Glycine N-acyltransferase-like 3 is responsible for long-chain N-acylglycine formation in N18TG2 cells[S]

Long-chain fatty acid amides are signaling lipids found in mammals and other organisms; however, details of the metabolic pathways for the N-acylglycines and primary fatty acid amides (PFAMs) have remained elusive. Heavy-labeled precursor and subtraction lipidomic experiments in mouse neuroblastoma N18TG2 cells, a model cell line for the study of fatty acid amide metabolism, establish the biosynthetic pathways for the N-acylglycines and the PFAMs. We provide evidence that the N-acylglycines are formed by a long-chain specific glycine-conjugating enzyme, glycine N-acyltransferase-like 3 (GLYATL3). siRNA knockdown of GLYATL3 in the N18TG2 cells resulted in a decrease in the levels of the N-acylglycines and the PFAMs. This is the first report of an enzyme responsible for long-chain N-acylglycine production in cellula. The production of the PFAMs in N18TG2 cells was reported to occur by the oxidative cleavage of the N-acylglycines, as catalyzed by peptidylglycine α-amidating monooxygenase (PAM). siRNA knockdown of PAM resulted in an accumulation of [13C18]N-oleoylglycine and decreased levels of [13C18]oleamide when the N18TG2 cells were grown in the presence of [13C18]oleic acid. The addition of [1-13C]palmitate to the N18TG2 cell growth media led to the production of a family of [1-13C]palmitoylated fatty acid amides, consistent with the biosynthetic pathways detailed herein.

technique to mouse N 18 TG 2 cells because these cells produce fatty acid amides, including the long-chain N-acylglycines and PFAMs (12), and express many of the enzymes proposed to function in fatty acid amide metabolism (12,25,26). Knockdown of mGLYATL3 in the N 18 TG 2 cells resulted in decreases in the cellular levels of both the longchain N-acylglycines and the PFAMs.
Merkler et al. (12) proposed that the long-chain N-acylglycines are intermediates in the PFAM biosynthetic pathway: fatty acid → fatty acyl-CoA → N-fatty acylglycine → PFAM (Fig. 1). In this pathway, a long-chain N-acylglycine is oxidatively cleaved to the corresponding PFAM and glyoxylate in a reaction catalyzed by peptidylglycine amidating monooxygenase (PAM). PAM knockdown in the N 18 TG 2 cells resulted in a decrease in the cellular levels of the PFAMs and an increase in cellular levels of the longchain N-acylglycines; results consistent with the proposed pathway in Fig. 1. Our research provides key new information regarding the enzyme-catalyzed reactions that occur in cellula to convert fatty acids into the long-chain N-acylglycines and the PFAMs.
GLYAT catalyzes the conjugation of short-chain acyl-CoA thioesters and glycine to produce short-chain N-acylglycines (C 2 -C 10 ) (21)(22)(23). We proposed that a GLYAT-like (GLYATL) enzyme would catalyze the formation of long-chain Nacylglycines in a similar reaction in vivo (12,19). In 2010, Waluk, Schultz, and Hunt (24) described four human GLYATs named hGLYAT, hGLYATL1, hGLYATL2, and hGLYATL3. Interestingly, hGLYATL2 prefers oleoyl-CoA as a substrate and catalyzes the synthesis of N-oleoylglycine from oleoyl-CoA and glycine in vitro. A Basic Local Alignment Search Tool (BLAST) search of the hGLYATL orthologs against the mouse genome yielded two GLYATs, mouse (m)GLYAT and mGLYATL3. Recently, Dempsey et al. (23) characterized recombinant mGLYAT with glycine serving as the best amino donor substrate and benzoyl-CoA serving as the best amino acceptor substrate. The formation of N-acylglycines possessing acyl chains longer than six carbon atoms was not observed. Therefore, we proposed that the remaining GLYATL ortholog, mGLYATL3, might be the enzyme responsible for long-chain N-acylglycine formation in vivo. To test this hypothesis, we analyzed fatty acid amide levels before and after knocking down mG-LYATL3 expression with siRNA, a technique termed subtraction lipidomics. We applied this subtraction lipidomic the determinations along with the standard deviation. For statistical analyses, the Holm-Šidák method was used to correct for multiple comparisons (30,31).
RT-PCR of GLYATL3 transcripts. Ambion MicroPoly(A) Purist kit was used to purify mRNA from N 18 TG 2 cells grown as described above. An Ambion Retroscript kit was used to generate a cDNA library for RT-PCR experiments. Identification of mGLYATL3 transcripts was completed by nested RT-PCR. The outer PCR product was generated with a denaturation step at 95°C for 5 min followed by 30 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 1 min. A final extension step was performed at 72°C for 10 min. The nested product was subsequently generated with the same PCR conditions. The outer primers (sense, CAAATA-CAAGGGTTGCAAAGTG; antisense, TAACCTTTCCTGCGATG-GTC) were designed and synthesized by Eurofins MWG Operon to amplify a 603 bp product. The amplified product from the RT-PCR reaction was purified from the outer primers using ultrafiltration with a 100 kDa cut-off filter. The nested primers (sense, GCGAGAGAGAGAGGCTGAGA; antisense, ATGGAAGC-CAGGTTGTCATC) were designed and synthesized by Eurofins MWG Operon to amplify a 380 bp product. This product was analyzed by a 1.3% agarose gel and the band was visualized using 0.5 g/ml ethidium bromide under UV light. The positive bands at 380 bp were cut out of the gel, purified by the Promega Wizard SV Gel and PCR Clean-up system, and sequenced by Eurofins MWG Operon.
siRNA knockdown of GLYATL3 in N 18 TG 2 cells. The N 18 TG 2 cells were transfected with anti-mGLYATL3 siRNA using siPORT amine transfection reagent in order to achieve knockdown of mGLYATL3. N 18 TG 2 cells were grown to 80% confluency and then collected by scraping into fresh medium. In each well of a 6-well plate, 2 × 10 5 cells were transfected with 10 l transfection reagent and 20 nM siRNA (sense, GGUUGCAAAGUGAAUUAU-ATT; antisense, UAUAAUUCACUUUGCAACCCT) in growth medium containing 0% FBS via a "reverse transfection method." Cultures were incubated for 5, 8, 12, or 24 h before the cells were collected by scraping and centrifugation. After washing with PBS, the cell pellets were stored at 80°C until the long-chain fatty acid amides were extracted, purified, and analyzed by LC/QTOF-MS as described above. Negative controls include: a) incubation with transfection reagent only or b) transfection reagent with Si-lencer® Select Negative Control No. 1 (scrambled siRNA). Three separate independent experiments were performed for each set of conditions. siRNA knockdown of PAM and purification of N-oleoylglycine in N 18 TG 2 cells. The N 18 TG 2 cells were transfected with anti-PAM siRNA to achieve knockdown of mouse PAM. N 18 TG 2 cells were grown to 80% confluency in 6-well plates. The medium was removed and replaced with 0.5% FBS medium containing 2.5 mM [ 13 C 18 ]oleate and 2.5 mM BSA. After 8 h, anti-mPAM siRNA (sense, GCGGGAAGUUCAUCACUCATT; antisense, UGAGUGAUG-AACUUCCCGCTT) was added to each well in medium with 0% FBS. After a total incubation time of 56 h, the cells were collected by scraping and centrifugation. After washing with PBS, cell pellets were stored at 80°C until the labeled fatty acid amides were extracted. Labeled and unlabeled oleamide was extracted from the cell pellet and subsequently purified using the methods of Farrell et al. (28). Labeled and unlabeled N-oleoylglycine was extracted and purified using preparative TLC. Dried lipid extracts were applied to a prewashed (chloroform:acetone 1:1) 10 cm EMD analytical normal phase TLC plate. Chloroform:methanol:acetic acid (95:5:1) was used until the mobile phase reached 5.5 cm horseradish peroxidase and -actin loading control antibody were from Genscript. N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) and silica were from Supelco. Heptadecanoic acid-d 33 was from C/D/N Isotopes. Fatty acid-free BSA (BSA), [ 13 C 18 ]oleic acid, [1-13 C]palmitic acid, protease inhibitor cocktail (P8340), and 6-thioguanine were from Sigma-Aldrich. DMEM and penicillin/ streptomycin were from Mediatech Cellgro. Mouse neuroblastoma N 18 TG 2 cells were from DSMZ (Deutsche Sammlung von Mikrooganism und Zellkuturen GmBH). Polyvinylidene fluoride membranes were from Millipore. SuperSignal chemiluminescent horseradish peroxidase substrate was from Pierce. MagicMark XP molecular weight ladder was from Invitrogen. All other reagents and cell culture supplies were of the highest quality available from commercial suppliers.

Methods
N 18 TG 2 cell culture. Cells were grown in 225 cm 2 culture flasks in an incubator with a temperature of 37°C and 5% CO 2 . The medium was DMEM supplemented with 100 M 6-thioguanine, 10% FBS, 100 IU/ml penicillin, and 1.0 mg/ml streptomycin. Aliquots of cells were taken for cell counting using a hemocytometer and trypan blue dye.

Identification and quantification of endogenous long-chain fatty acid amides in N 18 TG 2 cells by LC/QTOF-MS.
Lipids, including the long-chain fatty acid amides, were extracted from the N 18 TG 2 cells using a previously published extraction method (27,28). A solid-phase extraction method described earlier (5,27) was subsequently used to further purify the long-chain fatty acid amides from the dried lipid extracts. The fatty acid amide-containing fractions were combined, dried under N 2 at 40°C, and then stored at 20°C until analyzed in more detail. An Agilent 1260 LC system with a Kinetex™ 2.6 m C18 100 Å (50 × 2.1 mm) reverse phase column was connected to an Agilent 6540 quadrupole TOF (QTOF) mass spectrometer equipped with a Dual Agilent Jetstream ESI source in positive ion mode. The LC/ QTOF-MS methods that we used are described in Jeffries et al. (5). Extracted ion chromatograms were obtained from the total ion chromatograms for each of the long-chain fatty acid amides using Agilent MassHunter Qualitative Analysis B.04.00. The fatty acid amides in each extraction were identified by comparison to synthetic standards by molecular ion m/z and retention time. Synthetic standards of the N-acylglycines, Nacyldopamines, and N-acylethanolamines were commercially available. The PFAMs were synthesized according to Jeffries et al. (5). Run-to-run variation for all metabolites was less than ±0.1 min.
Quantification of each identified fatty acid amide was performed by integrating the area under the chromatographic peak and comparing that value against a standard curve constructed using the same fatty acid amide. Standard curves were in the linear range of 0.1-10 pmoles on the column (r 2 > 0.99). N-arachidonoylglycine-d 8 , 1 pmole per 10 l injection, was spiked into each extraction to measure instrument performance and for data normalization. Solvent and slip additive blanks were run to evaluate background oleamide and palmitamide levels (29). In addition, blank samples were run between standards and extracts to eliminate memory effects from the HPLC column. Background levels of these PFAMs were subtracted from analyte levels and were <10% of the endogenous levels in the N 18 TG 2 cells. The levels of the N-acyldopamine reported herein represent the sum total of the amount of the N-acyldopamine and the amount of the N-acyldopamine quinone. The amounts of endogenous fatty acid amides in the N 18 TG 2 cells were reported as the average of [1- 13 C]palmitate feeding study in N 18 TG 2 cells. N 18 TG 2 cells were grown to 80% confluency and fresh medium containing 0.5% FBS, 2.5 mM [1-13 C]palmitate, and 2.5 mM BSA was added to the culture flask. BSA, prepared as described in Farrell et al. (28), was used as a carrier for [1- 13 C]palmitate into the N 18 TG 2 cells. After a 48 h incubation, the medium was removed and the cells were washed with PBS. Cells were collected by scraping, followed by centrifugation and washing with PBS. Cell pellets were stored at 80°C until analyzed by LC/QTOF-MS, as described above. Similar to the endogenous analysis of the N-acyldopamines, the levels of [

Identification and quantification of endogenous long-chain fatty acid amides in N 18 TG 2 cells by LC/QTOF-MS
A panel of long-chain fatty acid amides was identified in N 18 TG 2 cell extracts by comparison of molecular ion m/z values and chromatographic retention times against synthetic standards via LC/QTOF-MS analysis ( Table 1). The chromatographic retention time and mass spectrum of endogenous palmitamide identified in an N 18 TG 2 cell extract matched those of a palmitamide synthetic standard (Fig. 2). These data are representative of the data collected for each endogenous fatty acid amide identified in N 18 TG 2 cells. Tandem MS analyses further validated the identification of these metabolites (supplementary Table 1). The identification of the PFAMs and N-acylglycines is consistent with the previous identification of these metabolites in N 18 TG 2 cell extracts, as well as their proposed role as brain signaling lipids (12,28,32,33). This is the first report of the identification of N-palmitoyldopamine, N-oleoyldopamine, and N-arachidonoyldopamine in N 18 TG 2 cells. A previous report identified N-arachidonoylethanolamine in the N 18 TG 2 cells at levels below our limit of detection (32). Endogenous from the bottom of the plate. The plate was dried and placed in the second running solvent, hexane:diethyl ether:acetone (60:40:5), and run until the mobile phase reached 8 cm. After drying, the final solvent, hexane:diethyl ether (97:3), was run until the mobile phase reached 9.5 cm. An N-oleoylglycine standard was run alongside the cell extract and visualized using KMnO 4 to test for position of the N-acylglycines. The appropriate area was scraped from the TLC plate into anhydrous isopropanol and sonicated in a water bath sonicator for 5 min. After initial sonication, the sample was centrifuged to pellet the silica. The supernatant liquid was syringe filtered with a 0.22 m filter and dried under nitrogen at 40-50°C. The silica was subjected to another round of sonication in isopropanol, centrifugation, and syringe filtering before adding to the previous supernatant. Dried samples were trimethylsilylated with 100 l BSTFA. Samples were flushed with dry N 2 , BSTFA added, flushed briefly again, and allowed to react at 55-60°C for 1 h before running on GC-MS. Samples were spiked with an internal standard, heptadecanoic acid-d 33 , to monitor instrument performance and then injected onto a Shimadzu QP-5000 GC-MS using the conditions described in Farrell et al. (28). Quantification of analytes was achieved by integration of the area under the chromatographic peak compared with a standard curve. The major product of oleamide derivitization was oleonitrile, and oleamide-TMS was also observed (28). The observed sum of these products represents the level of oleamide in the N 18 TG 2 cell extract, as determined by standards run with chemically pure oleamide. The level of the N-oleoylglycine reported herein represents the sum total of the amount of the observed mono-and di-TMS derivatives of N-oleoylglycine. Methods were validated by running spiked controls, and extraction efficiency was determined to be approximately 90% from picomole levels of metabolites in solution.
Western blotting for RNAi confirmation. Knockdown resulting from the RNAi experiments was monitored via Western blotting. Ten percent SDS-PAGE was run and, subsequently, electroblotted to a polyvinylidene fluoride membrane. The membranes were blocked with 5% nonfat dry milk, incubated with the respective primary antibody, and treated with a secondary antibody conjugated to horseradish peroxidase. Visualization of bands was achieved with the SuperSignal chemiluminescent system on radiographic film. -Actin antibody was used as a loading control.

N 18 TG 2 cells express GLYATL3 transcripts
Nested PCR primers were designed to amplify a 380 bp mGLYATL3 product from a cDNA library constructed from the mRNA extracted from an N 18 TG 2 cell pellet. After the first round of PCR, the product was filtered and then used as the template in a second reaction with nested primers. Nested primers were necessary due to the low abundance of mGLYATL3 transcripts in N 18 TG 2 cells. The 380 bp product isolated from the agarose electrophoresis gel was positively identified to be mGLYATL3 by DNA sequencing (supplementary Fig. 1).
amounts of the identified long-chain fatty acid amides were obtained by integrating the area under the chromatographic peak and comparing the integrated value against a standard curve prepared using the respective synthetic standard (supplementary Table 2). Known to be potent cell-signaling lipids, these long-chain fatty acid amides are found at cellular levels on the order of picomoles per 10 7 cells. Palmitamide was the most abundant fatty acid amide with an endogenous amount of 1,300 ± 98 pmoles per 10 7 cells, while palmitoleamide was the least abundant. unclear and there are many factors contributing to these differences. One possibility could be differences between the rates of hydrolysis of N-acylglycines relative to PFAMs by fatty acid amide hydrolase (FAAH) (34). Other than changing the cellular levels of the N-acylglycines and PFAMs, the knockdown of mGLYATL3 expression resulted in no change in N-oleoylethanolamine levels and an increase in N-acyldopamine levels. The increases in N-acyldopamine levels were unexpected, but could be due to an increase in acyl-CoAs available for conjugation to dopamine by an unidentified arylalkylamine N-acyltransferase-like (AANATL) enzyme. In Drosophila, AANATL2 catalyzes the formation of long-chain N-acyldopamines in vitro (35,36) and we propose that a mammalian AANATL2 ortholog could be responsible for N-acyldopamine formation in the N 18 TG 2 cells.
Controls for this siRNA experiment indicate that overall protein expression was unaffected by the knockdown of mGLYATL3, as -actin expression was unchanged (Fig. 3). Average cell viability during knockdown was 80% relative to the experiment with cells transfected with scrambled siRNA. At the 24 h time point, average cell viability was 88% when the cells were returning to normal metabolism. Negative controls included incubation with transfection reagent only or transfection reagent with scrambled siRNA. Long-chain fatty acid amide levels were unchanged in these control experiments ( Table 2). When the transient knockdown of mGLYATL3 is no longer present in the cell culture (at the 24 h time point), N-acylglycine and PFAM levels increase. This further validates the role of mGLYATL3 in the production of the N-acylglycines as intermediates in the biosynthesis of the PFAMs.
These data indicate that N 18 TG 2 cells express mGLYATL3 transcripts.

siRNA knockdown of GLYATL3 results in decreased levels of N-acylglycines and PFAMs
Transfection of siRNA against mGLYATL3 into N 18 TG 2 cells resulted in the knockdown of mGLYATL3 expression for 12 h (Fig. 3). While mGLYATL3 expression levels decreased 103 ± 13% during knockdown, -actin levels only changed 12 ± 7% (Fig. 3). After 24 h, the expression of mGLYATL3 returned to normal levels, which limited our ability to do a heavy-labeled precursor study during mGLYATL3 knockdown. The mGLYATL3 primary antibody was specific for recombinant maltose binding protein-tagged mGLYATL3, but not recombinant mGLYAT (23) (supplementary Fig. 2). After 8 and 24 h, the transfected cells were collected and the long-chain fatty acid amides extracted, purified, and analyzed by LC/QTOF-MS. A panel of identified fatty acid amides were quantified and compared with endogenous levels (supplementary Table 2). At the 8 h time point, mGLYATL3 expression was silenced and there were significant changes in N-acyl glycine and PFAM levels. N-oleoylglycine levels decreased 99.4% relative to the untransfected control, while oleamide levels decreased 93.6%. Similarly, N-palmitoylglycine levels decreased 90.4% relative to the untransfected control, while palmitamide levels decreased 86.9%. These data (Fig. 4) indicate that: a) mGLYATL3 has a major role in N-acylglycine formation in these cells; and b) that the N-acylglycines are precursors for PFAMs. Both are consistent with the biosynthetic pathways for the N-acylglycines and PFAMs that we have proposed (Fig. 1). The differences in the percentage changes between oleoylated or palmitoylated amides are  (Fig. 6). These data provide additional evidence that PAM is involved in the conversion of N-acylglycines to PFAMs in cellula.

C]palmitoylated fatty acid amides
The N 18 TG 2 cells were incubated with [1-13 C]palmitate in order to investigate metabolic flux after being exposed to a heavy-labeled precursor. After a 48 h incubation with [1-13 C]palmitate using BSA as a carrier, the lipids were extracted from the cells to analyze for fatty acid amides that contained the 13 C-N-palmitoyl chain as well as the other unlabeled metabolites. [ 13 C]N-palmitoylglycine, [ 13 C]Npalmitamide, and [ 13 C]N-palmitoyldopamine were identified in cell extracts, showing that these metabolites are synthesized from the exogenous [1-13 C]palmitate ( Table 3). The total palmitamide ( 13 C-labeled and unlabeled) in the N 18 TG 2 cells that were incubated with [1-13 C]palmitate was

siRNA knockdown of PAM results in an accumulation of N-oleoylglycine and a decrease in oleamide levels
Transfection of siRNA against PAM into N 18 TG 2 cells resulted in the knockdown of PAM expression for 48 h (Fig. 5). The average viability after 48 h was 56% relative to untransfected cells. Because a large number of peptide hormones are -amidated for biological activation (37), a decrease in cell viability was expected upon the knockdown of PAM. After a total of 56 h of incubation with anti-PAM siRNA and 48 h with [ 13 C 18 ]oleate, the cells were collected and the amounts of labeled and unlabeled N-oleoylglycine and oleamide were analyzed and quantified. In a previous study, the incubation of N 18 TG 2 cells with a PAM inhibitor, trans-4-phenyl-3-butenoic acid, resulted in the accumulation of N-oleoylglycine (12). Consistent with these earlier results, decreased expression of PAM resulted in the accumulation of N-oleoylglycine and a decrease in oleamide levels relative to endogenous amounts the conjugation of the resulting acyl-CoA with glycine. It is known that long-chain fatty acid amides can be produced by more than one route; there is evidence for their biosynthesis by the sequential oxidation of N-acylethanolamines (16)(17)(18)20). There is evidence that this alternative pathway does take place in the N 18 TG 2 cells (28). The subtraction lipidomic data herein suggests that the glycinedependent route is the predominant route for long-chain N-acylglycine synthesis in these cells because N-acylgycine levels were not rescued by an alternative biosynthetic pathway during mGLYATL3 knockdown. This is the first report of an enzyme responsible for the biosynthesis of longchain N-acylglycines in cellula. In vitro data obtained using purified recombinant GST-tagged mGLYATL3 shows the production of N-oleoylglycine from oleoyl-CoA and glycine providing support for our in cellula results (supplementary Table 3). Furthermore, Waluk, Schultz, and Hunt (24) previously described a human enzyme, hGLYATL2, that catalyzes the conjugation of long-chain acyl-CoA thioesters to glycine in vitro. Interestingly, our BLAST searches did not identify orthologs of hGLYATL2 in organisms known to produce long-chain N-acylglycines, such as the mouse and the rabbit. However, these species did have putative orthologs of mGLYATL3. A similar strategy employed by Saghatelian and Cravatt [discovery metabolic profiling, see (38)] led to the identification of N-acyltaurines as substrates for FAAH, despite the relatively low k cat /K M values obtained in vitro for the FAAH-catalyzed hydrolysis for the N-acyltaurines. As these authors point out (39), in vitro substrate specificity data obtained using a purified enzyme may not perfectly predict in vivo substrates for the same enzyme because proteins often function as partners in multi-protein complexes and networks and may be regulated in ways not fully captured in in vitro experiments. A subsequent decrease in PFAM levels with the knockdown of mGLYATL3 gives support to the proposed role for the N-acylglycines as intermediates in the biosynthesis of the long-chain PFAMs. These data are further validated by the siRNA-mediated knockdown of the 990 ± 480 pmoles per 10 7 cells, consistent with the endogenous amount of palmitamide reported herein, 1,300 ± 98 pmoles per 10 7 cells. A similar trend was also observed for N-palmitoyldopamine. A low percentage, 0.1%, of the incubated [1-13 C]palmitate was converted to labeled N-palmitoylglycine, palmitamide, or N-palmitoyldopamine by N 18 TG 2 cells. CONCLUSIONS mGLYATL3 catalyzes the formation of long-chain Nacylglycines in cellula. Subtraction lipidomic data observed a decrease in N-acylglycine and PFAM levels in N 18 TG 2 cells after the siRNA-mediated knockdown of mGLYATL3. These data support the proposed biosynthetic pathway of activation of a fatty acid by acyl-CoA synthetase followed by  the following biosynthetic pathway: fatty acid → fatty acyl-CoA → N-fatty acylglycine → PFAM (Fig. 1).