Multiplex analysis of sphingolipids using amine-reactive tags (iTRAQ).

Ceramides play a crucial role in divergent signaling events, including differentiation, senescence, proliferation, and apoptosis. Ceramides are a minor lipid component in terms of content; thus, highly sensitive detection is required for accurate quantification. The recently developed isobaric tags for relative and absolute quantitation (iTRAQ) method enables a precise comparison of both protein and aminophospholipids. However, iTRAQ tagging had not been applied to the determination of sphingolipids. Here we report a method for the simultaneous measurement of multiple ceramide and monohexosylceramide samples using iTRAQ tags. Samples were hydrolyzed with sphingolipid ceramide N-deacylase (SCDase) to expose the free amino group of the sphingolipids, to which the N-hydroxysuccinimide group of iTRAQ reagent was conjugated. The reaction was performed in the presence of a cleavable detergent, 3-[3-(1,1-bisalkyloxyethyl)pyridine-1-yl]propane-1-sulfonate (PPS) to both improve the hydrolysis and ensure the accuracy of the mass spectrometry analysis performed after iTRAQ labeling. This method was successfully applied to the profiling of ceramides and monohexosylceramides in sphingomyelinase-treated Madin Darby canine kidney (MDCK) cells and apoptotic Jurkat cells.


Amine-reactive tagging of sphingoid base
Dried samples were resuspended in a mixture of 20 µl of 0.5 M triethylammonium bicarbonate buffer and 30 µl of ethanol. In case of samples hydrolyzed with SCDase, 10 µl of 0.5 M triethy lammonium bicarbonate buffer and 30 µl of ethanol were added to lysosphingolipids in aqueous phase. iTRAQ reagents were resuspended in 70 µl of ethanol, and 30 µl of the reagents were added to the samples. The tagging reaction was carried out by incubation at room temperature for 1 h, followed by 30 min incubation after the addition of 0.1% trifl uoroacetic acid aqua to hydrolyze excess iTRAQ reagent and PPS. The labeled sphingolipids were combined and injected onto a solid-phase extraction column (NOBIAS RP-OD1D, Hitachi High-Technologies Corp., Tokyo, Japan) to remove salt and excess reagents. After washing with 40% methanol aqua, the labeled sphingolipids were eluted with chloroform-methanol (9:1, v/v). To remove residual PPS, the eluted solution was injected onto a Si column (InertSep Si, 50 mg / 1 ml, GL Sciences, Tokyo, Japan), washed with chloroformmethanol (9:1, v/v), and eluted with methanol. The eluted sphingolipids were dried and stored at 2 20°C until use.

Mass spectrometry
An Agilent 1100 series LC (Agilent Technologies, Santa Clara, CA) coupled to a 4000 QTRAP hybrid triple quadrupole/linear ion trap mass spectrometer (AB SCIEX) was used to analyze the lipid samples. The samples were injected onto a reversed-phase C18 column (CAPCELL PAK C18 MG III, 2.0 × 50 mm, Shiseido Co., Ltd., Tokyo, Japan) at 0.3 ml/min. Solvent A [methanolwater-formic acid (58:41:1, v/v/v) with 5 mM ammonium formate] and solvent B [methanol-formic acid (99:1, v/v) with 5 mM ammonium formate] were used as eluent. The samples were eluted through the following gradient condition: Solvent A/B (6:4) 0.5 min, followed by a linear gradient until A/B (0:10) over the next 2.5 min. After 5 min at 100% solvent B, the gradient was brought back to A/B (6:4) over 0.5 min and the column was then equilibrated for 3.5 min. The mass spectrometer was run in the positive ion mode with the following instrument parameters: curtain gas of 30, ion spray voltage of 3,500, temperature of 450, nebulizer gas of 50, auxiliary gas of 50, and interface heater on. Multiple reaction monitoring (MRM) of sphingolipids was performed determined is the sum of the intensity from multiple samples. In the course of MS, iTRAQ tags fragment to release tag-specifi c reporter ions. The ratios of these reporter ions are representative of the proportions of each peptide in the individual samples. The reactive group on the iTRAQ tags is an N-hydroxysuccinimide moiety, which interacts with primary amine groups. iTRAQ is widely employed for the quantifi cation and profi ling of proteins ( 21,22 ). Recently, iTRAQ has been applied successfully to the labeling of aminophospholipids ( 23 ).
In the present study, we employed iTRAQ to quantify ceramides and monohexosylceramides. Ceramides and monohexosylceramides were hydrolyzed with sphingolipid ceramide N -deacylase (SCDase) to release fatty acids ( 24 ), and the resulting compounds with their free amino groups were labeled with iTRAQ reagents. Simultaneous measurement of multiple samples after sphingomyelinase treatment revealed the formation of ceramides characterized by their sphingoid base structure. We also applied this method to measure the increase of ceramide during apoptosis.

Hydrolysis of N -acyl linkage of sphingolipids by SCDase
SCDase hydrolysis was performed by the aqueous-organic biphasic method described previously ( 25 ) with modifi cation. An amount of 10 µl of 50 mM sodium acetate, pH 6.0, containing 1% PPS and 5 mU of SCDase were added to dried lipids. After mixing, 100 µl or 500 µl of n-decane were added, and the biphasic mixture was incubated for appropriate intervals at 37°C. To facilitate hydrolysis, the upper organic solution was exchanged several times during incubation. The reaction was monitored by analyzing the lipids in aqueous phase by TLC with chloroform-methanol-25% NH 4 aqua (90:20:0.5, v/v/v) (for ceramide, glucosylceramide, and galactosylceramide analysis) or with chloroform-methanol-25% NH 4 aqua (5:4:1, v/v/v) (for sphingomyelin analysis). The lipids were visualized using copper sulfate spray, and then scanned using a LAS 4000 Mini Biomolecular Imager (GE Healthcare, Waukesha, WI). twice with PBS, and then scraped into 1 ml of 2 mM EDTA aqua. Lipid extraction was performed from 800 µl of the cell suspension by the Bligh and Dyer method ( 29 ), and protein concentrations were determined from remaining suspension by a Protein Assay Kit (BioRad, Hercules, CA). The lipid solutions were evaporated under N 2 gas and stored at 2 20°C until use. To remove endogenous sphingoid bases, the lipid mixture was fractionated with a Si column (InertSep Si, 50 mg / 1 ml, GL Sciences). The lipids in chloroform-methanol (9:1, v/v) were injected onto the Si column, and the fl ow-through fraction was used for analysis of ceramide and monohexosylceramide.

Anti-Fas induced apoptosis of Jurkat cells
Jurkat cells were maintained in medium B [RPMI 1640 medium (Sigma-Aldrich) supplemented with 10% FCS and penicillin/ streptomycin] ( 30 ). Apoptosis was induced by the addition of 500 ng anti-Fas antibody to 1 × 10 6 cells in 5 ml of medium B. At appropriate intervals, cells were washed twice with PBS by centrifugation. The pellet was resuspended in 1 ml PBS. The lipids were analyzed as described above. under optimal conditions as described previously ( 26 ). The acquisition parameters in MRM and chromatographic retention times of the iTRAQ-labeled sphingoid bases are listed in Table 1 . Data acquisition and analysis were performed using Analyst Software version 1.4.1 (AB SCIEX). Lipid identifi cations were determined by MS/MS in enhanced product ion mode. Relative quantifi cation was performed with peak areas observed in MRM using m/z 114.1, 115.1, 116.1, and 117.1 as Q3 parameters each. The peak area ratios were corrected with Cramer's rule described previously for overlapping isotopic contribution ( 27 ).

Sphingomyelinase treatment of MDCK cells
Madin Darby canine kidney (MDCK) cells were maintained in medium A (DMEM low glucose (1 g/l) supplemented with 10% FCS and penicillin/streptomycin) ( 28 ). Cells were seeded at 1.2 × 10 6 cells in 60 mm dishes. After two days, cells were washed twice with HBSS, and then treated with 1 U/ml SMase for 0, 15, 30, and 90 min at 37°C. After SMase treatment, cells were washed   between fatty acids and sphingoid bases ( 24,34,35 ). An aqueous-organic biphasic system has been used to improve the enzymatic hydrolysis because the condensation reaction is inhibited by the diffusion of fatty acids in the organic phase ( 25,36 ). However, the degree of hydrolysis did not exceed 70% under our experimental conditions ( Fig. 2 ). Thus, we tried to improve the effi ciency of the reaction. Fig. 3A shows the effects on hydrolysis of vortex mixing, sonication, and the exchange of the upper organic solvent. Although vortex mixing and sonication did not signifi cantly improve the reaction, exchange of the upper organic solvent greatly increased hydrolysis. As shown in Fig. 3B , multiple-exchange of the upper phase further accelerated hydrolysis. These results indicated that the elimination of free fatty acids facilitates the activity of SCDase. Fig. 4 shows the effect of the fatty acid chain length of glucosylceramide on the hydrolysis effi ciency achieved using SCDase. After four times solvent exchange, approximately 80% of the C12:0 glucosylceramide was hydrolyzed, whereas more than 90% of C24:1 glucosylceramide was degraded. However, all of the glucosylceramides tested were hydrolyzed more than 94% after eight times exchange of organic solvent. Thus, the multiple solvent exchanges resulted in almost complete hydrolysis of glucosylceramide irrespective of the fatty acid composition. We then examined the hydrolysis of different sphingolipids using SCDase. Ceramide, glucosylceramide, galactosylceramide, and sphingomyelin were hydrolyzed with SCDase at 37°C for 16 h ( Table 2 ). The solvent was exchanged every 2 h

Fluorescence microscopy
An amount of 100 m l of 500 ng/ml anti-Fas antibody in medium B was added to 5 × 10 4 Jurkat cells in 100 m l medium B. At appropriate intervals, 4 m g/ml Hoechst 33342 (fi nal concentration) and Alexa Fluor 488-conjugated Annexin V were added. Specimens were observed under an LSM 510 confocal microscope equipped with a C-Apochromat 63XW Korr (1.2 NA) objective (Carl Zeiss, Oberkochen, Germany). Fig. 1 shows the scheme of the quantitative analysis of sphingolipids using iTRAQ. To expose the primary amine group, lipids were hydrolyzed with SCDase ( 24,25 ). SC-Dase has an advantage over alkaline hydrolysis (31)(32)(33) in removing fatty acid from sphingolipids because the sugar moiety of glycosphingolipids remains intact. The activity of SCDase is enhanced by the addition of detergents, such as sodium taurodeoxycholate or sodium cholate ( 25 ). However, common detergents are diffi cult to remove from the sample solution and hamper MS. We employed PPS, a cleavable detergent that is chromatographically removable after acidic cleavage. Fig. 2 shows the effective concentration of PPS for the hydrolysis of glucosylceramide using SCDase. Hydrolysis was maximum in the presence of 1% PPS.
Labeling with iTRAQ changes molecular properties such as mass, polarity, and ionization effi ciency. Thus we examined the effect in the labeling of typical sphingoid bases. Cer/Sphingoid Internal Standard Mixture I, containing Sphingosine d17:1, sphinganine d17:0, sphingosine-1phosphate d17:1, and sphinganine-1-phosphate d17:0, was labeled with an iTRAQ reagent and analyzed in MRM mode. The changes of parameters before and after labeling are summarized in Table 3 and Fig. 5 . There were little changes in retention time on the reversed-phase chromatography, while the labeling provided considerable enhancement of sensitivity. Especially, the ratio of peak area of sphinganine during incubation. Under these conditions, more than 90% of ceramide, glucosylceramide, and galactosyceramide were hydrolyzed, whereas 60% of sphingomyelin became lyso-sphingomyelin. The hydrolysis of glucosylceramide, galactosylceramide and sphingomyelin under our conditions was much more effi cient than that of previous results ( 24 ). Note that ceramide was very effi ciently hydrolyzed in our study whereas SCDase was previously reported to hardly hydrolyze ceramide ( 24 ). The effi cient hydrolysis    . 7. The profi les of ceramides (A) and hexosylceramides (B) during anti-Fas antibody-induced apoptosis of Jurkat cells. Temporal profi les of ceramides and hexocylceramides were detected as described in Fig. 6 . Data represent the mean ± SD (n = 4). ** P < 0.01, * P < 0.02. could detect minor sphingoid species, such as d20:0 dihydroceramide and t16:0 and t20:0 phytoceramides. The temporal profi les based on ratiometric quantifi cation of iTRAQ revealed a rapid increase in ceramides (Fig. 6A) followed by an increase in monohexosylceramides (Fig.  6B). The peak areas of the monohexosylceramides are summarized in Table 5 . Fig. 6 suggests that the increase in hexosylceramides is secondarily derived from the rapid change in ceramides, as previously described (37)(38)(39). The ceramide profi le indicates that the kinetics of ceramide formation depend on the structure of the sphingoid base; i.e., ceramides and dihydroceramides rapidly increased within 15 min, followed by a gradual decrease. In contrast, phytoceramides continued to slowly increase after 15 min incubation. Fas (CD95) engagement by Fas ligand has a crucial function in the apoptotic elimination of T cells undergoing environmental trauma ( 40 ). The increase of ceramides during Fas-induced apoptosis is a matter of debate (41)(42)(43)(44). Fig. 7 shows the increase of ceramides and hexosylceramides during anti-Fas antibody-induced apoptosis of dramatically increased. Because of the lack of double bond, ionization effi ciency of sphinganine is lower than that of sphingosine. As a result, the sensitivity of MS to sphinganine in the positive ion mode is low. It is speculated that the introduction of two nitrogen atoms of iTRAQ reagent to sphinganine signifi cantly enhanced the sensitivity by increasing the affi nity of the molecule to proton.
Sphingomyelin is enriched in the outer leafl et of the plasma membrane ( 30 ) and is cleaved to ceramide by adding SMase to the medium. We applied the established method to the profi ling of ceramides and monohexosylcermides in SMase-treated MDCK cells ( Fig. 6 ) . Lipids were extracted from cells and separated from free sphingoid base by means of a Si column. The samples (0, 15, 30, and 90 min) were hydrolyzed with SCDase and labeled with four different iTRAQ reagents. The labeled lipids were combined and applied to MS. We identifi ed nine ceramides and fi ve monohexosylceramides, containing dihydroceramide and phytoceramide, with three different chain-length sphingoid bases. The relative peak areas of these ceramides are summarized in Table 4 . This method iTRAQ and molecular species analysis is able to provide a quantitative and detailed profi le of ceramide metabolism that is crucial for understanding the complex physiological roles of ceramides.
In this study, we report a new application of the iTRAQ technique to the quantitative analysis of ceramides and monohexosylceramides after hydrolysis with SCDase. iTRAQ allows simultaneous measurement of up to eight samples, and it dramatically reduces measurement time. Moreover, iTRAQ enables the accurate profi ling of lipids, even if abundance is low. This study represents the fi rst application of iTRAQ to analyze sphingolipid metabolism.
In addition to the heterogeneity of fatty acids, ceramides contain various sphingoid bases. The physiological significance of different sphingoid bases in ceramides and other sphingolipids has been reported. The major sphingoid bases of mammalian cells are sphingosine and dihydrosphingosine. In contrast to sphingosine-conjugated ceramides, dihydroceramides (which contain dihydrosphingosine) do not induce apoptosis ( 6,(48)(49)(50). Phytosphingosine-based sphingolipids are also observed in skin, intestine, and kidney (51)(52)(53). Although the importance of phyto-type sphingolipids in these cells is recognized ( 54, 55 ) (e.g., synthetic phytoceramides are more cytotoxic than ceramides ( 56 )), the precise role of the lipids remains unsolved. The function of the different chain lengths of sphingoid bases is also not clear. Recent study indicates that serine palmitoyltransferase containing subunit SPTLC3 generates short chain sphingoid bases ( 57 ). However, the role of short chain sphingolipids is not known. Our method is applicable to study the detailed profi les of ceramides and monohexosylceramides with minor sphingoid bases.
Furthermore, iTRAQ can be used to directly analyze endogenous sphingoid bases, such as sphingosine, S1P, and other atypical sphingoid bases, without needing previous enzymatic reaction. Although the enzymatic procedure requires additional steps, each step is simple, and multiplex analysis of labeled samples provides minimal experimental error. Such analysis will help to carry out a detailed kinetic analysis of endogenous sphingolipids under physiological conditions. In addition, iTRAQ can be used as a precise tool for analyzing the kinetics of lysoglycosphingolipid formation in lysosomal storage disorders ( 58,59 ).
Jurkat T cells. Apoptosis was monitored by measuring the fragmentation of the nucleus and the exposure of phosphatidylserine to the cell surface, as monitored by the binding of Alexa Fluor 488-conjugated Annexin V ( 45 ) ( Fig. 8 ). Fragmentation of the nucleus and the exposure of phosphatidylserine were observed 1 h after the addition of the anti-Fas antibody, and most of the cells became annexin V-positive after 4 h incubation. Fig. 7A indicates the time-dependent increase in ceramides. This result is consistent with previous results using HPLC ( 42 ) and a diacylglycerol kinase method ( 43 ) ( 44 ). In contrast to ceramides, the content of hexosylceramides was not signifi cantly altered during apoptosis, as previously observed ( 44 ) ( Fig. 7B ).
One of the drawbacks of using iTRAQ to analyze sphingolipids is that fatty acid profi les cannot be obtained. Recent results suggest that ceramides with different fatty acid chain lengths might have distinct functions ( 46,47 ). Fig. 9 shows the relative amount of the various ceramide molecular species at time 0 and 6 h after anti-Fas treatment. Fig. 9 depicts a signifi cant increase in the d18:1/C16:0, d16:1/ C16:0 and d16:1/C24:1 ceramides, but not the d18:1/ C24:2 ceramides, during anti-Fas-induced apoptosis in Jurkat cells. These results indicate that the combination of Fig. 9. The profi les of ceramide molecular species during anti-Fas antibody-induced apoptosis of Jurkat cells. MRM data were obtained as described in Materials and Methods. Data represent the mean ± SD (n = 4).