Alkyne lipids as substrates for click chemistry-based in vitro enzymatic assays.

Click chemistry is evolving as a powerful tool in biological applications because it allows the sensitive and specific detection of compounds with alkyne or azido groups. Here we describe the use of alkyne lipids as substrates for in vitro enzymatic assays of lipid modifying enzymes. The small alkyne moiety is introduced synthetically at the terminus of the hydrocarbon chain of various substrate lipids. After the assay, the label is click-reacted with the azide-bearing fluorogenic dye 3-azido-7-hydroxycoumarin, followed by the separation of the lipid mix by thin-layer chromatography and fluorescence detection, resulting in high sensitivity and wide-range linearity. Kinetic analyses using alkyne-labeled substrates for lysophosphatidic acid acyltransferases, lysophosphatidylcholine acyltransferases, and ceramide synthases resulted in Michaelis-Menten constants similar to those for radiolabeled or natural substrates. We tested additional alkyne substrates for several hydrolases and acyltransferases in lipid metabolism. In this pilot study we establish alkyne lipids as a new class of convenient substrates for in vitro enzymatic assays.


Lipid extraction, click reaction, and TLC
All assay reactions were stopped by the addition of 500 l chloroform/methanol 1/3 (v/v) to the assay mixture. Samples were then incubated for 5 min in an ultrasonic bath and 500 l 1% acetic acid were added, followed by vortexing and centrifugation at 14,000 g for 2 min. One hundred microliters of the chloroform phase were transferred to a new 1.5 ml reaction vessel. The aqueous vivo followed by fl uorescence imaging ( 14,15 ). In addition to these applications, it is especially worthwhile to test the feasibility of click-based in vitro assays in lipid metabolism, for which other labeled substrates often are not readily at hand.
As the alkyne moiety is a small, neither overly hydrophobic nor hydrophilic label, a good affi nity of enzymes to alkynemodifi ed substrates can be expected. For the fi rst time this study addresses the usage of alkyne lipids in enzymatic assays through the determination of Michaelis-Menten kinetics for these substrates. Here we describe a click-based in vitro enzymatic assay to study enzymes of lipid metabolism, notably from the families of lysophosphatidic acid acyltransferases (LPAATs), lysophosphatidylcholine acyltransferases (LPCATs), and ceramide synthases (CerSs).

Alkyne lipids
The chemical syntheses of the alkyne lipids are available in the supplementary material to this study, or were described previously ( 13 ).

Escherichia coli culture and preparation of microsomal fractions
E. coli (strain Rosetta TM 2 pLysS, Merck Millipore) cultured in Luria Broth at 37°C to an OD 600 of 1.6 were harvested by centrifugation at 3,000 g for 10 min. Microsomal fractions were prepared according to Lewin,Wang,and Coleman ( 16 ), frozen in liquid nitrogen and stored at Ϫ 80°C. The protein concentration in the microsomes was determined with the BioRad Bradford assay kit using BSA as a standard.

Cell culture and preparation of cell lysate
HuH7 cells were grown in RPMI 1640 (PAN Biotech P04-17500) supplemented with 10 mM HEPES, 0.1 mM nonessential amino acids, 2 mM L -glutamine, 10% FBS, at 5% CO 2 . Cells were washed, scraped into ice-cold buffer (20 mM HEPES/NaOH, pH 7.0, 200 mM sucrose) and homogenized in a cooled EMBL cell cracker (HGM, Heidelberg, Germany) with fi ve double strokes and a maximum clearance of 18 m. The lysate was centrifuged at 500 g for 5 min at 4°C, the supernatant frozen in liquid nitrogen and stored at Ϫ 80°C. The protein content of the lysate was determined with the bicinchoninic acid assay kit (Pierce) using BSA as a standard.

Animals and preparation of tissue microsomal fractions
Membrane fraction samples were prepared of the brain, liver, and kidneys of C57BL/6 wild-type (+/+) and CerS2-defi cient mice ( Ϫ / Ϫ ) (one individual each, littermates, 13-14 weeks old) according to the protocol described by Imgrund et al. ( 17 ), frozen in liquid nitrogen and stored at Ϫ 80°C. A total lysate of wildtype liver was also prepared. The protein content of the microsomal fractions, and of liver lysate was determined with the bicinchoninic assay kit (Pierce) using BSA as a standard.

Enzymatic assays
All assays were performed in 1.5 ml reaction tubes in a total volume of 100 l. The reaction was started by addition of the prewarmed reaction mix to the enzyme preparation and the tubes were incubated in a heating block (Eppendorf Thermomixer comfort) under shaking (1,100 rpm). Incubation times

Assay setup and method of quantifi cation
The general workfl ow, as outlined in Fig. 1 , consists of the enzymatic reaction using alkyne substrate, lipid extraction, and reaction of the alkyne moieties with the fl uorogenic dye 3-azido-7-hydroxycoumarin in a quantitative copper(I)-catalyzed cycloaddition. After separation by TLC, lipids are analyzed by fl uorescence detection.
The high sensitivity and wide linear dynamic range of this procedure regarding the click reaction, TLC, and imaging were established earlier ( 13 ) and are comparable to methods using [ 3 H]-labeled lipids. For an accurate quantifi cation of the assay products, a reliable and near complete recovery is necessary. To determine the effi ciency of the lipid extraction protocol, defi ned amounts of alkyne lipids that represent products of the LPAAT, LPCAT, or CerS assay were subjected to the standard procedures (in vitro enzymatic reaction, extraction, click reaction, detection), but using heat-inactivated enzyme preparations. The extraction procedure was evaluated by quantifi cation of fl uorescent signal against that of defi ned amounts of alkyne-oleate that had not been subjected to assay incubation and extraction. High recovery rates were measured for all tested lipids, alkyne-OOPA (97%), alkyne-dhCer (85%), alkyne-PC (88%), and pPC (82%), with good to very good reproducibility. This demonstrates the applicability of the extraction protocol (which had been optimized for polar lipids) for the assays performed in this study, and of the quantifi cation method. Thus, an exact correlation of fl uorescent signal to the amount of lipid produced in the assay is achieved.

LPAAT assay with alkyne-OLPA or alkyne-oleoyl-CoA
In E. coli , the conversion of lysophosphatidic acid (LPA) to phosphatidic acid (PA) is catalyzed by the gene product of plsC , 1-acyl-sn -glycero-3-phosphate acyltransferase ( 20 ). Its activity has been assessed with [ 32 P]LPA ( 21 ). To the best of our knowledge, kinetic constants have not been determined for this enzyme.
We established a LPAAT assay with alkyne-OLPA ( Fig. 2A ), the labeled analog of sn -1-oleoyl-lysophosphatidic acid using E. coli microsomes ( Fig. 2B ), and determined the phase was again extracted with 200 l chloroform. The combined organic phases were dried in a speed-vac and the click reaction performed as described previously ( 13 ): To the dried extracts, 7 l chloroform were added to redissolve lipids. Then 30 l click reaction mix (10 l of 2 mg/ml 3-azido-7-hydroxycoumarin, 250 ul of 10 mM [acetonitrile] 4 CuBF 4 in acetonitrile, 850 l ethanol) were added and the tubes incubated in a heating block (Eppendorf Thermomixer comfort) at 43°C without shaking until all solvent was condensed under the lid (3 h).
After brief centrifugation, addition of 30 l chloroform, and 5 min incubation in an ultrasonic bath, the samples were applied onto TLC silica plates (Merck 1.05721.0001). Plates were developed in the respective solvent system: a ) for all assays except CerS: fi rst chloroform/methanol/water/acetic acid 65/25/4/1 for 13 cm, then hexane/ethyl acetate 1/1 for 18 cm, with gentle drying between the two solvents; or b ) for the CerS assay: chloroform/methanol/water 80/10/1 ( 2 ).

Extraction effi ciency experiments
Diluted solutions (0.5 M) of 1,2-di-(nonadec-9-cis -en-18-ynoyl)sn -glycero-3-phosphate (alkyne-OOPA), alkyne-18:0-dihydroceramide (dhCer), alkyne-PC, and propargyl-phosphatidylcholine (pPC) in ethanol were prepared freshly, 10, 20, or 40 l of this solution added to a reaction vessel, the solvent evaporated, and the lipids dissolved in the appropriate assay buffer (see above). After the addition of heat-inactivated enzyme preparation, the reaction mix was incubated according to the respective assay protocol and subjected to extraction, click reaction, and TLC. Quantifi cation against alkyne-oleate was performed as given below, using the same molar amounts of alkyne-oleate as for the extracted lipids.

Detection and quantifi cation
Shortly before fl uorescence detection, the dry TLC was soaked for 5 s in 4% (v/v) N , N -diisopropylethylamine in hexane and excess solvent was allowed to evaporate in a hood. The system used for the imaging of the TLC plates and image quantifi cation was described earlier ( 13 ). Briefl y, standard LEDs (10 × 1 W 420 nm LEDs; Roithner Lasertechnik, Vienna, Austria) fi ltered through a colored glass fi lter (HEBO V01, Hebo Spezialglas) were used for excitation. Images were acquired with a Rolera MGI plus electron multiplying charge-coupled device (EMCCD) camera (Decon Science Tec), equipped with a 494/20 (channel for detection of the coumarin fl uorescence) and 572/28 (channel for detection and correction of background fl uorescence) bandpass emission fi lter wheel, all under control of GelPro analyzer software (Media Cybernetics). Note that the high sensitivity of the EMCCD camera (which is part of our electrochemiluminescence Western blot detection system) is not necessary to record the images, which typically can be seen already by visual inspection, but its large dynamic range is advantageous for image quantifi cation. We also use a system with a much simpler camera with good results. Fluorescent signals were correlated to the lipid amount detected by drying two different defi ned amounts of alkyne-oleate solution in the speed-vac, subjecting them to the click reaction, and applying them to separate lanes in the TLC. All signals of a TLC were quantifi ed against the weighted mean signal of the alkyne-oleate signals.

Statistical analysis and nonlinear regression
Unless stated otherwise (extraction effi ciency measurements), all data are presented as mean values ± standard deviations (n = 3). Michaelis-Menten kinetics were assessed in three independent measurements, a representative graph is shown. For each measurement, the kinetic constants K m and V max were calculated from the data using nonlinear regression in Microsoft Excel ( 19 ), and the mean value and standard deviation of these three calculations are given. headgroup-labeled PLpPC ( Fig. 3A ). Note that because PLpPC is derived by enzymatic headgroup exchange from egg yolk PC, it contains a mixture of fatty acids at sn -1, dominated by palmitate ( 28 ). Figure 3B shows the application of the two substrates in a LPCAT assay on HuH7 lysate using oleoyl-CoA as the acyl chain donor. The K m value we measured ( Fig. 3C ( 29 ). PLpPC had a higher apparent K m (5.46 ± 2.23 M), which might refl ect a somewhat stronger effect of the alkyne label when incorporated in the headgroup of a phospholipid rather than at the terminus of an acyl chain. However, the differences between the K m values for alkyne-OLPC and PLpPC were not statistically signifi cant and the differential acyl chains at the sn -1 positions have to be taken into account. In addition, we observed a significant difference ( P < 0.05) in activity ( V max ) for the two substrates, 0.51 ± 0.08 nmol/(mg·min) for alkyne-OLPC compared with 0.30 ± 0.07 nmol/(mg·min) for PLpPC. On the basis of these results, alkyne-OLPC may be considered as the substrate of fi rst choice for enzymatic assays compared with PLpPC. Nevertheless, the latter also displayed near to natural kinetic constants and allows for the direct combination of enzyme kinetics with microscopic imaging after labeling with propargyl-choline ( 14 ).

Ceramide synthase assay with alkyne-sphinganine
Ceramide synthases catalyze the acyl-CoA-dependent synthesis of dihydroceramide from sphinganine. The six mammalian isoforms display different tissue distributions and acyl chain specifi cities ( 30 ). CerS activity has been measured with radioactive and NBD-labeled sphinganine analogs ( 2,31 ). We synthesized the new substrate alkynesphinganine ( Fig. 4A ) and employed it in a CerS assay on mouse liver microsomes using nervonoyl-CoA ( Fig. 4B ). In kinetic measurements in the same system, we determined an apparent K m of 2.29 ± 2.08 M and a V max of 0.71 ± 0.38 nmol/(mg·min) for alkyne-sphinganine ( Fig. 4C , Table 1 ). In the liver, CerS2 is the dominant isoform with activity toward long-chain acyl-CoAs; hence the activity we measured in this assay may almost solely be attributed to CerS2 ( 32 ). The K m value is in agreement with recent studies: using tritiated sphinganine, Lahiri et al. ( 31 ) reported K m values in the range of 2-5 M for all six (murine or human) CerS isoforms, even for those which use multiple acyl-CoAs. That study reported a K m of 4.8 ± 0.4 M for human Cers2 with lignoceroyl (24:0)-CoA. Kim et al.
( 2 ) developed a fl uorescent assay with NBD-sphinganine and nervonoyl-CoA and found K m to be comparable to natural sphinganine (3.61 ± 1.86 M vs. 3.05 ± 0.81 M, respectively). However, V max was elevated for the NBD derivative.
CerS2 is expressed in various tissues besides the liver, with high mRNA levels in the kidney and moderate levels in the brain ( 30 ). Imgrund et al. ( 17 ) created a CerS2defi cient mouse and measured the CerS activity toward tritiated sphinganine and various acyl-CoAs (16:0, 18:0, 20:0, 22:0, and 24:1) in brain and liver microsomes. To constants V max and K m for this substrate ( Fig. 2C , Table 1 ). Conversely, by applying alkyne-oleoyl-CoA as an acyl chain donor, we measured V max and K m for PLPA. The apparent K m for alkyne-OLPA (0.53 ± 0.18 M) was signifi cantly lower than for PLPA (3.64 ± 0.24 M). Also, V max was apparently increased for alkyne-OLPA compared with PLPA, but the differences did not reach statistical signifi cance due to large variations between different enzyme preparations. In other organisms, K m had previously been determined for LPAAT enzymes using radiolabeled substrates. A similar acyl chain specifi city in the acceptor was found in Spinacia oleracea ( 22 ); with oleoyl-CoA as the donor, a K m of 5.3 M was measured for [ 14 24 )]. For both human isoforms, a preference of OLPA over PLPA as the acceptor was detected as well ( 24,25 ). Compared with palmitate and stearate, oleate is preferred by the E. coli enzyme also as the donor acyl-CoA ( Fig. 2D ) for the incorporation at the sn -2 position, in accordance with previous fi ndings ( 26 ).
Our results demonstrate that alkyne-OLPA and alkyneoleoyl-CoA are good substrates for the E. coli LPAAT. They indicate that the bacterial enzyme may have a lower K m for lysophosphatidic acids (especially OLPA) than the plant or mammalian enzymes investigated so far. The two alkyne substrates described above represent the two strategies that are available for an acyltransferase assay with alkyne lipids, i.e., labeling of the acyl chain acceptor or the acyl chain donor. While the latter is the more general approach, enabling the investigation of all acyltransferase reactions with a relatively small set of labeled alkyne-acyl-CoAs, it suffers from the limited storage stability of acyl-CoAs and the frequent observation of signifi cant amounts of labeled side products arising from both enzymatic and nonenzymatic conversion of the acyl-CoAs. In our experiments, the use of alkyne-oleoyl-CoA led to a few fl uorescent signals derived from side reactions of the reactive compound including the hydrolysis to oleate. The labeled acceptors on the other hand expressed high specifi city for the enzymatic reactions of interest. They essentially gave one or two strong product signals besides the fl uorescent educt ( Figs. 2B, 3B, 4B ), which might be benefi cial for automated image analysis. In addition, alkyne-labeled acyl acceptors show better stability for long-term storage.

LPCAT assay with alkyne-OLPC or PLpPC
Lysophosphatidylcholine (LPC) is converted to phosphatidylcholine (PC) by lysophosphatidylcholine acyltransferases. In mammals, four isoforms (LPCAT1-4) are known. HuH7 cells express LPCAT1 and LPCAT3, but not LPCAT2 ( 18,27 ). We synthesized two alkyne-labeled LPC species, i.e., side chain-labeled alkyne-OLPC and demonstrate that alkyne-sphinganine can be used as a convenient alternative substrate for such applications, we performed a similar small screen with brain, liver, and kidney microsomes and several acyl-CoAs (16:0, 18:0, 18:1, and 24:1). Figure 4D shows fl uorescent images of the TLC plates of this pilot study. Like Imgrund et al. ( 17 ), we observed a strong activity toward 24:1-CoA in the wild-type liver and a weak one in the brain. We also measured a strong 24:1 activity in kidney microsomes. For all tissues, 24:1-CoA was not converted to dhCer in the CerS2-deficient samples. For the acyl-CoAs of shorter chain length, the CerS activity was similar for wild-type and Ϫ / Ϫ mice, refl ecting that it was not derived from CerS2, but other isoforms. We detected activity toward 16:0-CoA in kidney (moderate) and liver (weak), but not in brain. Weak activity was also found for 18:0-CoA in brain and liver, both in wild-type and Ϫ / Ϫ mice. 18:1-CoA was converted to dh-Cer in kidney and very little in liver. Because we show here that alkyne-sphinganine is a useful substrate to screen for CerS activity in tissue samples, more detailed quantitative investigations can be carried out in the future.

Application of the method in assays for other lipid-modifying enzymes
We utilized mouse liver microsomes or mouse liver lysate to demonstrate the principal feasibility of assays with four additional examples of alkyne-labeled substrates ( Fig. 5 ). All substrates were converted by one or more enzymes in the liver fractions to alkyne-labeled products.
Alkyne-PAPA (structure depicted in Fig. 5B ) was converted to DAG in the presence of liver lysate ( Fig. 5B ), presumably by phosphatidate phosphatases ( 34 ). In parallel, we observed the release of alkyne-palmitate from the sn -1 position, either by a phospholipase 1 activity acting on the labeled PA, or by a DAG lipase activity acting on the released DAG.
N -oleoylethanolamide and N -palmitoylethanolamide are bioactive lipids that are cleaved by the enzymes fatty acid amide hydrolase (FAAH) ( 35 ) and N -acylethanolaminehydrolyzing acid amidase ( 36 ) yielding the corresponding fatty acids. Both the lipids and the enzymes have gained increasing interest as therapeutic targets for the control of pain, infl ammation, or food intake ( 37,38 ). We were able to survey the hydrolysis of the fatty acid amides alkyne-oleoylethanolamide and alkyne-palmitoylethanolamide in a time-course experiment ( Fig. 5C ) upon incubation with mouse liver microsomes. Several enzymatic assays are reported to follow this reaction [see ( 35 ) for a review and ( 39 )]. We will conduct further studies concerning the kinetic properties of these alkyne-labeled substrates, which would expand the existing techniques by a very convenient and direct assay procedure.
The various substrates discussed above therefore demonstrate how the scope of the assay can be extended to more acyltransferases (MGAT), lipases (MAGL, phospholipases A1 and A2) and other hydrolases (PA phosphatase, FAAH). This general applicability will prove useful for the kinetic characterization of many enzymes in lipid metabolism.

Scope and limitations of the method
All alkyne lipids that we tested in this pilot study were used as substrates by lipid-modifying enzymes. No substantial shift in affi nity ( K m ) of the enzymes compared with the natural substrates was detected in our kinetic studies. We thus introduce enzymatic assays with alkyne lipids as a reliable and convenient alternative to fl uorescent or radioactive assays.
As with the latter two methods, the TLC separation in our assay does not achieve species resolution of the lipids. This can however be overcome by a subsequent mass-spectrometric analysis [compare ( 13 )]. Depending on the location of the alkyne label in the molecule, not all reactions can be followed with every alkyne lipid. ␤oxidation of fatty acids leads to the loss of the label ( 13 ), as does the headgroup cleavage of pPC to PA by phospholipase D.
Our studies suggest that the alkyne label does not generally interfere with the affi nity of enzymes to the substrates, especially if the alkyne label is attached to the terminus of the alkyl chain. Such fatty acids of different length and unsaturation are synthetically available ( 13,40 ) and can serve as a basis of a versatile toolbox of tailor-made lipid substrates with the acyl chain(s) of choice at defi ned sn -positions, including phospholipids, lysophospholipids, acyl-CoAs, and acylglycerols, using standard chemical and enzymatic procedures. Similar to this strategy, alkyne-sphingolipids can be synthesized from alkyne-sphinganine. Hence, we expect alkyne lipids to become more and more commercially available in the coming years, both as substrates for enzymatic assays as well as standards for identifi cation after TLC separation . The migration behavior of lipids coupled to the coumarin dye is shifted, but, in general, the separation is possible with the same solvent mixes as for radiolabeled lipids or slightly adapted versions thereof and the retention factors of the substances usually are in the regular order. The adaptation of radiolabeled assays to the use of alkyne lipid substrates should therefore be neither costly nor overly time-consuming.
Alkyne lipids are promising substrates for use in in vitro enzymatic assays, as they are inexpensive, versatile, convenient PLpPC (headgroup-labeled). B: Fluorescent image of a TLC that shows a LPCAT assay using either alkyne-OLPC or PLpPC, oleoyl-CoA, and HuH7 cell lysate. For simplicity of the labeling of the TLC plate, we used aPC and aLPC for fatty acid-labeled PC and LPC, respectively, and pPC and LpPC for the headgroup-labeled analogs. Products were identifi ed using comigrating synthetic standards (Std.). "ori" depicts the origin of the TLC. For assay details, see Materials and Methods. C: Michaelis-Menten kinetics measured for alkyne-OLPC (squares) and PLpPC (triangles). Line graphs show the reaction rate (v) calculated with the Michaelis-Menten equation using the values for V max and K m obtained by nonlinear regression fi tting. Data are mean ± SD of triplicate determinations. Fig. 4. CerS assay using alkyne-sphinganine. Alkyne-sphinganine [structure in panel (A)] was applied in a CerS assay using microsomes from mouse tissues. B: Fluorescent TLC image of the assay performed with liver microsomes from a C57BL/6 wild-type mouse. Products were identifi ed using comigrating synthetic standards (Std.). "ori" depicts the origin of the TLC. For assay details, see Materials and Methods. C: Michaelis-Menten kinetics measured for alkyne-sphinganine (squares). The line graph shows the reaction rate (v) calculated with the Michaelis-Menten equation using the values for V max and K m obtained by nonlinear regression fi tting. Data are mean ± SD of triplicate determinations. D: CerS activity toward the acyl-CoAs 16:0, 18:0, 18:1, and 24:1 was measured in microsomes from brain, kidney, and liver from wild-type (+/+) or CerS2 knockout ( Ϫ / Ϫ ) mice. The image shown here is slightly overexposed for the two strongest signals to enhance the visibility of weaker signals. The setup allows the fast detection of the CerS activity profi le for various acyl-CoAs in multiple tissues. and, as we show here for the fi rst time, display the kinetic characteristics of the natural substrates.
The authors would like to thank Christiane Kremser and Prof. Dr. K. Willecke for providing the mice used in the CerS assay.  5. Application of alkyne lipids in assays for other lipid-modifying enzymes. Four different alkyne lipids were incubated with mouse liver microsomes or lysate (C57BL/6 wild-type) to test their applicability as substrates in enzymatic assays. A: Alkyne-1-OMAG deacylation and acylation. Alkyne-1-OMAG was incubated with liver microsomes in the absence ( Ϫ ) or presence (+) of acyl-CoAs as indicated. The data indicate both hydrolysis to oleate (Ole) in the presence of liver microsomes and acylation to DAG if an acyl-CoA is present. B: Alkyne-PAPA deacylation and dephosphorylation. Alkyne-PAPA was incubated with different amounts of liver lysate as indicated. We observed both release of alkyne-palmitate (Pal) and dephosphorylation to DAG. C: Time course of the hydrolysis of fatty acid ethanolamides by liver microsomes. Alkyne-oleoylethanolamide (Ole-EA) and alkyne-palmitoyle thanolamide (Pal-EA) were hydrolyzed upon incubation with the microsomes, but not in their absence, to give the alkyne-labeled fatty acids. A-C: Products were identifi ed using comigrating synthetic standards (Std. EA) . "ori" depicts the origin of the TLC. For assay details, see Materials and Methods.