Separation and quantification of 2-acyl-1-lysophospholipids and 1-acyl-2-lysophospholipids in biological samples by LC-MS/MS.

Lysophospholipids (LysoGPs) serve as lipid mediators and precursors for synthesis of diacyl phospholipids (GPs). LysoGPs detected in cells have various acyl chains attached at either the sn-1 or sn-2 position of the glycerol backbone. In general, acyl chains at the sn-2 position of 2-acyl-1-LysoGPs readily move to the sn-1 position, generating 1-acyl-2-lyso isomers by a nonenzymatic reaction called intra-molecular acyl migration, which has hampered the detection of 2-acyl-1-LysoGPs in biological samples. In this study, we developed a simple and versatile method to separate and quantify 2-acyl-1- and 1-acyl-2-LysoGPs. The main point of the method was to extract LysoGPs at pH 4 and 4°C, conditions that were found to completely eliminate the intra-molecular acyl migration. Under the present conditions, the relative amounts of 2-acyl-1-LysoGPs and 1-acyl-2-LysoGPs did not change at least for 1 week. Further, in LysoGPs extracted from cells and tissues under the present conditions, most of the saturated fatty acids (16:0 and 18:0) were found in the sn-1 position of LysoGPs, while most of the PUFAs (18:2, 20:4, 22:6) were found in the sn-2 position. Thus the method can be used to elucidate the in vivo role of 2-acyl-1-LysoGPs.

and adjusting apparent pH to 4.0 by adding formic acid (approximately 1,160 l) using a pH meter (D-21; Horiba).

Animals
C57BL/6 mice were purchased from CLEA Japan and maintained according to the Guidelines for Animal Experimentation of Tohoku University . The animal protocol was approved by the Institutional Animal Care and Use Committee at Tohoku University (No. 17-Pharm-Animal-2012).

LC-MS/MS analysis
LC-MS/MS analysis was principally performed as described previously with minor modifi cations ( 11,12 ). The LC-MS/MS system consisted of a NANOSPACE SI-II HPLC (Shiseido) and a TSQ Quantum Ultra triple quadrupole mass spectrometer (Thermo Fisher Scientifi c, San Jose, CA) equipped with a heated-electrospray ionization-II (HESI-II) source. The positive and negative HESI-II spray voltages were 3,500 and 2,500 V, respectively, the heated capillary temperature was 350°C, the sheath gas pressure was 60 psi, the auxiliary gas setting was 40 psi, and the heated vaporizer temperature was 350°C. Both the sheath gas and auxiliary gas were nitrogen. The collision gas was argon at a pressure of 1.5 mTorr. The LC-MS/MS system was controlled by Xcalibur software (Thermo Fisher Scientifi c) and data were collected with the same software. LysoGP analyses were performed in the multiple reaction monitoring (MRM) mode, in positive ion mode for LPC and in negative ion mode for LPA, lysophosphatidylethanolamine (LPE), lysophosphatidylglycerol (LPG), LPI, and LysoPS. The collision energy was optimized for each compound to obtain optimum sensitivity using argon as collision gas. The collision energy settings, tube lens offsets, and MRM transitions for all analytes are summarized in Table 1 . LC separation was performed using a reverse-phase column [Capcell Pak ACR (250 mm × 1.5 mm inner diameter , 3 m particle size; Shiseido)] with a gradient elution of solvent A (5 mM ammonium formate in water, pH are unstable and are quickly converted to the corresponding 1-acyl-2-LysoGPs by a spontaneously occurring intra-molecular acyl migration reaction ( 10 ), yielding a mixture of 1-acyl-2-LysoGPs and 2-acyl-1-LysoGPs. Plückthun and Dennis ( 10 ) demonstrated that 1-palmitoyl-2-lysophosphatidylcholine (LPC) prepared by PLA 2 is actually the equilibrium mixture consisting of approximately 90% of the 1-acyl-2-lyso isomers and 10% of the 2-acyl-1-lyso isomers . They also showed that the rate of the acyl migration was pH dependent, with a minimum around pH 4 to pH 5. Their technique was based on 31 P NMR and thus could be used only on pure samples. In the present study, we developed a simple and versatile method to separate and quantify 2-acyl-1-and 1-acyl-2lyso isomers using LC-MS/MS under conditions that completely eliminate the acyl migration reaction. Using this method, we determined the distribution of 2-acyl-1-LysoGPs and 1-acyl-2-LysoGPs with various head groups in biological samples. To the best of our knowledge, this is the fi rst report to precisely determine the distribution of acyl chains between the sn -1 and sn -2 positions.

Materials
All phospholipids {dioleoyl-phosphatidylserine (PS), dilinoleoyl-GPs [phosphatidic acid (PA), phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), phosphatidylinositol (PI), and PS], 17:0-LPA, and 17:0-LPC were purchased from Avanti Polar Lipids. Phospholipids were dried in borosilicated glass tubes (9831-1207; Iwaki Glass) under nitrogen gas, dissolved in PBS using a water bath sonicator, and stored at Ϫ 20°C. Other materials were purchased from Wako Pure Chemical Industries, unless otherwise noted. Acidic methanol (pH 4.0) was prepared by mixing 1 ml of 1 M ammonium formate and 99 ml of methanol (Kanto Chemical), Oleoyl-LysoPS was detected in two peaks in the chromatogram. Because the faster migrating peak was dominant when PLA 1 was used and the slower peak was dominant when PLA 2 was used, the faster peak was attributed to 2-oleoyl-1-LysoPS and the slower peak was attributed to 1-oleoyl-2-LysoPS. 4.0) and solvent B (5 mM ammonium formate in 95% (v/v) acetonitrile, pH 4.0) at 150 l/min. Solvent A was prepared by mixing 95 ml of Milli-Q (Millipore) water and 5 ml of 100 mM ammonium formate, and adjusting to pH 4.0 with formic acid (approximately 9.6 l). Similarly, solvent B was prepared by mixing 95 ml of acetonitrile and 5 ml of 100 mM ammonium formate, and adjusting to apparent pH 4.0 with formic acid (approximately 1,160 l). The initial condition was set at 55% B.

Quantifi cation of lysophospholipids in various tissues
Mouse tissues (approximately 100 mg) were placed in 1.5 ml siliconized sample tubes. Then, nine volumes of acidic methanol (pH 4.0), 1 M (fi nal concentration) of 17:0-LPA, and 10 M (fi nal concentration) of 17:0-LPC were added to the tube. The obtained mixture was homogenized for 10 min by Micro Smash MS-100R (TOMY) (3,000 rpm at 4°C). After an initial centrifugation at 1,000 g for 10 min at 4°C, the supernatants were further centrifuged at 21,500 g for 10 min at 4°C. The resulting supernatant was passed through a fi lter (0.2 m pore size, 4 mm inner diameter; YMC), and 10 l of the fi ltrate was subjected to LC-MS/MS.
To obtain a serum sample, a blood sample was collected with a noncoated capillary (910-01-75; Hirschmann Laborgerate), incubated at 37°C for 1 h, allowed to stand at 4°C for 12 h, and centrifuged at 1,500 g for 10 min at 25°C. Plasma was collected with a heparinized capillary (1-040-7500-HC; Drummond), followed by addition of 1 mM of EDTA (fi nal concentration), and centrifuged at 1,500 g for 10 min at 25°C. The plasma and serum samples (10 l each) were placed in 1.5 ml siliconized sample tubes, deproteinized by mixing with 90 l acidic methanol (pH 4.0), homogenized for 3 min in an ultrasonic bath, and centrifuged at 21,500 g for 10 min at 4°C (ice-cold water). The supernatants were fi ltered and 10 l of the fi ltrate was subjected to the LC-MS/MS.

Method validation
A mixture of 1 mM 18:1-LPC, 100 M 18:1-LPA, LPE, LPG, LPI, and LysoPS was prepared in methanol. It was diluted with methanol The following solvent gradient was applied: maintained 55% B for 10 min and then followed by a linear gradient to 85% B from 10 to 30 min, and hold at 85% B for 7 min. Subsequently, the mobile phase was immediately returned to the initial conditions and maintained for 3 min until the end of the run. Column effl uent was introduced into the mass spectrometer between 3 and 37 min after injection.

Evaluation of the acyl-migration reaction
To examine the effect of pH and temperature, a mixture of dilinoleoyl-GPs (PA, PC, PE, PG, PI, and PS) was incubated with 3.0 g/ml (660 lipase units/ml) Rhizomucor miehei lipase (Sigma; L4277), which has intrinsic PLA 1 activity, at 37°C for 3 h. The reactant was dissolved in methanol, whose pH was adjusted between 3 and 10 by adding formic acid or 1 M NaOH. After samples were left at Ϫ 80°C, Ϫ 20°C, 4°C, or room temperature (approximately 25°C) for 0, 1, 2, and 7 days, LysoGPs were analyzed by LC-MS/MS.
To examine the effects of albumin on the stability of 2-acyl-1lyso-GPs, dioleoyl-PS was incubated with recombinant PS-specifi c PLA 1 (PS-PLA 1 ) at 37°C for 30 min and then mixed with 100 g/ml Olristat (Roche; a lipase inhibitor) to stop the reaction. The solution was mixed with 0.1% (w/v) BSA (Sigma; fatty acid-free grade, A6003) and further incubated at 37°C. At the indicated time points, an aliquot (10 l) was placed into a 1.5-ml sample tube. Then, acidic methanol (pH 4.0, 190 l) was added to deproteinize the sample. The obtained mixture was homogenized for 3 min in an ultrasonic bath (ice-cold water). After centrifugation at 21,500 g for 10 min at 4°C, the supernatant was subjected to LC-MS/MS.

Preparation of recombinant PS-PLA 1 enzyme
Recombinant PS-PLA 1 proteins were prepared as described previously ( 9 ). Briefl y, HEK293 cells were transfected with expression plasmids encoding mouse wild-type PS-PLA 1 or mutant catalytically inactive PS-PLA 1 (PS-PLA 1 S166A), which has substitution of the catalytic center serine (

Separation of 2-acyl-1-lyso-GPs and 1-acyl-2-lyso-GPs by reverse-phase HPLC
We previously established a method to separate a wide range of LysoGP species including LPC, LPE, LPI, LysoPS,  LPG, LPA, and sphingosine 1-phosphate ( 11,12 ). In this method LysoGPs are separated by octadecyl (C18) reverse-phase LC and detected by MS/MS. This method was found to separate 1-acyl-2-LysoGPs from the corresponding 2-acyl-1-lyso isomers ( Fig. 1 ). Digestion of dioleoyl-PS with PLA 2 and monitoring of LysoPS (with m/z 522.2 in negative mode) resulted in two separate peaks (elution times at 10.7 and 11.2 min, respectively ). The fi rst peak was dominant when PLA 1 was employed and the second peak was dominant when PLA 2 was employed, indicating that the first and the second peaks correspond to 2-acyl-1-LysoPS and 1-acyl-2-LysoPS, respectively. The order of elution in reverse-phase LC was consistent with a previous report ( 13,14 ). The presence of 2-acyl-1-LysoPS in a PLA 2 reactant and 1-acyl-2-Lys-oPS in a PLA 1 reactant indicated that acyl chain migration occurred during the phospholipase reactions. 1-Acyl-2-LysoGPs and 2-acyl-1-LysoGPs with other polar head groups could also be separated under these conditions, and 2-acyl-1-LysoGPs always eluted faster than the corresponding 1-acyl-2-LysoGPs (data not shown). The standard curve for each LysoGP was constructed by plotting against the peak area ratio of the analytes to the internal standard ( Fig. 2 ). The standard curves for LPC species were linear from 0.1 to 500 M. The regions in which the standard curves for other LysoGPs were linear were similar ( Table 2 ). Intra-and inter-day assay precision were 0.6-13.8% and 0.1-39.3%, respectively ( Table 3 ). In the following experiments, we quantifi ed the amount of LysoGPs within the calibration range.

2-Acyl-1-LysoGPs are stable in acidic and low temperature conditions
2-Oleoyl-1-LysoPS deteriorated rapidly at room temperature but remained stable for at least 1 week at Ϫ 80°C ( Fig. 3A ). 2-Oleoyl-1-LysoPS deteriorated rapidly at pHs of 7 and above, but remained stable at pHs of 5 and below ( Fig. 3B ). 1-Acyl-2-LysoGPs with other head groups were also stable at pH 4 ( Fig. 3C ). By contrast, at neutral and pH 9, they were quickly converted to the corresponding 1-acyl-2-lyso isomers at different rates, with rates in the order LPE > LPS > LPC = LPA, LPG, LPI ( Fig. 3C ). Thus, in the following experiments we extracted lipids using methanol with an apparent pH 4.0 on ice and kept the samples at Ϫ 20°C before analyses.

Quantifi cation of 2-acyl-1-LysoGPs and 1-acyl-2-LysoGPs in biological samples
LysoPSs with 16:0, 18:0, 18:1, 18:2, 20:4, 22:5, and 22:6 were detected in mouse liver. Interestingly, LysoPSs with was detected in various tissues, contained signifi cant amounts of 1-acyl-2-LPC ( Fig. 5B ). In addition, most saturated LPGs detected in the brain and spleen were found to be 2-acyl-1-LPG ( Fig. 5D ). As expected, the uneven fatty acid distribution of LysoGPs in serum was completely lost, probably because of the 1 h incubation at 37°C. In sharp contrast, freshly prepared plasma contained both saturated and unsaturated fatty acids that were distributed unevenly between the sn -1 and sn -2 positions, as was the fatty acid distribution of LysoGPs from various tissues . To demonstrate the utility of this method, we examined the level of LysoGPs in the samples ( Table 5 ). Some of the LysoGPs that were detected could not be quantified because their concentrations were below the limit of quantitation. These LysoGPs were labeled as not quantified (NQ) in Table 5 , and those that were not detected were labeled ND.
We also examined the LysoGP species in various mouse tissues and freshly prepared plasma and serum ( Fig. 5A-F ). The fatty acid distributions of other LysoGPs were similar to those observed for LysoPS in the mouse liver: saturated fatty acids (16:0 and 18:0) were detected mainly in the sn -1 position and unsaturated fatty acids (18:2, 20:4, 22:6, etc.) were detected mainly in the sn -2 position. Monounsaturated fatty acids (18:1, 16:1 in the case of LPC) distributed differently depending on the polar heads and origin of tissues ( Fig. 5B ). Interestingly, however, we observed some exceptions. For example, saturated LPC detected in various tissues contained significant amounts of 2-acyl-1-LPC, and unsaturated LPC, which

Albumin accelerates acyl migration
Finally, we attempted to detect 2-acyl-1-LysoGPs at the cellular levels. We chose PS-PLA 1 as an enzyme to produce 2-acyl-1-LysoPS on the cell surface. PS-PLA 1 specifi cally cleaves PS and is capable of cleaving PS on the outer surface of plasma membrane without any effect on cell viability ( 15 ). When HEK293 cells were treated with recombinant PS-PLA 1 , a signifi cant amount of 2-oleoyl-1-LysoPS and a lesser amount of 2-DHA-1-LysoPS were detected in the cells but not in the cell supernatant. 1-Oleoyl-2-LysoPS was also detected, but the level was less than the level of 2-oleoyl-1-LysoPS. These LysoPS species were produced by PS-PLA 1 because none of them were detectable when catalytically inactive PS-PLA 1 mutant was used ( Fig. 6A ). Interestingly, when the conditioned media were supplemented with albumin, which binds and extracts LysoGPs from the cell surface, most of the Lys-oPS was detected in the cell supernatant as a 1-oleoyl-2-isomer ( Fig. 6B ). This suggested that 2-acyl-1-LysoPS was fairly stable on the cell membrane in the absence of albumin and that albumin added to the media extracted 2-acyl-1-LysoPS and converted it into 1-acyl-2-LysoPS. This was confi rmed by the fi nding that the acyl migration was dramatically accelerated in the presence of 0.1% BSA ( Fig. 7 ).

DISCUSSION
It has been difficult to quantify the level of 2-acyl-1-LysoGP species because of acyl migration reaction. In the current study, we minimized the acyl migration reaction and established a simple and versatile method for separation and quantifi cation of 1-acyl-2-LysoGPs and 2-acyl-1-LysoGPs. The preceding paper by Plückthun et al. indicated that the acyl migration reaction of LPC was highly sensitive to pH and was effectively suppressed by lowering pH to 4 or 5 (10) . In this study, we utilized this property and determined the pH-dependent stability of all the LysoGP classes, including LPC, LPA, LPE, LPG, LPI, and LysoPS. The result was essentially the same as was reported by the preceding paper for LPC (10). However, we obtained new information about the acyl migration reaction of LysoGPs. For example, the speed of the acyl migration reaction is highly dependent on the head group of LysoGPs (a rank order of migration rate is as follows: LPE > LysoPS > LPC > LPI > LPG, LPA; Fig. 3C ). Furthermore, we found that the acyl migration reaction was sensitive to the temperature. By keeping the temperature below Ϫ 20°C, we succeeded in suppressing the reaction completely. The method utilized in the preceding paper required 31 P-labeled LysoGPs and NMR equipment ( 10 ). Thus, it was not readily applicable to biological samples. In contrast, the present method is applicable to biological samples, such as tissues and cells in culture, by measuring LysoGPs directly.
In the current study, we established an LC method for separating 2-acyl-1-LysoGPs and 1-acyl-2-LysoGPs and examined the conditions in which the acyl migration was minimal ( Fig. 1 ). The previous study by Creer and Gross ( 14 ) showed that 1-acyl-2-LPC and 2-acyl-1-LPC could be separated on reverse-phase LC . However, the following modifi cations were applied. By comparing several reverse-phase columns, we chose CAPCELLPAK C18 ACR (Shiseido), which enabled us to separate the 1-acyl-2-lyso isomers and 2-acyl-1-lyso isomers of all the LysoGP species (data not shown). Indeed, the present method made it possible to analyze as much as 100 LysoGP species with different head groups and acyl chains within the range from several picomoles to nanomoles. In addition, we used the LC solvents with pH 4.0 to avoid possible acyl migration reaction during LC separation. Moreover, our method, adding the methanol only, is very easy and has high extract effi ciency ( Table 5 ). In our laboratory, by using an autosampler, simultaneous analyses of several hundred samples are possible.
It was revealed in this study that the pattern of acyl chain distribution in fresh plasma was totally different from that in serum. Most PUFA-containing LysoGPs detected in freshly prepared mouse plasma were 2-acyl-1-lyso isomers and most saturated LysoGPs were 1-acyl-2-lyso  isomers, indicating that plasma (or naïve blood) has a mechanism for maintaining an asymmetric distribution. We previously demonstrated that multiple phospholipase activities are involved in the production of LysoGPs (mainly LPC) in plasma. LCAT (which has PLA 2 activity) and lipases (which have PLA 1 activity) are responsible for continuous production of LysoGPs in blood ( 19 ). The present method will elucidate the synthetic pathway of such plasma LysoGPs.
The present study confi rmed that the fatty acid asymmetrical distribution between the sn -1 and the sn -2 of LysoGPs was kept in most tissues, while it was partially lost in plasma and almost completely lost in serum. 1-Acyl-2-LysoGPs were more stable in organic solvents such as chloroform than in aqueous solution (data not shown). Thus, LysoGPs appear to be more stable in a hydrophobic environment than in an aqueous environment. If this is the case, it is reasonable to assume that 1-acyl-2-LysoGPs are In summary, we developed a simple and versatile method to measure 1-acyl-2-LysoGPs and 2-acyl-1-LysoGPs. The method can be used to determine the precise ratio of 1-acyl-2-LysoGPs and the corresponding 2-acyl-1-LysoGPs in various biological samples, and thus, aids in better understanding of the biological signifi cance of 2-acyl-1-LysoGPs .