Lipid Discovery by Combinatorial Screening and Untargeted LC-MS/MS

We present a method for the systematic identification of picogram quantities of new lipids in total extracts of tissues and fluids. It relies on the modularity of lipid structures and applies all-ions fragmentation LC-MS/MS and Arcadiate software to recognize individual modules originating from the same lipid precursor of known or assumed structure. In this way it alleviates the need to recognize and fragment very low abundant precursors of novel molecules in complex lipid extracts. In a single analysis of rat kidney extract the method identified 58 known and discovered 74 novel endogenous endocannabinoids and endocannabinoid-related molecules, including a novel class of N-acylaspartates that inhibit Hedgehog signaling while having no impact on endocannabinoid receptors.


Chemicals and standards
Solvents of LC grade were purchased from Sigma-Aldrich (Munich, Germany) and Fisher Scientific (Schwerte, Germany). Endocannabinoid standards were from Cayman Chemical Company (Ann Arbor, MI).

Rat kidney homogenization
To 500µL of 0.1% formic acid in H 2 O were added to 50 mg of rat kidney tissue and the sample was twice homogenized at 4°C on TissueLyser II (QIAGEN) with 30 freq/sec for 1.5 min with 30 sec break.

Extraction of endocannabinoid-related compounds
Extraction was performed at 4 °C in a cold room. 750 μl of ethyl acetate/n-hexane (9:1 v/v) containing 0.1% formic acid were added to the kidney homogenate and vortexed for 30 sec followed by 10 min centrifugation at 14,000 g. Samples were incubated on dry ice for 10 min and the upper (organic) phase was collected and dried in a vacuum centrifuge. The samples were re-dissolved in 90 μL of water/ acetonitrile/ iso-propanol/ formic acid (6:3.6:0.4:0.1, v/v/v/v) mixture, centrifuged for 5 min at 14,000 g, transferred into a new Eppendorf tube, centrifuged for 5 min at 14,000 g and then 90 μL of the solution transferred into a 300 μL glass vial for LC-MS/MS analyses.

a) N-acylaspartate binding to CB 1 R nd CB 2 R
For cannabinoid receptor studies, tissues were resuspended in 2 mM Tris-EDTA, 320 mM sucrose, 5 mM MgCl 2 (pH 7.4), and then they were homogenized in a Potter homogenizer and centrifuged at 4°C sequentially at 800xg (10 min), and 10,000xg (30 min). The resulting pellet was resuspended in assay buffer (50 mM Tris-HCl, 2 mM Tris-EDTA, 3 mM MgCl 2 , pH 7.4). The membrane preparation was divided in aliquots, quickly frozen in liquid nitrogen, and stored at -80 °C for no longer than 1 week. The membrane fractions (100 µg per test) were used in rapid filtration assays with the synthetic cannabinoid [ 3 H]CP55.940. In all assays, homogenates were pre-incubated for 15 min with each compound tested. In all experiments, unspecific binding was determined in the presence of an excess (1 µM) of ''cold'' agonist, and also in the presence of selective antagonists (SR141716A for CB 1 and SR144528 for CB 2 ) (Pucci et al., 2012).

b) Inhibition of FAAH by N-acylaspartate
Mouse brain homogenates (40 μg per test) were pre-incubated for 15 min at room temperature with each compound tested, then they were incubated with 10 μM [ 14 C-ethanolamine]anandamide for 15 min at 37 °C, in 500 μl of 50 mM Tris-HCl buffer (pH = 9). The reaction was stopped by the addition of 1 mL of ice-cold methanol/chloroform (2:1, v/v). The mixture was centrifuged at 3000xg for 5 min, the upper aqueous layer was put in a vial containing liquid scintillation cocktail (Ultima Gold XR, Perkin Elmer Life Sciences), and radioactivity was quantified in a β-counter. FAAH activity was expressed as pmol [ 14 C-ethanolamine] released/min per mg of protein (Gattinoni et al., 2010).

Inhibitory activity of N-acylasparates against Hh signalling
Shh-LIGHT2 cells (Taipale, 2000), which represent a reporter for Hedgehog (Hh) signalling pathway activity, were maintained in Dulbecco's Modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS), 150 µg/mL Zeocin (Invitrogen), and 400 µg/mL G418 (Invitrogen). Twenty-four hrs prior to assay, cells were plated at a density of 70,000 cells per well in 96-well plates. Cells were then switched to serum-free medium (DMEM + 1% ITS-X) and supplemented with conditioned medium from mock transfected HeLa cells (background), Sonic Hedgehog (Shh) transfected HeLa cells (which release a processed but non-sterol-modified form of Shh), Smoothened agonist (SAG) (Cayman) or DMSO (background). Shh and SAG activities were assayed in the presence or absence of different endocannabinoid classes and species. 20:4 ethanolamide (NAE 20:4) served as a control (Khaliullina, 2015). Luciferase activity was measured in cell lysates after 24 hrs, as instructed by the manufacturer (Dual Glo Luciferase Assay, Promega). The resulting Hedgehog (Hh) pathway activity was determined as the ratio between Firefly : Renilla luciferase.

XIC alignment algorithm employed by the Arcadiate software
Arcadiate employs an original two-step XIC alignment algorithm. First, for every precursor a global correlative chromatogram is produced by computing the average of natural logarithms of intensities of the precursor and its fragments ions. The chromatogram identifies time ranges where precursor and fragments are detected simultaneously. Second, individual time shift values are computed by aligning XIC peaks of the precursor and its fragments within close proximity of the precursor peak. Overall, the alignment quality is assessed by a global detection score expressed in % of fragments with a successful time -based correlation to their precursor. If the time-shift of the optimal overlap between the precursor and a corresponding fragment peak candidate is less than the estimated chromatographic peak width, then the fragment is scored as detected (value 1). If the time shift is larger or the signal intensity of the fragment exceeds the precursor intensity, then it is scored as non-detected (value 0). The overall score termed as molecular marker detection score is computed by averaging the scores of all fragments and expressed in %. A molecular marker will have a detection score of 100% if a peak in its XIC is aligned with XIC of all expected fragments according to the criteria above. To save computation time only three most abundant peaks in the XIC of every precursor are tested for concurrency with fragment peaks and then the XIC peak with the highest detection score is reported along with its score value and absolute retention time.
The correlative chromatogram peak intensity and the retention time-score values tuple for every marker ion are both active numbers. One click onto them displays the overlay of the precursor and fragment chromatograms for a rapid visual evaluation. An example of aligned chromatograms as displayed by the Arcadiate graphical interface is shown below.
Here A, B and C are XICs of the precursor (molecular marker; in blue), fragment (in red) and their alignment, respectively. Interface panel D displays the compound name (here N-acetylethanolamine NAE 20:4); its m/z (348.2897 Th); the logarithmic abundance of correlative chromatogram of the precursor and its fragments; detection score (here 100%) and retention time (here 11.71 min) of the best correlating peak. The second table displays the fragment m/z (62.06 Th) and whether this fragment correlated with the precursor peak (1 for yes, 0 for no) for the precursor selected in the first column. Here mE321_02 is the name of the dataset used in this experiment.

Chemical synthesis of N-acylaspartate
All chemicals were obtained from commercial sources (Acros, Sigma,-Aldrich, Lancaster or Merck) and were used without further purification. Solvents for flash chromatography were obtained from VWR and dry solvents were obtained from Sigma. Deuterated solvents were obtained from Deutero GmbH, Karlsruhe, Germany. All reactions were carried out using dry solvents under an inert atmosphere unless otherwise stated in the respective experimental procedure. TLC was performed on precoated plates of silica gel (Merck, 60 F254) using UV light (254 or 366 nm) or a solution of phosphomolybdic acid in EtOH (10 g phosphomolybdic acid, in 100 mL EtOH) for analysis. Preparative column chromatography was performed using silica form Merck, Darmstadt, Germany (silica 60, grain size 0.063-0.200 mm) with a pressure of 1 -1.5 bar. 1 H-and 13 C-NMR-spectra were obtained on a 400 MHz Bruker UltraShield TM spectrometer. Chemical shifts of 1 H-and 13 C-NMR-spectra are referenced indirectly to tetramethylsilane.
J values are given in Hz and chemical shifts in ppm. Splitting patterns are designated as follows: s, singlet; d, doublet; t, triplet; q, quartett; m, multiplet; m c , centered multiplet; b, broad. 13 C-NMR-spectra were broadband hydrogen decoupled. High resolution ESI mass spectra were recorded on a Q Exactive tandem mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) in direct infusion mode using robotic ion source TriVersa Nanomate (Advion BioSciences, Ithaca NY). 1a,b,c; 2a,b,c and 3

General procedure for the synthesis of N-acyl-L-aspartic acid di(tert-butyl)ester derivatives 1a,b,c
A solution of ethyl cyano(hydroxyimino)acetate (oxyma pure) (0.15 eq) and HBTU (1.1 eq) in DMF (10 ml) was treated with a solution of the respective fatty acid (1.0 eq) in dry DMF (1 ml) while stirring at room temperature under an argon atmosphere. DIEA (2.0 eq) was added immediately afterwards and the reaction mixture stirred for 5 min. Stirring was continued for additional Over O/N after subsequent addition of L-aspartic acid di(tert-butyl)ester (1 eq). The reaction mixture was diluted with a mixture of EtOAc and H 2 O (1:1, 100 ml) and the layers were separated. The organic layer was washed with H 2 O mixed with 20 to 50 % brine (5 x 100 ml) and saturated NaCl solution (1 x 50 ml) and dried over Na 2 SO 4 .
The solvent was removed under reduced pressure and the residue purified by flash chromatography using the eluent cyclohexane/EtOAc 3:1. The respective target compounds were obtained as colorless oils 1a, 1b or as white solid 1c.
N-ararchidonyl-L-aspartic acid di(tert-butyl)ester (1a) Yield: 510 mg, 59%. NMR spectra of pure NAAsp showed extreme signal broadening in a wide temperature range, suggesting several conformations in equilibrium and were therefore excluded from the analytical data.