Measurement of lysophospholipid acyltransferase activities using substrate competition

Abstract

Lysophospholipid acyltransferases (LPATs) incorporate fatty acyl chains into phospholipids via a CoA-dependent mechanism and are important in remodeling phospholipids to generate the molecular species of phospholipids found in cells. These enzymes use one lysophospholipid and one acyl-CoA ester as substrates. Traditional enzyme activity assays engage a single substrate pair, whereas in vivo multiple molecular species exist. We describe here an alternative biochemical assay that provides a mixture of substrates presented to the microsomal extracts. Microsomal preparations from RAW 264.7 cells were used to compare traditional LPAT assays with data obtained using a dual substrate choice assay using six different lysophospholipids and eight different acyl-CoA esters. The complex mixture of newly synthesized phospholipid products was analyzed using LC-MS/MS. Both types of assays provided similar results, but the dual choice assay provided information about multiple fatty acyl chain incorporation into various phospholipid classes in a single reaction. Engineered suppression of LPCAT3 activity in RAW 264.7 cells was easily detected by the dual choice method. These findings demonstrate that this assay is both specific and sensitive and that it provides much richer biochemical detail than traditional assays.

Two distinctive branches of anabolic pathways dictate the quantity and diversity of phospholipids synthesized across multiple phyla of organisms. One branch, known as the de novo route, relies on the specificity of glycerophosphate- and acylglycerophosphate-acyltransferases to dictate the fatty acyl composition of downstream phospholipid end products (Kennedy pathway) (1). A second branch remodels products of the first branch, usually via the sequential action of a phospholipase and a lysophospholipid acyltransferase (LPAT), producing a different repertoire of phospholipid molecular species (Lands cycle) (2). Data from yeast clearly indicated that the remodeling pathway can operate independently of the de novo pathway when a significant source of extracellular lysophospholipids is provided (3, 4).

Although the initial enzymatic characterization of the remodeling pathway occurred nearly 50 years ago, little progress was made to identify the genes that encoded these enzymes until various complete genomes were sequenced and a collection of candidate genes and cDNAs were empirically tested for their enzymatic functions. Since 2005, several groups have performed extensive analysis of the function of these enzymes and discovered that LPATs belong to two different protein families, represented by the lysophosphatidic acid acyltransferases (LPAATs) and the membrane bound O-acyltransferases (MBOATs) (5–16). These families both contain membrane-bound proteins that have not been purified from tissues (17), but their encoding cDNAs have been cloned and expressed in a variety of host cells including yeast (10, 13, 18).

Enzyme activity assays have been a traditional biochemical means by which to assess the presence of a particular protein in the biological extract and provide insight into the overall performance of this enzyme in terms of synthetic capacity and substrate specificity. LPAT activity assays are often performed by exposing a tissue extract, containing the enzyme of interest, to a pair of pure substrates (one lysophospholipid and one acyl-CoA ester) and measuring the reaction rate parameters for substrate conversion to product. The results of this experiment are then compared with results from an identical set of experiments, but with a different substrate. Eventually, a picture emerges of preferred substrates. Thus, many substrates need to be individually studied, and a concern with this approach is that any competition between substrates for the enzyme is not revealed.

We present here an alternative biochemical assay that provides a mixture of substrates to the microsomal extract. Phospholipid products of unique masses are identified and quantitated using LC/MS/MS. This strategy of a substrate choice enzymatic assay is particularly useful for study of the LPATs because these enzymes use two substrates to make the final phospholipid product. Therefore, with two substrates this greatly expands the total number of analyses that need to be carried out by this single substrate enzyme activity strategy if a complete picture of enzymatic activity is desired. This substrate choice enzyme activity assay presents microsomal enzymes with a mixture of different substrates in the same reaction, with the profile of product abundance indicating substrate preference. Although exact information may differ somewhat between these two assay approaches, this choice assay does reveal potential for product formation when specific substrates are added to the tissue extract. This substrate choice assay is easily carried out due to advances in electrospray MS/MS. In this report, we describe a dual substrate choice assay using microsomal extracts containing LPATs expressed in mammalian cells to test the activity of these proteins in a single experimental model and compare the results with the more traditional means of determining substrate specificity and enzymatic rate.

MATERIALS AND METHODS

Materials

Fatty acyl-CoA esters and phospholipids were from Avanti Polar Lipids (Alabaster, AL): tetradecanoyl (myristoyl; 14:0-CoA), hexadecanoyl (palmitoyl; 16:0-CoA), octadecanoyl (stearoyl; 18:0-CoA), (9Z)-octadecenoyl (oleoyl; 18:1-CoA), (9Z,12Z)-octadecadienoyl (linoleoyl; 18:2-CoA), (5Z,8Z,11Z,14Z)-eicosatetraenoyl (arachidonoyl; 20:4-CoA), (5Z,8Z,11Z,14Z,17Z)-eicosapentaenoyl (20:5-CoA), (4Z,7Z,10Z,13Z,16Z,19Z)-docosahexaenoyl (22:6-CoA), 1-(10Z)-heptadecenoyl-2-hydroxy-lysophosphatidic acid (17:1-LPA), 1-(10Z)-heptadecenoyl-2-hydroxy-lysophosphatidylcholine (17:1-LPC), 1-(10Z)-heptadecenoyl-2-hydroxy-lysophosphatidylethanolamine (17:1-LPE), 1-(10Z)-heptadecenoyl-2-hydroxy-lysophosphatidylglycerol (17:1-LPG), 1-(10Z)-heptadecenoyl-2-hydroxy-lysophosphatidylinositol (17:1-LPI), 1-(10Z)-heptadecenoyl-2-hydroxy-lysophosphatidylserine (17:1-LPS), 1-(9Z)-octadecenoyl-2-hydroxy-lysophosphatidic acid (18:1-LPA), 1-(9Z)-octadecenoyl-2-hydroxy-lysophosphatidycholine (18:1-LPC), 1-(9Z)-octadecenoyl-2-hydroxy-lysophosphatidyserine (18:1-LPS), (2R)-[²H₃₁]3-hexadecanoyl-2-(9Z)-octadecenoyl-PA ([²H₃₁]POPA), (2R)-[²H₃₁]3-hexadecanoyl-2-(9Z)-octadecenoyl-PC ([²H₃₁]POPC), (2R)-[²H₃₁]3-hexadecanoyl-2-(9Z)-octadecenoyl-PE ([²H₃₁]POPE), (2R)-[²H₃₁]3-hexadecanoyl-2-(9Z)-octadecenoyl-PG ([²H₃₁]POPG), (2R)-[²H₃₁]3-hexadecanoyl-2-(9Z)-octadecenoyl-PI ([²H₃₁]POPI), (2R)-[²H₃₁]3-hexadecanoyl-2-(9Z)-octadecenoyl-PS ([²H₃₁]POPS), 1-heptadecanoyl-2-eicosatetraenoyl-sn-glycero-3-phosphate (17:0/20:4-PA), 1-heptadecanoyl-2-(5Z,8Z,11Z,14Z)-eicosatetraenoyl-phosphatidylcholine (17:0/20:4-PC), 1-heptadecanoyl-2-(5Z,8Z,11Z,14Z)-eicosatetraenoyl-phosphatidylethanolamine (17:0/20:4-PE), 1-heptadecanoyl-2-(5Z,8Z,11Z,14Z)-eicosatetraenoyl-phosphatidylglycerol (17:0/20:4-PG), 1-heptadecanoyl-2-(5Z,8Z,11Z,14Z)-eicosatetraenoyl-phosphatidylinositol (17:0/20:4-PI), and 1-heptadecanoyl-2-(5Z,8Z,11Z,14Z)-eicosatetraenoyl-phosphatidylserine (17:0/20:4-PS). DMEM was purchased from Corning Cellgro (Manassas, VA). RAW 264.7 cells were obtained from ATCC (Manassas, VA). EDTA-free protease inhibitor cocktail was purchased from Roche (Madison, WI). All other chemicals and solvents were purchased through Fisher Scientific.

Cell culture and microsome preparation

RAW 264.7 and HEK293T cells were cultured in DMEM (with 4.5 g/l glucose and 100 µM sodium pyruvate) supplemented with 10% heat-inactivated FBS. The cells were grown in humidified air with 5% CO₂ at 37°C. For microsomal preparations, cells were pelleted at 300 g, resuspended in homogenization buffer [50 mM Tris-HCl, pH 7.4, 250 mM sucrose, 1 mM EDTA, 20% (w/v) glycerol, and protease inhibitor cocktail], and lysed using a Sonics Vibra-Cell probe sonicator (Newtown, CT). Whole cells and cellular debris were pelleted at 12,000 g for 20 min at 4°C. The supernatant was transferred to an ultracentrifuge tube and centrifuged at 100,000 g for 60 min at 4°C. The microsomal pellet was resuspended in assay buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA), protein amount was determined using the bicinchoninic acid assay (Thermo Scientific, Rockford, IL), and microsomes were stored at −20°C until used.

Single choice LPAT assay

Stock solutions were made as follows: 60 µM arachidonoyl-CoA in 100% methanol; 200 µM lysophospholipids in assay buffer (prepared and sonicated immediately prior to the assay); 1.25 mM fatty acid-free BSA (BSA) in water; and internal standard mixture of 10 ng/µl each of [²H₃₁]16:0/18:1-PA, [²H₃₁]16:0/18:1-PC, [²H₃₁]16:0/18:1-PE, [²H₃₁]16:0/18:1-PG, [²H₃₁]16:0/18:1-PI, and [²H₃₁]16:0/18:1-PS in methanol. The final concentrations of the reaction components were 10 µg total protein from microsomes; 3 µM LPC, LPE, or LPS; 3 µM arachidonoyl-CoA; and 12.5 µM BSA, in assay buffer to a total volume of 200 µl. The CoA ester solution was made in methanol, leading to a concentration of 5% in the final reaction. When compared with reactions with no methanol, this 5% level resulted in only minor differences found in 2 of the 48 products. The acyltransferase assay was performed at 37°C for 10 min. The reaction was stopped with 750 µl of methanol-chloroform (2:1, v/v), internal standard mixture (2.5 µl) was added, and products were extracted by the Bligh and Dyer method (19). After the samples were dried under a stream of nitrogen, they were resuspended in 100 µl of 75% solvent A (isopropanol-hexanes 4:3, v/v) and 25% solvent B (isopropanol-hexanes-water 4:3:0.7, v/v/v, containing 5 mM ammonium acetate). Samples were analyzed by LC/MS/MS as described subsequently.

Dual substrate choice assay

Acyltransferase activity was tested in microsomal preparations as described for the single substrate choice except for the following alterations. Stock solutions of reaction materials were made as follows: 6 µM, 20 µM, or 60 µM equimolar mixture of eight acyl-CoAs (14:0, 16:0, 18:0, 18:1, 18:2, 20:4, 20:5, and 22:6) in methanol; 20 µM or 200 µM equimolar mixture of six lysophospholipids (LPA, LPC, LPE, LPG, LPI, and LPS) in assay buffer. Final lysophospholipid concentrations used were 0.3 µM, 1 µM, 3 µM, or 10 µM, as indicated, with final concentrations of 50 ng/µl total protein from microsomes, 3 µM of each fatty acyl-CoA ester, and 12.5 µM BSA, in a total volume of 200 µl. To assay different microsomal protein amounts, the assay was performed at final concentrations of 3 µM lysophospholipid mix, 3 µM fatty acyl-CoA ester mix, and 12.5 µM BSA, while using microsomal protein concentrations from 5 ng/µl, 15 ng/µl, and 50 ng/µl. To assay different fatty acyl-CoA ester concentrations, we used 50 ng/µl total protein from microsomes, 3 µM of each lysophospholipid, and 12.5 µM BSA, while using fatty acyl-CoA ester concentrations 0.3 µM, 1 µM, or 3 µM each. The final volume of methanol was kept constant. To assay different time points, the components were at final concentrations of 50 ng/µl total protein from microsomes, 3 µM of each lysophospholipid, 3 µM of each fatty acyl-CoA ester, and 12.5 µM BSA. The assay was performed at 37°C for 0 min, 1 min, 3 min, 10 min, or 30 min. Samples were analyzed by LC/MS/MS as described subsequently.

LC/MS

For normal-phase separation, samples were injected onto an Ascentis-Si HPLC column (150 × 2.1 mm, 5 µm; Supelco) at a flow rate of 0.2 ml/min at 25% solvent B. Solvent B was maintained at 25% for 5 min, increased to 60% over 10 min, and then to 95% over 5 min. The system was held at 95% B for 20 min prior to reequilibration at 25% for 14 min. For reversed-phase separation, solvent C was methanol-acetonitrile-water 60:20:20 (v/v/v), containing 2 mM ammonium acetate, and solvent D was methanol containing 2 mM ammonium acetate. The samples were injected onto an Ascentis-C18 HPLC column (150 × 2.1 mm, 5 µm; Supelco) at a flow rate of 0.2 ml/min at 75% solvent D. Solvent D was maintained at 75% for 1 min and increased to 98% over 5 min. The system was held at 98% D for 20 min prior to reequilibration at 75% for 10 min. Phospholipid products of the LPAT assay were measured using an API3200 triple quadrupole mass spectrometer (AB Sciex, Framingham, MA) in negative ion mode using multiple reaction monitoring (MRM) of the m/z transitions shown in supplementary Table I. Quantitation was performed using AB Sciex MultiQuant software and using 17:0/20:4-PA, 17:0/20:4-PC, 17:0/20:4-PE, 17:0/20:4-PG, 17:0/20:4-PI, and 17:0/20:4-PS reference standards for dilution curve analysis, as described previously (20). Quantitative analysis of the reference standards was performed in the absence or presence of microsomes, and quantitative values did not differ significantly between the two conditions (not shown).

Knockdown of LPCAT3 gene expression in RAW 264.7 cells

HEK293T cells were transfected using TurboFect (Thermo Fisher) with lentiviral packaging vectors and either a vector coding for a nontargeting shRNA (Sigma SHC002) or pLKO.1-puro vector coding for shRNA targeted for the mouse LPCAT3/MBOAT5 sequence 5′-CCGGGCCAATCTACTACGATTGTATCTCGAGATACAATCGTAGTAGATTGGCTTTTTG-3′. Medium containing the lentivirus was collected from the HEK293T cell cultures, and 1,5-dimethyl-1,5-diazaundecamethylene polymethobromide (polybrene) was added at a concentration of 8 mg/ml. The receiving RAW 264.7 cells were treated with 8 mg/ml polybrene for 30 min at 37°C. The medium was then replaced with the lentivirus-containing medium and incubated overnight. The medium was replaced with standard medium containing 2 µg/ml puromycin. Cells were kept under constant selection of 2 µg/ml puromycin for at least three passages. Microsomes where prepared as described previously. Total RNA was extracted from nontargeted and LPCAT3/MBOAT5-targeted cells using Life Technologies (Grand Island, NY) Trizol reagent. Total RNA was converted to cDNA with the BioRad iScript reverse transcription supermix. Quantitative PCR (qPCR) was performed using the BioRad iTaq SYBR Green supermix. Primers for mouse GAPDH and LPCAT3 were purchased from Integrative DNA Technologies. Samples were analyzed with a BioRad iQ5 thermal cycler according to the instructions provided with the iTaq SYBR Green supermix.

RESULTS

The characterization of acyltransferase activities has been performed traditionally using a single lysophospholipid and a single fatty acyl-CoA ester, one of which would be radiolabeled to allow phospholipid product detection and quantitation (21). Recent work has taken advantage of MS to analyze reaction products using a choice among multiple fatty acyl-CoA esters and lysophospholipids to investigate the substrate specificity of individual human isoforms of the MBOAT family (13, 22). The goal of the present work was to validate the dual choice assay by comparing it directly to the traditional assays and to further develop this method as a tool to analyze the mixture of acyltransferase activities present on the microsomes of the murine macrophage-like cell RAW 264.7. The fatty acyl-CoA esters chosen contained the acyl chains more commonly found in cellular membranes. They included 20:4, 20:5, and 22:6, which are of biological interest because the corresponding fatty acids are precursors of bioactive lipids. All the lysophospholipids contained the same sn-1 acyl chain, 17:1, which reduced the background signal from endogenous phospholipids present in the microsomal preparations. Two common molecular species (16:0/18:1 and 18:1/18:1) were also measured for each phospholipid class to check for possible changes in endogenous microsomal phospholipids during the incubation period.

The effectiveness of normal- and reverse-phase LC was evaluated to resolve both endogenous and newly synthesized phospholipids, which were then detected with a triple quadrupole mass spectrometer operated in the MRM mode. Typical reverse-phase and normal-phase chromatography profiles for the products of the assay are presented in supplementary Fig. I. In subsequent experiments, normal-phase separation was used because it provided separation of the individual phospholipid classes. In those cases where selected mass transitions overlapped (PE/PI and PA/PS), MS provided the identification and quantitation of products. Most of the m/z transitions were specific for each phospholipid measured, with two exceptions: m/z 747→281 ([²H₃₁]16:0/18:1-PE or 16:0/18:1-PG) and m/z 774→267 (17:1/22:6-PE or 17:1/18:1-PS). In these cases, quantitation was still possible because the corresponding phospholipids were resolved by normal-phase chromatography.

The deprotonated [M-H]⁻ precursor ions were selected as the target ions to quantitate all classes except for PC, in which case the acetate adduct [M+CH₃COO]⁻ ions were used. Upon collision-induced decomposition, all precursor ions yielded abundant product ions corresponding to both the sn-1 and sn-2 carboxylate anions. In preliminary experiments using standards, the sn-1 anion was found to be more abundant than the sn-2 anion for PA and PS, whereas the sn-2 anion was more abundant in the cases of PC, PE, PI, and PG under the conditions used. The transitions to the sn-1 anion (m/z 267) were used for PC and PE because endogenous plasmenyl (18:0p-containing) PC and PE species are isobaric with the 17:1-containing products newly synthesized. Endogenous plasmenyl phospholipids do not generate an m/z 267 production. Therefore, the transitions to this sn-1 acyl anion yielded robust signals that were not obscured by high background.

Quantitation of each of the phospholipid reaction products was performed by calculating the ratio of the integrated area of the phospholipid product peak (analyte) to that of the corresponding internal standard. Then, reference standard dilution curves were used to calculate the amount of phospholipid product made, as previously described (20). The calibration curves for the 17:0/20:4 phospholipid species of each of the six classes are presented in supplementary Fig. II.

Determination of enzymatic parameters of the dual substrate choice assay in microsomes from RAW 264.7 cells

Previous assays to test LPAT activity have used high levels of substrates with long incubation times to assay LPAT activity. In testing the dual substrate choice assay, reaction conditions were optimized to ensure that they were within the linear range, to better appreciate the substrate preferences of the LPATs present in the sample.

Four major variables that dictate product formation were analyzed, namely time, lysophospholipid concentration, fatty acyl-CoA concentration, and total microsomal protein. The standard reaction conditions used in testing each variable were 3 µM of each lysophospholipid and fatty acyl-CoA, 10 µg of microsomal protein, and a 10 min incubation. To illustrate the reaction kinetics for the dual substrate choice assay, the formation of 17:1/20:4-PC is presented in Fig. 1. The product formation was clearly dependent on reaction time, concentrations of either substrate, and amount of protein. The standard reaction conditions chosen were within the linear range of product formation and yielded robust signals for many of the possible reaction products. The standard reaction conditions used lysophospholipid concentrations that are significantly lower than concentrations used in previous biochemical assays and below the reported critical micellar concentration (23).

Fig. 1.

Reaction conditions of the dual substrate choice acyltransferase assay. Microsomes from RAW 264.7 cells were used to assay for LPAT activity using a mix of six lysophospholipids and eight acyl-CoA esters as described in the Materials and Methods. Four variables were changed as shown in the different panels: time (A), lysophospholipid concentration (B), fatty acyl-CoA concentration (C), and microsomal protein amount (D). Only the results for incorporation of arachidonoyl chain into PC are shown, but all 48 possible products were quantitated using LC/MS/MS. Three independent microsomal preparations were assayed in duplicate, and data shown are average ± SEM.

The absolute amount of microsomal protein present in the assay was used to normalize the amount of each phospholipid product newly synthesized, as is commonly done in enzymatic assays. The levels of the two endogenous molecular species measured for each of the phospholipid classes (16:0/18:1- and 18:1/18:1-phospholipids) showed excellent correlation with the amount of microsomal protein (supplementary Fig. III) and could be used as an alternative normalization factor. A small increase in endogenous 16:0/18:1-PC was observed, but when statistical analysis was performed between time 0 and time 3 min, a P value of 0.334 was found. When comparison was made for the formation of 17:1/20:4-PC at time 0 and time 3 min, a P value of 0.0006 was found, thus suggesting that the slight increase in 16:0/18:1-PC at 3 min was probably due to the error of microsomal addition to the samples.

The products of the dual substrate choice assay reflect the presence of many LPATs in RAW 264.7 microsomes

After determining the reaction conditions for the LPATs present in microsomes from RAW 264.7 cells, this assay was used to further characterize the enzymatic activities in these microsomes. Figure 2 shows the 48 products analyzed after a 30 min incubation. The data presented highlight different fatty acyl substituents at the phospholipid sn-2 position, supporting the presence of several LPATs that specifically incorporated acyl chain into each phospholipid class.

Fig. 2.

Acyl chain preference of RAW 264.7 microsomal acyltransferases. Microsomes from RAW 264.7 cells were incubated with lysophospholipid and acyl-CoA mixtures for 30 min. The 48 possible products of the reaction were quantitated, and data are presented to highlight the acyl chain preference within each of six phospholipid classes. Three independent microsomal preparations were assayed in duplicate, and data shown are average ± SEM.

In the PA class, most acyl chains were incorporated at high levels, with the exception of 20:5. The preferred acyl chains were 16:0 and 18:2, followed by 14:0 and 18:1. These results are consistent with the possible expression and activity of LPAAT1, LPAAT2, and LPAAT3 (6, 8). In PC, the preferred acyl chain was 20:4, and there was also robust incorporation of 18:2 and 20:5 chains, consistent with the activity of LPCAT3/MBOAT5 (6, 8, 13). The incorporation of saturated and monounsaturated chains suggested the activities of LPCAT1, LPCAT2, and/or LPCAT4 (6, 8). The incorporation of 18:2 and 20:4 chains into PE was characteristic of the activity of LPEAT2 (6, 8). LPEAT1 has been reported to incorporate 18:1-CoA into LPE and LPS (6, 8, 13, 24). However, both of these PE species were produced at a minimal level, suggesting negligible activity of this enzyme in RAW 264.7 cells. The PS products generated were possibly a result of the LPSAT activity of MBOAT5. The 20:4 and 20:5 incorporation into PI was consistent with the activity of MBOAT7. Finally, there was not a robust production of any of the molecular species measured for PG, which may be due to low levels of LPGAT present in the RAW 264.7 cells.

The incorporation of 18:0 chains into PE, PS, PI, and PC was surprisingly high (Fig. 2), providing evidence for additional acyltransferase enzymes because an acyltransferase with this activity has not yet been characterized. These phospholipid products could arise from other enzymes not previously described that could exhibit LPEAT, LPIAT, LPSAT, and LPCAT activities. It is also uncertain which enzyme incorporated 22:6 chains into PC and PI. Low activity with 22:6-CoA esters has been reported for incorporation into LPA by LPAAT3, and LPC by LPCATs 1–3 (8). The data demonstrate that many acyltransferase activities can be revealed by the dual substrate choice assay, which greatly enhances the chances of identifying previously unknown and novel enzymes. To confirm that these products were not an aberration of the dual choice assay because of the substrate mixtures or the use of nonendogenous, odd-chain lysophospholipids, single substrate assays were performed using more common lysophospholipids (supplementary Fig. IV). These data confirm that the use of 17:1-lysophospholipid in a dual choice setting recapitulates data obtained in the traditional enzymatic assays with natural lysophospholipid species.

An alternative presentation of the data makes it easier to illustrate phospholipid class specificity using each of the acyl chains (Fig. 3). Even though this experiment used a 10 min incubation, the apparent substrate preference of the combined LPAT activities in the microsomes remained very similar to that presented in Fig. 2, and the incorporation of 20:4 and 20:5 had very similar distribution patterns in terms of relative phospholipid class formation, suggesting that the same LPATs (probably LPCAT3/MBOAT5 and MBOAT7) were responsible for the incorporation of these acyl chains into phospholipids in these cells. The distribution pattern of newly synthesized phospholipids incorporating 14:0, 16:0, 18:1, 18:2, and 22:6 were also quite similar. The incorporation of 18:0 displayed a unique pattern as compared with the other seven acyl chains, highlighting the possible involvement of an unknown acyltransferase activity.

Fig. 3.

Polar head group preference of RAW 264.7 microsomal acyltransferases. Microsomes from RAW 264.7 cells were incubated with lysophospholipid and acyl-CoA mixtures for 10 min. The 48 possible products of the reaction were quantitated, and data are presented to highlight the head group preference for each of the acyl-CoA esters tested. Three independent microsomal preparations were assayed in duplicate, and data shown are average ± SEM.

Dual substrate choice assay comparison to alternative LPAT assays

The dual substrate choice assay was compared with three other LPAT assays. Specifically, the levels of a single phospholipid molecular species, 17:1/20:4-PC, were measured. The single substrate assay (Single Substrate) used the conditions of a traditional assay where only LPC and 20:4-CoA were added to the microsomes, thus allowing the formation of only one phospholipid species. The lysophospholipid choice assay (Lyso Choice) included 20:4-CoA and six lysophospholipids, therefore allowing the possible formation of six different phospholipid species. The acyl-CoA choice assay (CoA Choice) used LPC and the eight different fatty acyl-CoA esters, allowing the potential formation of eight different phospholipid species. Figure 4A shows an increased 17:1/20:4-PC production with decreased diversity of substrate options. This finding suggested substrate competition for the active site of the LPATs present in the microsomes. Although the single substrate assay yielded a somewhat more abundant product formation, quantitative evidence for substrate selectivity would require numerous individual assays that would less accurately reflect the true in vivo milieu, because within the intact RAW 264.7 cell the LPATs likely reside in an environment with multiple substrate choices.

Fig. 4.

Effect of substrate competition on acyltransferase activity. A: Microsomes from RAW 264.7 cells were incubated for 10 min with mixtures of lysophospholipids and CoA esters (Dual Choice), LPC and a mixture of fatty acyl-CoA esters (CoA Choice), a mixture of lysophospholipids and 20:4-CoA (Lyso Choice), or LPC and 20:4-CoA (Single Substrate). The production of 17:1/20:4-PC was quantitated using LC/MS/MS and is compared between the four assay conditions. B: LPC, LPE, LPI, and LPS were incubated independently with 20:4-CoA. The respective phospholipid products were quantitated using LC/MS/MS. C: A dual choice assay was performed with 6 lysophospholipids and 8 fatty acyl-CoAs. All products were quantitated by LC/MS/MS, and four of the products are displayed. Three independent microsomal preparations were assayed in duplicate, and data shown are average ± SEM.

Importantly, the substrate preference exhibited for arachidonate incorporation into the different phospholipids was essentially identical using either a series of single substrate assays (Fig. 4B) or the dual substrate choice (Fig. 4C). This indicates that the availability of a variety of substrates does not obscure the substrate preference of the enzymes present in the microsomes and supports the usefulness of this assay as a valid alternative to individual assays for each substrate.

Dual substrate choice as a tool to detect changes in LPAT activity

The dual substrate choice assay was used to detect changes in a specific LPAT activity through inhibition of specific LPAT gene expression. Lentiviral particles were used to transduce either a nontargeting shRNA construct (control) or an shRNA construct targeted against the mRNA coding for LPCAT3/MBOAT5 (knockdown) into RAW 264.7 cells. Microsomes from these cells were then tested for changes in LPAT activity by the dual choice assay. Microsomes from cells transduced by the nontargeting construct or the LPCAT3 construct (Fig. 5A, B, respectively) were found by the dual choice assay to incorporate 20:4-CoA into the six lysophospholipids quite differently. Gene expression inhibition was confirmed by measuring LPCAT3 mRNA levels using qPCR (Fig. 5, insert). Microsomes from RAW 264.7 cells were found to incorporate arachidonate into all six classes of phospholipids tested in this assay, albeit to very different extents. In the LPCAT3-targeted microsomes, there was a substantial decrease in the incorporation of 20:4, specifically into PC, PE, and PS, but no decrease in 17:1/20:4-PA, -PG, or -PI production was observed. These changes were clearly indicative of a decrease in LPCAT3/MBOAT5 activity, while other LPAT activities were unchanged in this microsomal preparation. In this case, the intensities shown are the raw ion counts from the detector in the mass spectrometer, with no quantitative corrections applied. Using the ratios of the product ion to the corresponding internal standards and applying calibration curves facilitated the comparison of phospholipid products between alternative species.

Fig. 5.

Reduced expression of LPCAT3/MBOAT5 and its effect on the acyltransferase activity of microsomes. RAW 264.7 cells were transduced with either a nontargeting shRNA construct (A; control, CNTL) or an shRNA targeted against the LPCAT3 gene (B; knockdown, KD). The levels of mRNA expression of LPCAT3, as determined by qPCR, are shown in the inset. The microsomes were incubated with mixtures of lysophospholipids and acyl-CoA esters. The normal-phase chromatographic profiles of the six phospholipids containing the 20:4 acyl chain are shown. The data shown are representative of a total of nine experiments performed with three independent microsomal preparations. The inset data are average ± SEM.

The phospholipid molecular species shown in Fig. 6 were quantified using the corresponding internal standards. The incorporation of 20:4 (Fig. 6A) and 16:0 (Fig. 6B) acyl chains into the six different phospholipids was found to be significantly altered in terms of production of 17:1/20:4-PC, 17:1/20:4-PE, and 17:1/20:4-PS between the control and knockdown microsomes. These are all known products of LPCAT3 catalysis. However, there were no changes in the incorporation of 16:0 chains into any of the phospholipids, consistent with this 16:0-CoA being a poor substrate for LPCAT3 and other enzymes being responsible for this activity on the microsomes. Changes in LPCAT3 activity were also revealed by the incorporation of 20:5-CoA and 18:2-CoA (supplementary Fig. V).

Fig. 6.

Quantitative effect of reduced expression of LPCAT3/MBOAT5 on the acyltransferase activity of microsomes. RAW 264.7 cells were transduced with either a nontargeting shRNA construct (CNTL) or an shRNA targeted against the LPCAT3 gene (KD), and microsomes were used in the dual substrate choice acyltransferase assay. Only the phospholipid molecular species containing 20:4 acyl chains (A) or 16:0 acyl chains (B) were quantitated. Four independent microsomal preparations were assayed for a total of nine experiments, and data shown are average ± SEM.

The change in phospholipid biosynthesis from this mixture of CoA esters and lysophospholipids using nontargeting shRNA control and LPCAT3 knockdown microsomes was also evident from the exact phospholipid molecular species made in the dual choice assay (Fig. 7). In this figure, the dual choice assay results are presented as the quantity of newly synthesized phospholipid molecular species catalyzed by the LPCATs knockdown microsomes after subtraction of the quantity made by the nontargeting shRNA RAW 264.7 cell microsomes when the dual choice assay was performed, which would be expected for a loss of LPCAT3 activity. The largest drop in molecular species occurred in the acylation of LPC and LPS with 18:2-, 20:4-, and 20:5-CoA chains. To a lesser extent, this occurred with acylation of LPE using these same polyunsaturated CoA esters.

Fig. 7.

Phospholipid molecular species production during reduced expression of LPCAT3/MBOAT5 in microsomes. RAW 264.7 cells were transduced with either a nontargeting shRNA construct (CNTL) or an shRNA targeted against the LPCAT3 gene (KD), and microsomes were used in the dual substrate choice acyltransferase assay. Four independent microsomal preparations were assayed. All 48 phospholipids products were quantitated by MS/MS and normalized by microsomal protein content. Each phospholipid molecular species produced by KD cells was subtracted from that measured in CNTL cells. The results are expressed as a histogram for each phospholipid molecular species as a less abundant product (bars going left of center) or more abundant product (bars going right of center).

DISCUSSION

Glycerophospholipids are the major components of all biological membranes, and specific molecular species within this family determine properties such as membrane fluidity and curvature. Glycerophospholipid molecular species composition also affects the activity of membrane-associated enzymes and the production of lipid mediators of inflammation (25). The acyl chain composition of phospholipids is the result of a combination of multiple enzymatic activities, including phospholipases, acyl-CoA synthetases, and LPATs, which lead to the large diversity of phospholipid species necessary for different membrane characteristics.

LPATs have been studied since the 1950s, by using assays in which the enzymatic activity was usually determined using substrates consisting of a single lysophospholipid and one fatty acyl-CoA ester. In the past decade, several LPATs, belonging to two distinct protein families, have been described and characterized. Detailed enzymatic analysis of these proteins, after cloning and expression of the corresponding genes, has revealed a wide variety of activities for specific lysophospholipid and fatty acyl-CoA substrates, sometimes catalyzed by the same enzyme, which makes single substrate analysis less useful for characterizing biological samples. Previous work by this and other groups has used different lysophospholipids and/or acyl-CoA esters to measure LPAT activity (13, 22). However, this method has not been formally validated by comparing its results to those of the corresponding individual single substrate assays, even though the competition of multiple substrates could change the observed use of substrates. Under the conditions used, the substrate competition did not significantly affect the relative incorporation of 14:0, 18:0, 20:4, or 22:6 into the different classes.

In this report, absolute amounts of the phospholipid products were determined by creating and applying standard curves for each of the six phospholipid classes measured. Because ionization efficiencies can differ greatly between the different phospholipid classes, it was important to use separate internal standards and standard curves for each phospholipid class. An additional discovery from our experimentation was the finding that normalization of the activity data can be achieved by measuring endogenous molecular species within each class because the absolute quantity of endogenous phospholipid was a direct measure of the quantity of microsomal preparation taken for assay. This study showed excellent correlation of the 16:0/18:1- and 18:1/18:1-phospholipids with the microsomal protein levels.

The microsomal membranes from RAW 264.7 cells contained a mixture of different enzymes, thus the integrated reacylation activities observed in the dual choice assay reflect this complexity. However, it was possible to identify products that implied the presence of particular LPAT isoforms. In many cases, the phospholipids produced in the reaction are not the most abundant phospholipids in the RAW 264.7 cell. A good example of this is 22:6-containing PE and PS. No LPAT activity has been described at this point that can catalyze the formation of these phospholipids. It is possible in other biological systems that there is an LPAT that can catalyze these reactions, but we do not know precisely which proteins are involved. An alternative explanation of some abundant phospholipids not being formed in this reaction would be that it takes multiple different enzyme subtypes to catalyze these reactions, for instance involving CoA-independent transacylases. Because we are using microsomal extracts and incubating the reaction for a short period of time, we may be missing the formation of these lipids, which would occur in a whole cell system and/or at longer time points.

The characteristic LPAT activity of LPCAT3/MBOAT5 was suggested by specific newly synthesized arachidonoyl-containing phospholipids by the microsomal preparation from the RAW 264.7 cells. This evidence was confirmed by the dramatic reduction of 20:5, 20:4, and 18:2-containing PC, PE, and PS products when LPCAT3 expression was specifically reduced by using the corresponding shRNA construct, whereas no change was observed in the majority of the other phospholipid products. This illustrates the possibilities of the assay in detecting changes in LPAT activity across different cells or experimental conditions. It provides a less targeted approach than the traditional biochemical assays, potentially very useful where the activity or expression of individual acyltransferases has not previously been shown to change.

Use of the dual choice assay also provided evidence for additional acyltransferase enzyme activities that have not been identified. The incorporation of stearoyl chains into PS, PI, and PC was unexpectedly high, but an acyltransferase catalyzing this reaction has not been reported. It is also uncertain which enzyme transfers 22:6 chains into PC and PI, although low levels of these activities for LPA have been reported for LPAAT3. Thus, by probing for multiple acyltransferase activities in one assay, using large combinations of substrate choices, it is now possible to detect novel enzymes with LPAT activity.

The dual choice assay provides a facile and more extensive biochemical approach to studying acyltransferase activity. This novel biochemical technique introduces substrate competition, which provides insight into the preference of substrates, because most biological systems have multiple fatty acyl-CoA esters and lysophospholipids present and is readily implemented using MS/MS.