Analysis of ECs and related compounds in plasma: artifactual isomerization and ex vivo enzymatic generation of 2-MGs.

The analysis of peripheral endocannabinoids (ECs) is a good biomarker of the EC system. Their concentrations, from clinical studies, strongly depend on sample collection and time processing conditions taking place in clinical and laboratory settings. The analysis of 2-monoacylglycerols (MGs) (i.e., 2-arachidonoylglycerol or 2-oleoylglycerol) is a particularly challenging issue because of their ex vivo formation and chemical isomerization that occur after blood sample collection. We provide evidence that their ex vivo formation can be minimized by adding Orlistat, an enzymatic lipase inhibitor, to plasma. Taking into consideration the low cost of Orlistat, we recommend its addition to plasma collecting tubes while maintaining sample cold chain until storage. We have validated a method for the determination of the EC profile of a range of MGs and N-acylethanolamides in plasma that preserves the original isomer ratio of MGs. Nevertheless, the chemical isomerization of 2-MGs can only be avoided by an immediate processing and analysis of samples due to their instability during conservation. We believe that this new methodology can aid in the harmonization of the measurement of ECs and related compounds in clinical samples.

on membrane phosphatidylinositols, and then being converted to 2-AG by the action of either of two isoforms of the same enzyme, the sn -1-diacylglycerol lipases ␣ and ␤ (DAGL ␣ and DAGL ␤ ) ( 15,16 ).
In clinical studies, the determination of ECs and ERCs is limited by methodological issues which particularly concern 2-MGs. Both their chemical isomerization and ex vivo generation are major issues that limit their inclusion as disease/ physiological biomarkers . The aim of the present work is to improve current available methodological approaches for a better understanding of the biological signifi cance of ECs.
A number of publications have been devoted to the analysis of ECs in plasma, and an effort has recently been made to establish reference intervals for fi ve ECs in human plasma ( 12 ). As the chemistry and biology of ECs are better understood, so is the need to improve measurements in human plasma through greater control of factors that introduce variability. At present, such factors still limit the interchangeability of EC plasma concentrations from clinical studies. It has already been established that the chemical properties of ECs, with respect to their stability during analytical procedures (extraction solvents, pH conditions, and evaporation of organic solvents) and their absorption by glassware and plastic materials, are very relevant factors to take into consideration. A comprehensive review has provided a full discussion on these issues ( 13 ) and several analytical methods have already taken these factors into account ( 12,14 ). Discrepancies among laboratories probably originate from preanalytical sample procurement protocols and compound-specifi c factors.
Concerning biological matrices, ECs are unstable in those where enzymes are involved in their synthesis and clearance: the fatty acid amide hydrolase hydrolyzes AEA to arachidonic acid and ethanolamine, and the MG lipase is responsible for the hydrolysis of 2-AG to arachidonic acid and glycerol. Nevertheless, accumulated experience suggests that the incorporation of unspecifi c enzyme inhibitors of amidases, esterases, and proteases, such as PMSF, to sample collection tubes is not justifi ed. However, blood-containing tubes not centrifuged immediately in cold conditions after withdrawal may cause artifactual exaggerated NAE concentrations due to ex vivo release of from erythrocytes or leukocytes ( 12 ). Additionally, it has been reported that ex vivo synthesis of 2-AG for plasma preserved at room temperature and abundant 2-AG/1-AG isomerization is due to sample analysis conditions ( 12,14 ). The two main ECs, AEA and 2-AG, are produced from different biosynthetic pathways. AEA is generated from N-arachidonoyl phosphatidylethanolamines by several possible biosynthetic routes with multiple enzymes implicated: the N-acyl phosphatidylethanolamine-specific phospholipase D, the ␣ , ␤ -hydrolase-4 (ABHD4), the glycerophosphodiesterase-1 (GDE1), a soluble phospholipase A2, an unidentifi ed phospholipase C, and phosphatases ( 15 ). In contrast, the biosynthetic precursors for 2-AG, the sn -1acyl-2-AGs, are mostly produced by phospholipase C ␤ acting

Sample preparation
Freshly extracted blood from human volunteers was collected in 10 ml K2E 18.0 mg (EDTA) BD Vacutainer tubes and centrifuged immediately for 15 min at 2,800 g in a refrigerated centrifuge (4°C). Plasma was then immediately separated from the blood and distributed in aliquots for further processing or stored at Ϫ 80°C. Discarded human plasma batches from the Blood Bank of Hospital del Mar of Barcelona were used for the validation experiments.
Plasma samples were thawed in less than 30 min at room temperature and processed on ice. Aliquots of 0.5 ml were transferred into glass borosilicate tubes, spiked with 25 l of ISTD mix

Human volunteers
Three human male volunteers were recruited for the procurement of blood samples following protocol MUESBIOL/1 (protocol for the collection of biological samples for biomedical research studies). Twenty-fi ve female healthy control volunteers with a BMI of <25 kg/m 2 were recruited for the procurement of blood samples following the TANOBE protocol. Both protocols were approved by the Ethical Committee of Parc de Salut Mar Barcelona (CEIC-PSMAR) and comply with the Declaration of Helsinki. An informed consent was obtained from the human subjects.

Linearity
The linearity of the method was assessed for seven surrogated analytes (SAs) by construction of calibration curves using plasma samples spiked with deuterated analogs of NAEs and MGs. Analysis was performed in quadruplicate for the following SAs: AEA-d4, LEA-d4, PEA-d4, OEA-d4, DHEA-d4, 2-AG-d5, and 2-OG-d5. The ISTDs were 2-AG-d8 and AEA-d8 (ISTD mix 2), which have additional deuterium atoms in their structure. 2-AG-d8 was used as ISTD of 2-MGs, and AEA-d8 as ISTD of NAEs. The regression analyses of the calibration curves were calculated with SPSS 12.0 with a 1/x weighting factor.

Quantifi cation
Experimental LC/MS-MS parameters for the detection of analytes and the deuterated analogs are presented in Tables 1 and 2 . The quantifi cation of the SAs was calculated by interpolation of the response ratios on the calibration curves. The quantifi cation of the authentic analytes was carried out by isotope dilution with the following formula: [EC]ng/ml = (ng ISTD × analyte response)/(ISTD response × RF × ml aliquot volume). The response factor (RF) was calculated as the ratio of the response area of the analyte divided by the response area of its ISTD for a standard solution mix directly injected without extraction into the LC/MS-MS system and in which equal amounts of the analyte 1 or mix 2, diluted up to 1 ml with 0.1 M Am. Ac. buffer (pH 4.0), extracted with 6 ml of TBME, and centrifuged (3,500 rpm, 5 min) at room temperature. The organic phase was transferred to clean tubes, evaporated (40°C, 20 min) under a stream of nitrogen, and extracts were reconstituted in 100 l of a mixture of water:acetonitrile (10:90, v/v) with 0.1% formic acid (v/v) and transferred to HPLC vials. Twenty microliters were injected into the LC/MS-MS system.

LC/MS-MS analysis
An Agilent 6410 triple quadrupole mass spectrometer (Agilent Technologies, Wilmington, DE) equipped with a 1200 series binary pump, a column oven, and a cooled autosampler (4°C) was used. Chromatographic separation was carried out with a Waters C18-CSH column (3.1 × 100 mm, 1.8 m particle size) maintained at 40°C with a mobile phase fl ow rate of 0.4 ml/min. The composition of mobile phase A was 0.1% (v/v) formic acid in water and mobile phase B was 0.1% (v/v) formic acid in acetonitrile. The initial conditions were 40% B. The gradient was fi rst increased linearly to 90% B over 4 min, then increased linearly to 100% B over 5 min and maintained at 100% B for 3 min, to return to initial conditions for a further 4 min with a total run time of 16 min. The ion source was operated in the positive electrospray mode. A desolvation gas temperature of 350°C and a gas fl ow rate of 10 l/min were used. The pressure of the nebulizer was set at 40 psi and the capillary voltage at 4,000 V. The multiple reaction monitoring  Accuracy was calculated as the percentage of difference between the observed concentration and the nominal concentration. The nominal concentration was calculated as the expected concentration on day 1 of the QC sample after the spiking process, taking into account the basal EC and ERC concentrations. A percentage of difference less than 15% for QC-M and QC-H, and less than 20% for QC-L was considered acceptable.
Imprecision was calculated as the standard error deviation of the QC sample replicates. A standard error deviation less than 15% for QC-M and QC-H, and less than 20% for QC-L was considered acceptable.

Recovery and matrix effect
Recovery and matrix effect were evaluated in plasma from six different sources with deuterated analogs as SAs analyzed in triplicate. First, each batch of plasma was divided into two pools; one pool was spiked with ISTD mix 1 and mix 2 and extracted, while the other pool was spiked with ISTD mix 1 and mix 2 after extraction. Second, ISTD mix 1 and mix 2 were also spiked into clean glass tubes, evaporated, and reconstituted. Finally, extracted samples and pure standards were analyzed by LC/MS-MS. Recovery was calculated as the response of the SAs of samples spiked before extraction versus samples spiked after extraction. The matrix effect was calculated as the response of the SAs of samples spiked after extraction versus the pure standards. Additionally, the CV of the ratio of the SAs with the ISTD of the six plasma sources was calculated.

Stability of the analytes on reinjection
The stability of the reconstituted extract solutions on HPLC vials was tested with the reinjection of a batch in which the vials were kept at 4°C for 24 h and another batch with the vials kept at Ϫ 20°C for 10 days.

Stability of the isomers 1 and 2 of MGs in the biological matrix
The stability of the isomers 1 and 2 of MGs to isomerization (or acyl migration) in the biological matrix was assessed by measuring the ratio between the isomers after preservation of the plasma at different times and temperatures. For that, one batch of human plasma from the blood bank was spiked with 2-OG-d5 and 2-AG-d5 and another batch was spiked with 1-OG-d5 and 1-AG-d5. Samples were analyzed by LC/MS-MS and the ratios of the 1 and 2 isomers of both batches were calculated. The spiked standards, dissolved in mobile phase, were also injected directly into the LC/MS-MS system and the and ISTD were present. A deuterated form was not commercially available for some analytes, so a deuterated analog of another NAE or MG with a similar structure was used as ISTD. ISTD mix 1 was used for the quantifi cation of authentic analytes. This fi tfor-purpose approach could be employed due to the fact that the basic structure of the NAEs and the MGs is the same, the only difference being the length of the hydrocarbon chain and the number and position of double bonds. For some analytes the RF was considered 1.0 because the differences in the absolute response were less than 10% ( Table 1 ). We found that the responses of 2-AG-d8 and AEA-d8 were considerably lower than their nondeuterated forms (approximately 10-fold), although, as they were not used in the isotope dilution quantifi cation method, calculations were not affected. The decreased response was probably due to the different position of the deuterium atoms in the structure of the d8 analog (and next to the double bonds) compared with the d4 and d5 analogs ( Fig. 2 ). In our LC/MS-MS conditions, responses of the 1-MG and 2-MG isomers were the same.

Limits of detection and quantifi cation
The mathematical estimates of the limits of detection (LODs) and lower limits of quantifi cation (LLOQs) of the SAs were inferred from the equations of the curves by the following formulas: LOD ng/ml = (SD of the replicates of the lowest concentration on calibrator/slope) × 3 and LLOQ ng/ml = (SD of the replicates of the lowest concentration on calibrator/slope) × 10. Additionally, the LLOQs of the SAs were verifi ed experimentally by a six replicate analyses of plasma spiked with d4 or d5 deuterated forms of NAEs and 2-MGs at the following concentrations: PEA-d4, 0.1 ng/ml; OEA-d4, 0.1 ng/ml; LEA-d4, 0.1 ng/ml; AEA-d4, 0.02 ng/ml; DHEA-d4, 0.02 ng/ml; 2-AG-d5, 0.75 ng/ml; and 2-OG-d5, 2.5 ng/ml. The samples were further spiked with ISTD mix 2, which contained 2-AG-d8 and AEA-d8, and were analyzed by LC/MS-MS. The ratio of the SAs and their ISTDs was calculated. A coeffi cient of variation (CV) of the ratios of less than 20% and a signal to noise ratio greater than three were considered acceptable.
A dilution integrity experiment was carried out for lower sample volumes down to 50 l, with no signifi cant differences in concentration. However, the standard volume of the method was set at 0.5 ml in order to be able to quantify the ECs and ERCs with lower endogenous concentrations.

Accuracy and imprecision
The within-day and between-day accuracy and imprecision of the method were evaluated by the quadruplicate analysis of quality control (QC) samples at three concentration levels  The stability of the isomerization of isomer 2 to isomer 1 during extraction and evaporation steps was assessed by analyzing the ratio (ISO1/ISO2) of plasma spiked with standard solutions of 2-AG-d5 and 2-OG-d5 subjected to liquid-liquid extraction compared with standard solutions dissolved in mobile phase injected directly into the LC/MS-MS system. ISO1, isomer 1; ISO2, isomer 2 .
a Reused clean tubes. b New silanized tubes.
KT109, and GSK 264220A. Stock inhibitor solutions were prepared in ethanol, except D609 that was prepared in water.

Stabilization of MG measures in plasma with Orlistat
Blood extracted from 25 human female volunteers was collected in 10 ml K2E 18.0 mg (EDTA) BD Vacutainer tubes and centrifuged immediately at 2,800 g in a refrigerated centrifuge (4°C). Plasma of each volunteer was separated immediately from the blood and two equal 0.6 ml aliquots were obtained. One aliquot was spiked at 3.35 M with 5 l of Orlistat solution (200 g/ml, ethanol). Both aliquots were stored at Ϫ 80°C until EC analysis with our standard procedure.

Method development
While some solid phase extraction methods were tried, recovery was diffi cult to optimize due to the varying structures of the analyzed compounds (NAEs and MGs). The isomer ratio was calculated. Additionally, in another experiment, the isomerization of endogenous MGs (2/1-AG, 2/1-LG, and 2/1-OG) was assessed in freshly obtained plasma samples from human volunteers preserved 2 h at 4ºC or room temperature .

Inhibition experiments of the ex vivo generation of MGs from plasma
Blood was obtained from human volunteers and processed immediately. The rate of production of MGs from the separated plasma was assessed by EC analysis after incubation for 2 h at RT under agitation. Basal EC levels (time 0) were estimated by immediate EC analysis. A set of potential inhibitors of MG production were tested by spiking the plasma before incubation for 2 h at RT at different concentrations of inhibitors. In several experiments, the following phospholipase and lipase inhibitors were tested: edelfosine, D609, FIPI, Orlistat, RHC 80267, KT172,  Mean ± SD of the recovery and matrix effect of the surrogate analytes in plasma of six different sources and analyzed in triplicate; CV of the ratio of the SAs with the ISTDs of the six plasma sources. authors ( 12,14 ). However, this is most likely due to the longer evaporation time of toluene compared with TBME, which in our experiment was relevant due to the amount of solvent used (6 ml) in the liquid-liquid extraction. The standard solutions were prepared in acetonitrile because methanol and other protic solvents promote the isomerization of isomer 2 into isomer 1 of MG ( 13 ). No differences were observed between extraction tubes that were clean, reused ones, or new silanized tubes.
We found optimal recoveries and peak shapes with the reconstitution of the extract in a mixture of water: acetonitrile (10:90) with 0.1% formic acid. The EC and ERC profi les were separated by reverse phase gradient chromatography in a C18 column ( Fig. 3 ), because with a C8 column complete separation of the MG isomers was not possible. Acetonitrile was used as the organic mobile phase, as we observed that methanol also promoted 2-MG/1-MG isomerization if present in the mobile phase. Formic acid at 0.1% v/v was employed as an additive of the mobile phase to promote the positive ionization of NAEs and MGs. The parent ion adducts selected for fragmentation in the mass spectrometer were in the form of [M+H] + . The product ion m/z 62, which corresponds to ethanolamine, is characteristically generated by fragmentation of NAEs; while a neutral loss of 92 Da , which corresponds liquid-liquid extraction methods provided the best overall recoveries for ECs and ERCs. In order to assess a method to stabilize the original isomeric ratio after the extraction and evaporation steps, 0.5 ml plasma samples were spiked with the deuterated analog isomers 2-AG-d5 and 2-OG-d5, extracted, and analyzed by LC/MS-MS to evaluate the generation of 1-AG-d5 and 1-OG-d5 by chemical isomerization or acyl migration ( Table 3 ). Several liquid-liquid extraction methods were tested: TBME or toluene as extraction solvents, Am. Ac. buffer 0.1 M at pH 4.0, or water as aqueous solvents, and new silanized borosilicate glass tubes (or clean reused tubes) to test for the activity of catalytic silanols. Aqueous solvents were used to reduce viscosity and control plasma pH. Elevated temperatures, presence of serum albumin in the sample, and high pH values have been reported to accelerate 2-AG/1-AG acyl migration ( 13,14,17 ). It is worth noting that commercially available solutions of 2-AG and other 2-MG usually contain 5-10% 1-MG. Therefore, it seems impossible to completely avoid 2-MG/1-MG isomerization. The solvent that best preserved the original deuterated MG isomeric ratio of the sample after extraction was the mixture 6:1 TBME: 0.1 M Am. Ac. buffer (pH 4.0). In our experimental conditions, we observed that TBME preserved the 2-MG/1-MG ratio better than toluene, which was the solvent of choice for other  Blood was collected in EDTA or heparin tubes. The separated plasma was analyzed immediately (time 0) or after incubation for 2 h at room temperature with or without addition of Orlistat to the plasma collection tube. Orlistat was added at 3.5 M to EDTA-plasma and at 15 M to heparin-plasma. Data are presented as mean ± SD of 2/1-MG concentrations (ng/mL) of plasma from one volunteer in an experiment performed in duplicate.
The quantifi cation of ECs, as for all endogenous analytes, is challenging due to the absence of a blank matrix. Some authors have developed strategies of depletion of the analytes by processing the plasma using fi ve cycles of activated charcoal ( 12 ). Other authors have used a surrogated analysis approach ( 18 ) or carried out quantifi cation by isotope dilution ( 19,20 ). It is to be noted that EC-depleted plasma still contains MGs, probably due to its high concentrations ( 12 ). In this work, we have assessed the linearity, LOD, LLOQ, recovery, matrix effect, and MG acyl migration stability of the method using deuterated analog forms as SAs and ISTDs. Our approach is valid because the original unaltered matrix can be used and, theoretically, the deuterated forms have the same properties as the authentic analytes. However, there are a limited number of deuterated analogs of ECs and ERCs, and in order to use this approach different deuterated analog versions for each analyte are necessary. For this reason, in the analysis of samples of clinical studies (inhibition experiments), and in determining the accuracy and imprecision of the method, quantifi cation was carried out with isotope dilution as described in the Materials and Methods.

Method validation
The method was linear for the ECs and ERCs whose quantifi cation was standardized. Results, which include the mathematically derived LODs and LLOQs, are shown in supplementary Table I. The experimentally verifi ed LLOQs of the method are the following: 0.02 ng/ml for AEA, DEA, DGLEA, EPEA, ␣ -LEA, DHEA and POEA; 0.1 ng/ml for LEA; 0.5 ng/ml for OEA and SEA; 0.75 ng/ml for 2-AG; 1 ng/ml for PEA; and 2.5 ng/ml for 2-OG and 2-LG. The LLOQs of PEA, OEA, and SEA were set at a higher concentration than their mathematical LLOQs due to small basal contaminant concentrations found in the solvents and glassware as reported by other authors ( 12,21 ). No signifi cant carry over was detected. No differences in the concentration values were found after reinjecting vials kept at 4°C for 24 h. Vials kept at Ϫ 20°C were stable for all analytes except SEA. Recoveries were high (>80%) for all the analytes and matrix effect was substantial (40%) in some analytes such as 2-OG-d5. Minimal differences in the matrix effect of the six plasma sources were, however, observed due to being compensated by the use of deuterated analogs with similar ISTD structure ( Table 4 ). Within-run and between-run accuracy and imprecision values of NAEs and MGs are presented in Table 5 and fi t current standard requirements for analytical method validation. With respect to the MGs, we found decreases in the concentration of the separate 2-MG isomers from day 2 of the validation protocol due to acyl migration during conservation of the plasma. This is explained in the section on stability of the isomeric ratio of MGs. The method is deemed fi t for the determination of the EC and ERC profi le in human plasma samples.

Inhibition experiments of the ex vivo generation of MGs from plasma
Fanelli et al. ( 12 ) reported the generation of 2-AG in plasma in the absence of blood cells, with increases in 2/1-AG for plasma preserved for 4 h at 4°C or room temperature. In the course of our MG stability experiments, we observed to glycerol, is common to the fragmentation of MGs. The specifi c MRM transitions are listed in Tables 1 and 2 . The method is selective and specifi c for each analyte with no cross contamination between MRM channels and, in most cases, a single chromatographic peak corresponding to the endogenous analyte was found throughout the acquisition time. The variation in retention time of the individual analytes in a typical batch analysis was less than 0.5%. The retention time and product ion spectra of the endogenous analytes in the matrix matched the ones of the authentic standards.  ( 7,(25)(26)(27). Therefore, in terms of preventing 2-MG ex vivo formation, results obtained from plasma samples spiked with Orlistat are in agreement with this inhibitory enzymatic activity. The biosynthetic origin of 2-AG, however, is presumably not related to fat digestion but to phospholipids. Arachidonate DAGs, the precursors of 2-AG, are originated by the hydrolysis of membrane phosphoinositides and they are converted to 2-AG by the action of two sn -1 selective DAGLs, DAGL ␣ and DAGL ␤ ( 16 ). Further, a direct dietary origin of 2-AG seems unlikely because arachidonic acid, an essential fatty acid and backbone of 2-AG structure, is present at low amounts in the diet, and is mainly obtained through metabolism of triacylglycerols that contain acyl-linoleoyl in their structure . The linoleic acid released is then elongated and unsaturated to form arachidonic acid through the omega-6 pathway. Additionally, we investigated the effect of the blood collection tube on MG generation. We found that MG concentrations were higher in heparin-plasma than in EDTA-plasma. The differences were maintained either for samples analyzed immediately or after incubation at room approximately 5-fold increases in 2/1-LG and 2/1-OG concentrations in plasma preserved for just 2 h at room temperature compared with plasma preserved at 4°C. In regard to 2/1-AG, we observed differences in concentration when plasma was analyzed immediately after blood extraction versus some time afterwards (2 h), or after a freezing/thawing step even though cold chain (4°C) was maintained. We therefore investigated, in controlled experiments, several phospholipase and lipase inhibitors for their capacity to inhibit ex vivo MG production in plasma. We found no inhibition activity for the phospholipase inhibitors D609, FIPI, and edelfosine (IC 50 >30 M, IC 50 >125 nM, and IC 50 >50 M, respectively). On the other hand, we observed inhibition of the ex vivo generation of 2-AG, 2-LG, and 2-OG in plasma spiked with Orlistat, a gastric and pancreatic lipase inhibitor ( 22 ) and a potent nonspecifi c inhibitor of DAGL ␣ and DAGL ␤ ( 23 ). It is to be noted that the artifactual generation of MGs persisted even after immediate sample centrifugation that essentially eliminates all blood cells, which means that this MG buildup should be related to an enzymatic plasma activity. Because Orlistat is a general lipase inhibitor, we also tested the inhibition activity of RHC 80267, another nonspecifi c inhibitor of DAGL ␣ ( 23 ), KT172, and KT109, potent selective DAGL ␤ inhibitors ( 24 ). None of these compounds showed inhibition activity (IC 50 >30 M, IC 50 >15 M, and IC 50 >15 M, respectively), which means that the ex vivo MG production in plasma is a mechanism independent A pool of plasma was spiked separately with the deuterated analogs of the isomer 1 (ISO1) and isomer 2 (ISO2) of AG and OG. Aliquots were distributed in cryotubes and chemical stability of the isomer ratio (ISO1/ISO2 and ISO2/ISO1) was assessed upon conservation at time 0, 30 min at room temperature, 20 days at Ϫ 20°C, or 20 days at Ϫ 80°C. Data are presented as mean ± SD of replicate analysis. the 25 human female samples ( Table 7 ), with the exception of EPEA that, due to its low levels, could only be quantifi ed in 14 samples. In summary, data show that Orlistat addition as part of the sample collection protocol can be a tool to stabilize MG concentrations in plasma, and this can aid in the harmonization of EC and ERC measurements in clinical samples.

Stability of the isomeric ratio of MGs
The results of the stability experiment of MGs in plasma show that the chemical isomerization of isomer 2 to isomer 1 decreases with decreases in the preservation temperature of the plasma before analysis. However, isomerization is still observed, even if samples are stored at Ϫ 80°C ( Table 8 ). Furthermore, it has been reported that isomerization is also dependent on the amount of serum albumin present in the sample ( 17 ). It is, therefore, possible that even when samples are subjected to the same storage conditions, they could still have different chemical isomerization rates. Additionally, chemical isomerization takes place in plasma preserved for a very short time at room temperature.
We also investigated the endogenous origin of MG isomers in fresh plasma samples ( Table 9 ). As has been previously suggested ( 14 ), our data support the hypothesis that 1-AG does not have an endogenous origin and is the result of chemical isomerization during sample storage and processing, because in fresh plasma samples we found that 1-AG was present at the same isomeric ratio as the pure standard mixture. On the other hand, we observed that 1-LG and 1-OG were present at substantial concentrations. 1-MG originates from the in vivo isomerization of 2-MG during digestion and absorption. It has been estimated that approximately 25% of 2-MG is isomerized to the 1-MG form. However, 2-MG is the predominant form in which MGs are absorbed and resynthesized to triacylglycerols, while 1-MGs are eventually hydrolyzed by pancreatic lipase to free fatty acids and glycerol (25)(26)(27)30 ).
In summary, only EC analysis performed with fresh samples is able to quantify the original isomeric ratio of the sample. Studies that report MG concentrations should specify whether the concentration data are from the separate or combined 1 and 2 isomers. Due to the instability of isomerization during conservation, and the fact that the 1-MG isomer originates either in vivo or ex vivo from the 2-MG isomer, studies that report the concentration of the two isomers together may still provide meaningful data for the interpretation of its biological signifi cance in a fi t-for-purpose approach. Alternatively, clinical samples may be spiked with deuterated analogs of known isomer ratios before conservation in order to correct the concentration data. CONCLUSIONS MG analysis is a challenging issue; to our knowledge, this is the fi rst time that an enzymatic activity inhibited by the lipase inhibitor Orlistat and able to generate MGs in plasma in the absence of cells has been reported. Our fi ndings suggest that, as happens with EC brain concentrations ( 31 ), temperature. Orlistat was able to inhibit MG production in plasma originated from both kind of tubes, but due to the higher MG buildup in heparin-plasma, a higher concentration of Orlistat was needed to achieve full inhibition ( Table 6 ). Further, because heparin-plasma is commonly used for the assay of lipoprotein lipase due to its affi nity for heparin ( 28 ), we also tested the inhibition activity of GSK 264220A, an endothelial lipase and a lipoprotein lipase inhibitor ( 29 ), but we found no inhibition (IC 50 >15 M) on EDTA-plasma or heparin-plasma. The reason for the lower MG concentrations on EDTA-plasma is probably due to the chelate effect of EDTA on the cofactors needed for MG biosynthesis. We recommend, therefore, the use of EDTA blood tubes for collection in addition to Orlistat. In a second set of experiments, EDTA-plasma of three human volunteers was used for the calculation of the IC 50 of Orlistat for the ex vivo generation of 2-AG, 2-LG, and 2-OG. An inhibition model was obtained and the data is graphically presented in Fig. 4 . The IC 50 of Orlistat with the mean, its 95% confi dence interval, and the coeffi cient of determination (

Stabilization of MG measures in plasma spiked with Orlistat
The ex vivo generation of MG in plasma can be prevented by Orlistat. Because our MG assay was done at room temperature, we tested, in controlled conditions, whether Orlistat addition to the plasma storage tube had any effect on MG measures when a typical clinical sample collection protocol was followed. For that, 25 female blood samples were collected and processed in a matter of weeks, maintaining the cold chain until they were fi nally stored at Ϫ 80°C. EC analysis took place several weeks after all samples had been collected and was done with our standard sample preparation procedure. The results are presented in Table 7 and they show that Orlistat addition during the sample collection protocol leads to a signifi cant reduction of all MG measures (36-59%, P < 0.001). The NAEs, which are the other measures of our EC and ERC analysis, were not affected by the addition of Orlistat. We think that differences may have arisen due to the enzymatic activity that took in the freezing/thawing and processing steps. All the NAEs and MGs described in the method could be quantifi ed in The stability of the isomer 1/isomer 2 (ISO1/ISO2) ratio of endogenous 2/1-AG, 2/1-LG, and 2/1-OG was assessed in plasma of three different volunteers which was kept 2 h at 4°C or room temperature after extraction from the volunteer. Data are presented as mean ± SD of triplicate analyses. peripheral EC concentrations from clinical studies or animal models greatly depend on sample collection and sample time processing conditions that take place in the clinical and laboratory settings due to the natural presence of enzymatic activity in plasma. Because of the instability of EC concentrations in blood, EC studies need to follow strict harmonized sample collection and processing protocols in order to avoid artifi cial differences between samples. Finally, the collection of plasma samples with Orlistat may be a useful tool in the determination of real endogenous 2-MG concentrations. In addition to immediate centrifugation in refrigerated conditions and separation of plasma from blood to avoid the release of NAE from blood cells, we also recommend the addition of Orlistat to plasma collecting tubes and maintaining the cold chain until storage and processing. Orlistat is inexpensive, and thus may be a cost effective measure to aid in the harmonization of EC and ERC measurements in clinical research. Data suggest that the ex vivo generation of MG in plasma is a mechanism independent of DAGL, because besides the general lipase inhibitor Orlistat, other specifi c or unspecifi c DAGL inhibitors do not inhibit MG generation, and neither is the result of endothelial lipase or lipoprotein lipase activity. The full characterization of this enzymatic activity goes beyond the scope of this work, but due to the importance of the EC 2-AG as a biomarker, the understanding of this apparent alternative biosynthetic pathway of 2-AG, probably linked to lipid metabolism, would contribute to a better comprehension of the signifi cance of its blood concentrations .
We have validated a method for the determination of a range of MGs and NAEs in plasma. The developed method is able to preserve the original isomeric ratio of MGs. We have found that the chemical isomerization of MGs can only be avoided by immediate processing (at cold temperature and acid pH) and analysis of samples. The report of MGs as the sum of both isomers may be considered. Alternatively, appropriate isomerization controls can be used during sample collection and conservation in order to correct concentrations. Data suggest that isomer 1 of AG is not an endogenous compound, and most probably is the result of chemical isomerization during storage and sample processing. On the other hand, isomer 1 of OG and isomer 1 of LG are likely to be endogenous compounds that result from in vivo isomerization that takes places during digestion, and their concentrations can be detected in plasma.