was used in the calculations. All lipid samples were analyzed in triplicate. The LOD of the MRM analysis was about 1 pmole of SGG.

Sulfate ion scanning of testis lipids
Dried lipid samples were redissolved and injected into the same LC/MS/MS system as described above, with the mass spectrometer operating in the negative ion parent scan mode with the same source and MS/MS conditions. In these experiments, Q1 was tuned to unit resolution (10% valley between the PPG 13 C-isoforms), Q3 was tuned to lower resolution (no separation between the PPG 13 C isoforms), Q1 was scanned ( m/z 400-1500, 0.3 Da step size, 6 s/scan), and Q3 was set to transmit the m/z 97.1 ion corresponding to HSO 4 Ϫ .

Fertility assessment of Cgt +/ ؊ male mice
Mating experiments. Five pairs of Cgt +/ Ϫ males and their wildtype male littermates from fi ve different litters (two pairs of three-month-old and three pairs of nine-month-old males) were used. Each male was individually caged with an eight-week-old wild-type female in the evening. The females were checked for vaginal plugs the following morning. If positive, this was counted as day 1 of gestation. As females of this strain are infanticidal, the pregnant females were euthanized between day 14 and 18 of gestation, prior to the expected delivery date of the pups. An accurate census of litter sizes was achieved by counting the number of fetuses or implantation sites in the uteri.
In vitro fertilization. To compare fertilizing ability of Cgt +/ Ϫ sperm with that of the wild-type, in vitro fertilization (IVF) was carried out as previously described ( 33,34 ). Four pairs of Cgt +/ Ϫ males (two pairs of three-month-old and two pairs of nine-month-old males) and their wild-type male littermates were used. Mature eggs were collected from superovulated CD-1 females. Fertilization rate was defi ned as percentage of total eggs fertilized in each sample. A minimum of 80 eggs were used for the assessment of sperm fertilizing ability for each genotype. SGG, sulfogalactosylglycerolipid. a Thirty micrograms of partially purifi ed pig testis SGG was weighed and divided into two equal fractions, one of which was used to prepare the highly purifi ed sample, as described in "Materials and Methods" (see Fig. 2 for HPTLC patterns and sulfate ion ESI-MS/MS spectra). All assays were done in triplicate. The SGG amounts in the original fraction (15 µg by weight) were reported as mean ± SD of the triplicate results. b SGG quantifi cation by both methods was done in triplicate on total mouse testis lipids and SGG partially purifi ed by preparative HPTLC from the total testis lipids, as described in "Materials and Methods" (see Fig. 3 for HPTLC patterns and sulfate ion ESI-MS/MS spectra). The amounts of lipids used for each SGG quantifi cation method were in the same ranges as those described for pig testis SGG. Data were expressed as mean ± SD of SGG per g of testis wet weight from the triplicate results. small overestimation by the colorimetric procedure (1.2and 1.4-fold for highly purifi ed and partially purifi ed pig testis samples, respectively) was observed ( Table 1 ). This was likely the result of the presence of minor SGG isoforms and possibly trace levels of other Azure A-positive components in the samples. ESI-MS/MS parent ion (sulfate) scanning of partially and highly purifi ed pig testis samples gave similar spectra, with C16:0/C16:0 SGG being the main molecular species ( m/z 795) detected ( Fig. 2 , bottom  panel). The presence of a single Azure A-stained SGG band on the HPTLC plate corroborated these MS results ( Fig. 2 , top panel, lanes c and d). Some of the other small MS signals in the m/z 750-850 range were likely to be minor SGG isoforms: C17:0/C18:0 for m/z 837; C16:0/C18:0 for m/z 823; C17:0/C16:0 for m/z 809; C15:0/C16:0 for m/z 781, and C14:0/C16:0 or C16:0/C14:0 for m/z 767 ( 3,4 ). There were a number of putative sulfate-containing ions in the m/z range 450-600 in the pig testis SGG samples that were reduced in abundance but not completely removed by the additional purifi cation to produce the highly purifi ed sample ( Fig. 2 , bottom panel, A versus B). While the m/z 557 ion matched the lyso form of C16:0/C16:0 SGG, the identities of other signals in this m/z range were unknown. The decrease of these low-range molecular mass sulfolipids in the highly purifi ed pig testis SGG sample would partly account for the lower sulfolipid amounts obtained from the Azure A assay compared with the partially purifi ed SGG sample ( Table 1 , 24% decrease). However, this decrease could also come from the removal of other Azure A-reactive anionic lipids during the purifi cation process. A signifi cant reduction of Coomassie Blue-stained lipid bands on the HPTLC plate supported this statement ( Fig. 2 , top panel, lane a versus b). Nonetheless, this reduction was partly due to the lipid loss during the purifi cation. The amount of C16:0/C16:0 SGG determined by LC-MS/MS-MRM was decreased by 13% (9.6 versus 8.4 µg, Table 1 ) before and after preparative HPTLC purifi cation, indicating the lipid loss during purifi cation.
In the mouse testis lipid samples, the sulfolipid content, as quantifi ed by the Azure A procedure, was much higher than the amount of C16:0/C16:0 SGG obtained from LC-MS/MS-MRM (i.e., 7.3-and 4.6-fold difference for the total lipid extract and partially purifi ed samples, respectively) ( Table 1 ). HPTLC Coomassie Blue results revealed, as expected, a complex profi le for the total lipid extract and a less complex pattern for the partially purifi ed sample ( Fig. 3 , top panel, lanes  a and b). While the HPTLC Azure A profi le for both samples was dominated by the SGG band, additional but faint bands of low R f values were evident, especially in the total lipid extract ( Fig. 3 , top panel, lanes c and d). ESI-MS/MS parent ion (sulfate) scanning of total mouse testis lipids gave a complex spectrum dominated by C16:0/C16:0 SGG ( Fig. 3 , bottom  panel, A). Therefore, the presence of substantial amounts of other sulfolipids in the total mouse testis lipid extract was partly the cause of a great difference (7.3-fold) in the levels of total sulfolipids over C16:0/C16:0 SGG. However, ESI-MS/ MS parent ion (sulfate) scanning spectrum was relatively clean for the partially purifi ed mouse testis lipid, with the C16:0/C16:0 SGG signal at m/z 795 accounting for more Equivalent amounts of protein (20 µg) from each PNS sample were subjected to SDS-PAGE (10% polyacrylamide, 1.0 mm thick) ( 36 ), followed by electroblotting onto a nitrocellulose membrane ( 37 ). The nitrocellulose was blocked for 1 h with 5% fat-free milk in Tris-buffered saline (137 mM NaCl in 20 mM Tris-HCl, pH 7.6) containing 0.05% Tween 20. Immunoblotting was performed using anti-CGT antiserum (1:5,000 dilution) as the primary antibody, and goat anti-rabbit immunoglobulin conjugated with horseradish peroxidase (1:5,000 dilution, Bio-Rad Laboratories, Hercules, CA) as the secondary antibody. The primary and secondary antibodies were prepared in the blocking medium. The reactivity of antibody-antigen was detected by enhanced chemiluminescence (ECL) using an ECL kit (Pierce, Rockford, IL). Relative CGT expression levels in the Cgt +/ Ϫ samples compared with those in the wild-type samples were determined by densitometric analysis using AlphaImager Software (AlphaEase version 5.5, Alpha Innotech, San Leandro, CA).

CGT enzymatic activity
The post-nuclear supernatants (prepared as described above) of the Cgt +/ Ϫ , wild-type, and knockout mouse testes were used for the CGT activity assay, following a previously described protocol with some modifi cations ( 35 ). The assay was based on the ability of CGT to transfer galactose to C6-NBD-ceramide (N-[6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino] hexanoyl]D-erythro-sphingosine) to form NBD galactosylceramide (NBD-GC), a fl uorescent product. The post-nuclear supernatant (20 µg of protein) of each sample was incubated (25 min, 37°C) in a 200 µl reaction mixture containing homogenization buffer (described above) with 1% BSA, 2 mM UDP-galactose, 2 mM MgCl 2 , 2 mM MnCl 2 , and 50 µM C6-NBD-ceramide (Avanti Polar Lipids, Alabaster, AL), after which chloroform (250 µl) and methanol (500 µl) were added. After mixing vigorously, additional chloroform and water (250 µl of each) was added. The samples were mixed again before centrifugation (800 g , 5 min, room temperature). The lower chloroform phase was transferred to another glass tube and dried under a stream of nitrogen. The residue was resolubilized in 20 µl of chloroform and applied to a HPTLC plate on which standard C6-NBD-GC (Avanti Polar Lipids) was also loaded. The plate was developed using the same solvent system as described in the HPTLC section, and the chromatographed fl uorescent lipids were detected by a Typhoon 8600 Scanner (Molecular Dynamics). Densitometric analysis of NBD-GC in the samples was done using AlphaImager software.

Statistical analyses
Student's t -test was used to analyze signifi cant differences between two sets of data from the Cgt +/ Ϫ and wild-type samples. A P value of less than 0.05 was considered statistically signifi cant.

Quantifi cation of SGG by Azure A and LC-ESI-MS/MS-MRM
Azure A quantitation consistently overestimated the SGG content of the biological samples compared with LC-ESI-MS/MS-MRM, which was targeted for only C16:0/ C16:0 SGG. However, the magnitude of the overestimation was variable and depended on the purity of the lipid preparations, with high overestimation in crude lipid extracts ( Table 1 ). Both partially and highly purifi ed pig testis SGG samples were separated from major lipid classes via Biosil-A column chromatography. Therefore, relatively tially purifi ed mouse testis sample was not ascribed to the high levels of other sulfolipids. While the exact reason for this great discrepancy is still unknown, it was possible that there existed a signifi cant amount of anionic lipids that reacted with Azure A in this mouse testis sample. These results highlight the caution that should be exercised when the colorimetric assay is used for precise sulfolipid quantitation. In contrast, the clarity of the MRM signals for C16:0/C16:0 SGG and the internal standard, even when working with total crude lipid extracts ( Fig. 4 ), underscored the use of the MRM method.

Levels of C16:0/C16:0 SGG in Cgt +/ ؊ mouse testes and sperm
CGT is the initial enzyme in the SGG biosynthetic pathway that transfers galactose onto the lipid backbone of SGG (1-O -alkyl-2-O -acyl-sn -glycerol). Cgt null males do not contain testicular SGG, and their spermatogenesis is arrested at the pachytene stage ( 11 ). Cgt +/ Ϫ males, however, sire offspring and are routinely used for colony breeding, although it is not clear whether these heterozygous males are subfertile and whether they possess SGG at levels higher than 50% of the wild-type values. Therefore, it is important to know whether the level of SGG, especially the main molecular species C16:0/C16:0, is similar to that of the wild-type animals. LC-MS/MS MRM analyses rethan 90% of the total ion current ( Fig. 3 , bottom panel, B); this spectrum was even somewhat cleaner than the corresponding spectrum of the highly purifi ed pig testis SGG ( Fig.  2 , bottom panel, B). Using the amount of C16:0/C16:0 SGG to calculate the lipid loss during the partial purifi cation process of the mouse testis sample, this value was 16% ( Table 1 ), similar to that seen in the pig testis purifi cation sample (13%). Consequently, the 4.6-fold difference of the amounts of Azure A-reactive lipids over C16:0/C16:0 SGG in the par-

Fertility status of Cgt +/ ؊ male mice
Although Cgt +/ Ϫ male mice sire offspring, their fertility status may still be reduced compared with wild-type males. Results shown in supplementary Fig. II, however, argued against this possibility. Mating studies in fi ve pairs of wildtype and Cgt +/ Ϫ male littermates with wild-type females indicated no signifi cant difference in the litter sizes of the offspring (7.4 ± 0.9 versus 8.4 ± 0.9 pups/litter) (supplementary Fig. IIA). Sperm from Cgt +/ Ϫ males also had the same ability as sperm from wild-type animals to fertilize eggs in vitro (supplementary Fig. IIB). The numbers of caudal epididymal/vas deferens sperm retrieved from Cgt +/ Ϫ males were also not different from those collected from wild-type animals (21.1 ± 6.9 million sperm/mouse versus 18.5 ± 4.3 million sperm/mouse, n = 7). In addition, the testis weight of the heterozygous and wild-type animals were similar (103.7 ± 6.3 versus 110.1 ± 2.6 mg/testis, n = 7).

DISCUSSION
In this study, we described a simple but precise quantification method for C16:0/C16:0 SGG using LC-ESI-MS/ MS MRM with 2 H 3 -C16:0/C16:0 SGG as the internal standard. As expected, the internal standard and unlabeled SGG were virtually indistinguishable on the basis of reverse-phase retention time ( Fig. 4 ) and collisionally activated dissociation conditions ( 24 ). With this internal standard, it was possible to accurately quantitate SGG in crude total lipid extracts from tissues and cells without further purifi cation of the samples. Further, the internal standard included in the MRM assay can be used to adjust for losses during sample preparation; this is not possible with the colorimetric procedure. Normally the internal standard is added at the beginning of the procedure, but in this study, the internal standard was added after the preparation of the extracts so a comparison between Azure A and MRM data could be made with identical samples. Expansion of the MRM procedure to include other SGG isoforms would be straightforward and achieved by simple selection of the appropriate parent → fragment ion transitions. In contrast, caution should be exercised when attempting to quantitate SGG in crude lipid extracts using the Azure A assay. Results from the mouse partially purifi ed SGG sample indicated that it contained high levels of Azure A-reactive components that were not sulfolipids ( Fig. 3 , Table 1 ). Besides sulfolipids, Azure A is known for its binding to anionic compounds ( 15 ). Whereas most phospholipids are nonreactive, polyphosphate phospholipids such as cardiolipin, react with the dye ( 15,16 ). Cardiolipin is present in male germ cells at a level comparable to SGG, and it migrates on the HPTLC plate in close proximity to SGG ( 14 ). Preparative HPTLC, involving scraping chromatographed lipids around the SGG band, as used for partially purifying the mouse testis lipid samples, was likely ineffi cient in removing cardiolipin. In contrast, if the lipid sample is further purifi ed, such as the case for the pig testis samples ( Fig. 2 ), which were chromatographed through a BioSil-A column to remove other lipid classes vealed that Cgt +/ Ϫ mouse testes and sperm contained C16:0/C16:0 SGG at the levels of 78% (406.06 ± 23.63 µg/g testis) and 87% (0.13 ± 0.02 µg/million sperm) of the corresponding wild-type values (516.65 ± 30.62 µg/g testis and 0.15 ± 0.03 µg/million sperm), respectively ( Fig.  5 ). The lower amounts of C16:0/C16:0 SGG in Cgt +/ Ϫ mouse testes were signifi cantly different ( P < 0.05) from those in the wild-type testes, although the decreased levels of C16:0/C16:0 SGG in sperm of Cgt +/ Ϫ males were not signifi cantly different from the wild-type values. In contrast, the levels of total sulfolipids in both testes (2,556.72 ± 312.36 µg/g testis) and sperm (0.60 ± 0.14 µg/million sperm) of Cgt +/ Ϫ mice determined by the Azure A assay were not signifi cantly dissimilar from those of the wildtype mice (2,562.95 ± 308.00 µg/g testis, and 0.73 ± 0.21 µg/million sperm) ( Fig. 5 ).

Compensatory expression of the CGT polypeptide
Since the SGG and sulfolipid amounts in testes and sperm of Cgt +/ Ϫ mice were more than 50% and, in most cases, very similar to the wild-type values, we asked whether the compensation of the CGT expression occurred at the transcriptional or translational level. The Cgt transcript in the testis of Cgt +/ Ϫ mice was ‫ف‬ 40% of the wild-type level, as expected (supplementary Fig. I). However, the CGT polypeptide in the Cgt +/ Ϫ testis was expressed at more than 80% level compared with the wild-type. The CGT polypeptide in the Cgt +/ Ϫ testis also contained enzymatic activity, comparable to the enzyme in the wild-type tissue (supplementary Fig. I). The results indicated the compensatory mechanism of CGT expression at the translational level in Cgt +/ Ϫ testes.  In contrast, the amounts of C16:0/C16:0 SGG in testis was 78% of the wild-type values; however, spermatogenesis in these animals occurred at the same rate as in the wildtype, as revealed by the same numbers of sperm present in the epididymis and vas deferens and the same testicular weights. These results indicated that about 80% of normal testicular C16:0/C16:0 SGG is suffi cient for maintaining spermatogenesis. SGG biosynthesis occurs mainly in the primary spermatocytes ( 39,40 ). During spermiogenesis, when round spermatids become polarized and transformed into elongated spermatids and then testicular sperm, membrane reorganization, including the shedding process of residual bodies, takes place. Mature sperm would likely have reduced levels of SGG, compared with round testicular germ cells, as a result of membrane shedding. It is possible that the shedding of SGG-containing membranes occurred with slightly reduced effi cacy in round spermatids of Cgt +/ Ϫ males, thus correcting SGG levels in their testicular sperm to the normal values. However, both elongated spermatids and testicular sperm were minor components in the total testis homogenate, and the reduced C16:0/C16:0 SGG levels observed would most likely refl ect the decreased lipid levels in round testicular germ cells.
Our results suggest that compensatory translation of active CGT enzyme would be one factor contributing to the 78% SGG expression in the testis of Cgt +/ Ϫ mice, although the Cgt transcript in the heterozygous testis was only half of the wildtype level. Our unpublished results also reveal that the enzymatic activity and polypeptide level of ASA, the fi rst enzyme in the SGG degradation pathway, in the Cgt +/ Ϫ testis were not different from those in the wild-type testis, suggesting that SGG desulfation was not affected in these heterozygous mice. It is still unclear why the expression of C16:0/C16:0 SGG in Cgt +/ Ϫ testis was 20% lower than the wild-type levels. CGT's substrates, palmitylpalmitoylglycerol (PPG) and UDP-Gal, might be present at reduced levels in the Cgt +/ Ϫ testis. The 50% expression of the Cgt transcript might trigger the slowdown of PPG biosynthesis and/or acceleration of its degradation. In the same vein, the expression of the enzyme downstream to CGT in the SGG biosynthetic pathway [i.e., cerebroside sulfotransferase (CST)] may be interlinked with the level of the Cgt transcript, and a reduced level of the CST expression and activity could also result in the 78% level of SGG in Cgt +/ Ϫ mice. In addition, PPG and/or UDP-Gal might have been shunted to alternative synthesis pathways of other sulfated glycolipids to compensate for the reduced C16:0/ C16:0 SGG levels, a situation similar to that observed in the kidney of Cgt null mice ( 41 ). All these possibilities are ongoing studies in our laboratory. Regardless, the compensation of SGG biosynthesis observed in our study is in accordance with that of other structural, lipids such as phosphatidylcholine (PC), which has two synthetic pathways, one utilizing phosphatidylethanolamine methyltransferase (PEMT) and the other CTP-choline cytidyltransferase. Mice genetically null of Pemt still possess the same PC levels in their liver as wild-type animals due to the compensatory increase in the CTP-choline cytidyltransferase activity ( 42 ). Structural lipids are important for maintaining the integrity of the cell membranes and their microdomains, such as lipid rafts, and including phospholipids, the Azure A and MRM data sets were much more convergent ( Table 1 ). Nonetheless, the Azure A procedure for sulfolipid quantifi cation is much less sensitive than the MRM assay (0.5 nmole versus 1 pmole LOD). In general, the Azure A procedure remains a useful technique for obtaining a global picture of the sulfolipid components on an HPTLC plate.
Offsetting the advantages of specifi city and sensitivity is the need for appropriate quantitation standards for LC-ESI-MS/MS-MRM. Both SGG and SGC (commercially available) can be used interchangeably as a standard in the Azure A assay, as both are sulfoglycolipids and elicit similar molar responses in the colorimetric procedure. However, SGC is unsuitable as a quantitative standard in the MRM assay for SGG. SGC is an exceptionally stable gas phase ion, requiring stringent conditions for collisionally activated dissociation compared with SGG ( 18,22,38 ). In addition, besides the m/z 97 ion fragment, the collision-induced fragments of SGC are different from those of SGG ( 34 ). Because we were unable to purify a molecularly homogeneous preparation of SGG from natural sources, it was necessary to chemically synthesize C16:0/C16:0 SGG as a quantitative standard. To use it as an internal standard, it was prepared as 2 H 3 -C16:0/ C16:0 SGG by incorporating stable deuterium atoms into the molecule as part of the synthesis scheme ( 24 ). The MRM-determined concentration of C16:0/C16:0 SGG in mouse testes was about 550 µg/g testis ( Table 1 and Fig. 5 ), somewhat lower than that described by Tadono-Aritomi et al. ( 23 ) (i.e., ‫ف‬ 640 µg/g testis), who also used LC-MS/MS MRM for quantifi cation. Their slightly higher value would likely be attributed to the fact that their measurements were performed with purifi ed natural SGG as the quantitative standard, which in our experience is contaminated with low levels of other SGG molecular species.
The availability of LC-ESI-MS/MS-MRM to specifi cally quantify C16:0/C16:0 SGG allowed assessment of the level of this sulfoglycolipid that is required for maintaining male fertility. Cgt null males without the ability to synthesize SGG have their spermatogenesis disrupted and are thus infertile. In contrast, Cgt +/ Ϫ males sire offspring, although their detailed fertility status is unknown ( 11 ). It is possible that these heterozygous males are subfertile, a condition commonly found in men. However, the results from the mating and in vitro fertilization experiments reported herein argued against this possibility. Sperm from Cgt +/ Ϫ males could fertilize mature eggs to the same extent as wild-type sperm. The litter size of offspring generated from mating these heterozygous males with wild-type females was also the same as that produced by wild-type males in the parallel mating experiment. Notably, the C16:0/C16:0 SGG levels in sperm of Cgt +/ Ϫ males were indistinguishable from those in wild-type sperm. Our accumulated studies indicate the signifi cance of sperm surface SGG in egg interactions ( 1,8 ); it has direct affi nity for the ZP and also acts as a major participant in the formation of lipid rafts, which are ZP binding platforms on the sperm head surface ( 5,7,10 ). Therefore, the normal amounts of C16:0/C16:0 SGG would allow uncompromised function of sperm from Cgt +/ Ϫ males.