N-Acylethanolamines as novel alcohol dehydrogenase 3 substrates

Article abstract of DOI:10.1016/j.abb.2010.12.002

N-Acylethanolamines (NAEs) are members of the fatty acid amide family. The NAEs have been proposed to serve as metabolic precursors to N-acylglycines (NAGs). The sequential oxidation of the NAEs by an alcohol dehydrogenase and an aldehyde dehydrogenase wo

Full text of DOI:10.1016/j.abb.2010.12.002

Archives of Biochemistry and Biophysics 506 (2011) 157–164
Contents lists available at ScienceDirect
Archives of Biochemistry and Biophysics
journal homepage: www.elsevier.com/locate/yabbi
N-Acylethanolamines as novel alcohol dehydrogenase 3 substrates
1
⇑
Milena Ivkovic, Daniel R. Dempsey, Sumit Handa, Joshua H. Hilton, Edward W. Lowe Jr. , David J. Merkler
Department of Chemistry, University of South Florida, 4202 E. Fowler Ave., CHE 205, Tampa, FL 33620-5250, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 2 August 2010
and in revised form 1 December 2010
Available online 6 December 2010
N-Acylethanolamines (NAEs) are members of the fatty acid amide family. The NAEs have been proposed
to serve as metabolic precursors to N-acylglycines (NAGs). The sequential oxidation of the NAEs by an
alcohol dehydrogenase and an aldehyde dehydrogenase would yield the N-acylglycinals and/or the NAGs.
Alcohol dehydrogenase 3 (ADH3) is one enzyme that might catalyze this reaction. To define a potential
role for ADH3 in NAE catabolism, we synthesized a set of NAEs and evaluated these as ADH3 substrates.
NAEs were oxidized by ADH3, yielding the N-acylglycinals as the product. The (V/K)app values for the
NAEs included here were low relative to cinnamyl alcohol. Our data show that the NAEs can serve as alco-
hol dehydrogenase substrates.
Keywords:
Alcohol dehydrogenase 3
N-Acylethanolamine oxidation
N-Acylglycinal
Ó 2010 Elsevier Inc. All rights reserved.
N-Fatty acylglycine
Introduction
creted [1-14C]-oleamide indicating that these cells must contain
the enzymatic machinery required for oleamide biosynthesis.
The fatty acid amide bond has long been recognized in nature,
dating back decades to the work defining the structure of the cera-
mides and sphingolipids [1,2]. N-Palmitoylethanolamine, CH3–
(CH2)14–CO–NH–CH2–CH2–OH, was the first non-sphingosine fatty
acid amide isolated from a biological source, egg yolk in 1957 [3].
Interest in the fatty acid amides burgeoned with the identification
of N-arachidonoylethanolamine (anandamide) as the endogenous
ligand for the cannabinoid receptors, CB1 and CB2, in the brain
[4]. We now know that a family of N-acylethanolamines, in addi-
tion to anandamide and N-palmitoylethanolamine, is found in
the brain and other tissues. Most of the N-acylethanolamines
(NAEs)2 seem to play important signaling roles in mammals [5–7].
We have proposed that two classes of fatty acid amides are
linked in a biosynthetic pathway, the N-fatty acylglycines and
the PFAMs [8,9]. Oxidative cleavage of the N-fatty acylglycines at
While our data support the hypothesis that the N-fatty acylglycines
are biosynthetic precursors for the PFAMs, other possible pathways
for the in vivo production of the PFAMs have been suggested, most
notably the nucleophilic attack of fatty acyl-CoA thioesters
by ammonia as catalyzed by cytochrome c [12]: CoA–S–CO–
R + NH3 ? R–CO–NH2 + CoA–SH.
Short-chain N-acylglycines like N-butyrylglycine, N-isovaleroyl-
glycine, and N-hexanoylglycine, are mammalian metabolites [13–
15]. Trace levels of longer-chain N-acylglycines, N-octanoylglycine,
and N-decanoylglycine, are known only from patients with a defi-
ciency in medium-chain acyl-CoA dehydrogenase [16,17]. Longer-
chain N-fatty acylglycines were unknown in mammals until the
isolation of N-arachidonoylglycine from bovine and rat brain
[18]. Since the discovery of N-arachidonoylglycine, there have been
other reports of mammalian N-fatty acylglycines: N-linoleoylgly-
cine, N-oleoylglycine N-palmitoylglycine, N-stearoylglycine, and
N-docosahexaenoylglycine [18–20]. Bradshaw et al. [20] also iso-
lated N-oleoylglycine from mice, consistent with our earlier work.
The known N-fatty acylglycines are widely distributed throughout
the body with the highest levels being found in the skin, lung,
spinal cord, liver, and kidney [19,20].
the C –N bond of the glycyl residue would yield the corresponding
a
PFAM and glyoxylate. In vitro studies have shown that this reaction
is catalyzed by peptidylglycine
a-amidating monooxygenase,
(PAM, EC 1.14.17.3) with VMAX/KM values comparable to the pep-
tide precursor substrates [10]. In addition, elimination of PAM
activity in mouse neuroblastoma N18TG2 cells resulted in the accu-
mulation of N-oleoylglycine when these cells were grown in the
presence of oleic acid [9]. Previously, Bisogno et al. [11] had shown
that N18TG2 cells cultured in the presence of [1-14C]-oleic acid se-
The identification of mammalian N-fatty acylglycines leads to
the question regarding the pathway(s) for their biosynthesis. One
possible pathway involves the conjugation of glycine to the fatty
acid via nucleophilic attack of the glycyl a-amino group at the car-
boxylate of the fatty acid. Huang et al. [18] and Bradshaw et al. [21]
demonstrate that this reaction does take place in the cell; however,
the details of the chemistry are undetermined. It seems unlikely the
glycine would be directly conjugated to a fatty acid, but instead
would involve nucleophilic attack of an activated fatty acid, an
acyl-CoA thioester being an obvious possibility. Acyl-CoA:glycine
⇑
Corresponding author. Fax: +1 813 974 3203.
E-mail addresses: merkler@usf.edu, merkler@chuma1.cas.usf.edu (D.J. Merkler).
Present address: Vanderbilt University Center for Structural Biology, 465 21st
1
Ave., South BIOSCI/MRBIII, Rm. 5144F, Nashville, TN 37232-8725, USA.
2
Abbreviations used: ADH3, alcohol dehydrogenase 3; NAEs, N-acylethanolamines;
NAGs, N-acylglycines; PMS, phenazinemethosulfate.
0003-9861/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.abb.2010.12.002

158
M. Ivkovic et al. / Archives of Biochemistry and Biophysics 506 (2011) 157–164
N-acyltransferase (ACGNAT or GLYATL) [22,23], bile acid-CoA:ami-
no acid N-acyltransferase (BAAT) [24], and cytochrome c [25] are all
enzymes which catalyze glycine conjugation to an acyl-CoA. For
each enzyme, there are in vitro studies suggesting that each could
produce N-fatty acylglycines from the corresponding N-fatty acyl-
CoA and glycine, but data conclusively validating their role in N-
fatty acylglycine biosynthesis in vivo are currently lacking. Other
possible activated fatty acids for attack by glycine could be a fatty
acid adenylate similar to the amino acid adenylates involved in
charging tRNAs or in the biosynthesis of mycothiol or a fatty acid
thioester linked to a cysteine in a carrier protein, similar to interme-
diates observed in fatty acid biosynthesis. Alternatively, the N-fatty
acylglycines might be produced by a novel N-acylglycine synthase
via the initial formation of an acyl-enzyme intermediate which is
subsequently attacked by a bound glycine. Such a novel N-acylgly-
cine synthase is consistent with data from Huang et al. [18] and
Bradshaw et al. [21] showing that fatty acids are directly conjugated
DEAE cellulose column was loaded onto a Mimetic Blue II affinity
column (1.5 Â 20 cm) that was previously equilibrated with
10 mM Tris–HCl pH 8.0 and 2 mM DTT. The column was washed
with the same buffer and flow rate was adjusted to 1 ml/min. Pro-
tein content of fractions was monitored by measuring absorbance
at 280 nm. Once the A280 was below 0.2, a linear gradient of 0–
20 mM NAD+ (total volume 200 ml) was applied to the column in
order to elute ADH3.
Fractions (4.0 ml) were assayed for ADH3 activity by measuring
NADH production spectrophotometrically at 340 nm, using cin-
namyl alcohol as the substrate. Assays were carried out in
100 mM glycine–NaOH pH 10.0, 5.0 mM cinnamyl alcohol, and
2.5 mM NAD+ at 37 °C. Note that 1 unit of ADH3 activity corre-
sponds to 1 lmol of NADH produced/min under these conditions.
NADH production was followed at 340 nm using a JASCO V-530
UV–Vis spectrophotometer. Crude enzyme fractions are known to
contain ADH1 and ADH2 and were, thus, evaluated in presence
of 3.0 mM pyrazole, a known inhibitor ADH1/ADH2 [31]. In addi-
tion to this, rates for controls without cinnamyl alcohol were sub-
tracted from those obtained in the presence of substrate to
eliminate false positives.
Fractions with the highest specific activity were analyzed for
purity using SDS–PAGE, and those containing only a single protein
band corresponding to the weight of an ADH3 subunit (ꢀ40 kDa)
were pooled together. Protein concentration of purified ADH3
was determined using the Bradford dye binding assay [32] and
the concentrated enzyme (6.5 mg/ml) was stored at À80 °C.
to glycine and could enable a decrease in the pKa of the glycyl a-
amino group in the enzyme active site to increase the nucleophilic-
ity of this moiety. Amino acid promiscuity by a novel N-acylglycine
synthase might account for the plethora of N-fatty acylamino acids
recently identified in mammalian brain by Tan et al. [26].
Another suggested pathway for the biosynthesis of the N-fatty
acylglycines is the sequential oxidation of the NAEs by a fatty alco-
hol dehydrogenase and fatty aldehyde dehydrogenase [27]. The
first evidence for this pathway to the N-fatty acylglycines was pre-
sented by Burstein et al. [28] showing that anandamide was con-
verted to N-arachidonoylglycine in cultured liver cells. More
recent work by Bradshaw et al. [21] in RAW 264.7 and C6 glioma
cells again showed that anandamide is oxidized to N-arachidonoyl-
glycine. Finally, Aneetha et al. [27] found that purified human alco-
hol dehydrogenase 4, hADH4, catalyzed the NAD+-dependent
oxidation of ananadamide in vitro and will dismutate the interme-
diate N-arachidonoylglycinal to N-arachidonoylglycine. These data
all strongly suggest that the NAEs can serve as precursors to the N-
fatty acylglycines. Intriguingly, data from Bradshaw et al. [21]
show that N-arachidonoylglycine is produced by both anandamide
oxidation and glycine conjugation to arachidonic acid in cultured
mammalian cells.
We report here, for the first time, that a series of NAEs are sub-
strates for alcohol dehydrogenase 3 from bovine liver (ADH3).
Alcohol dehydrogenase 3 enzymes are found in virtually all
eukaryotes and prokaryotes and represent the ancestral enzyme
from which all the other medium-chain ADHs originated [29].
NAEs from N-acetylethanolamine to anandamide are oxidized by
ADH3 with a general trend of the longer chain NAEs exhibiting
the highest V/K values. N-Benzoylethanolamine and other N-aryla-
cylethanolamines are also ADH3 substrates with N-(tBOC)-etha-
nolamine having the highest V/K value from this class of novel
ADH3 substrates. ADH3-mediated oxidation of the NAEs results
in the formation of the corresponding N-fatty acylglycinals show-
ing that ADH3 does not catalyze the dismutation of the acylglyci-
nals to the N-fatty acylglycine and NAE.
Synthesis of the NAEs
A number of N-acylethanolamines were available commercially
in high purity and these were obtained and used without further
purification. N-Arachidonoylethanolamine and N-linoleoylethanol-
amine were purchased from Cayman Chemical Company,
N-acetylethanolamine was obtained from Sigma–Aldrich, and
N-propionylethanolamine was purchased from TCI.
The remaining N-acylethanolamines included here were
synthesized by the addition of ethanolamine to the acid
chloride as described [33,34]. [1,2-13C]-N-Octanoylethanolamine,
CH3–(CH2)6–CO–NH–13CH2–13CH2–OH, was prepared similarly ex-
cept that commercially available [1,2-13C]-ethanolmine–HCl was
first neutralized with a 10-fold molar excess of triethylamine be-
fore the addition of octanoyl chloride. Purity and identity of syn-
thesized NAEs were confirmed by 1H and 13C NMR using a Varian
iNOVA 400 MHz NMR. The acid chlorides, ethanolamine, and
[1,2-13C]-ethanolamine–HCl were obtained from Sigma–Aldrich
and were used without further purification. Anhydrous CH2Cl2
was purchased from EMD.
Synthesis of the N-acylglycinals
N-Benzoylglycinal and N-octanoylglycinal were prepared using
a procedure slightly modified from Brown [35]. N-Acylglycinal
diethyl acetals were prepared by reacting aminoacetaldehyde
diethyl acetal with the corresponding acyl chlorides under
basic conditions. N-Benzoylglycinal and N-octanoylglycinal were
obtained by acid-catalyzed deprotection of the correspond-
ing N-acylglycinal diethyl acetals. The purity and identity of
the N-acylglycinal diethyl acetal intermediates and the desired
N-acylglycinals were confirmed by 1H and 13C NMR spectra ob-
tained on Varian iNova 400 MHz.
Materials and methods
Enzyme purification
Bovine liver ADH3 was purified following a procedure modified
from Pourmotabbed and Creighton [30]: ammonium sulfate pre-
cipitation (45–67%), DEAE cellulose chromatography, Mimetic Blue
II chromatography, and blue dextran agarose chromatography. The
Mimetic Blue II column replaced the NAD+-agarose chromatogra-
phy in the published procedure from Pourmotabbed and Creighton
[30]. Dialyzed sample containing partially purified ADH3 from the
Spectrophotometric assay for the production of NADH
Kinetic constants for short- and medium-chain NAEs were
determined from initial reaction rates obtained by following

M. Ivkovic et al. / Archives of Biochemistry and Biophysics 506 (2011) 157–164
159
NAD+ reduction to NADH as the increase in absorbance at 340 nm
ent of 50 mM sodium phosphate pH 6.0:CH3CN from 90:10 (v/v) to
95:5 (v/v) over 20 min, or isocratic mobile phase of 50 mM sodium
phosphate pH 6.0:CH3CN (90:10 v/v) and the analytes were de-
tected by monitoring the UV absorbance at 230 nm.
(D
e
340 = 6.22 Â 103 MÀ1 cmÀ1) using a JASCO V-530 UV–Vis spec-
trophotometer. Reaction conditions were as follows: 100 mM gly-
cine–NaOH pH 10, 2.5 mM NAD+, and 15–30
lg/ml ADH3 in 1.0 ml
total reaction volume. NAE concentrations were between 0.1 KM
and 10 KM, and all assays were conducted at 37 °C. Glycine was
purchased from J.T. Baker, NAD+ was acquired from BioWorld,
and cinnamyl alcohol and 1-octanol were from Sigma–Aldrich.
All were used without further purification. Rates were normalized
Cinnamyl alcohol, cinnamyl aldehyde, and cinnamic acid were
separated using a linear gradient of 50 mM sodium phosphate pH
6.0:CH3CN from 75:25 (v/v) to 40:60 (v/v) over 15 min and the
analytes were detected by monitoring the UV absorbance at
265 nm.
using rate for 250
l
M cinnamyl alcohol as the standard.
N-Benzoylethanolamine and N-benzoylglycinalsemicarbazone
were separated using a linear gradient of H2O:CH3CN from 70:30
(v/v) to 40:60 (v/v) over 20 min and the analytes were detected
by monitoring the UV absorbance at 210 nm.
MTS-formazan assay for the production of NADH
Due to the low solubility of long-chain NAEs, a complete range
of kinetic assays at high [NAE] to determine steady-state kinetic
parameters (KM and VMAX) could not be performed. As the com-
pounds could only be assayed at relatively low concentrations
Analysis by GC–MS
GC–MS analyses were performed using a Shimadzu QP5000
(<100
l
M), the rates of NADH production were very low and could
GC–MS equipped with a DB-5 (0.25
l
m  0.25 mm  30 m) col-
not be confidently determined spectrophotometrically. Instead, a
more sensitive method for NADH detection was employed, cou-
pling the NADH production to the reduction of a tetrazolium dye
MTS ([3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-
2-(4-sulfophenyl)-2H-tetrazolium, inner salt) to the colored for-
umn. Compounds were extracted from reaction mixture using
either ethyl ether (cinnamyl alcohol and cinnamyl aldehyde) or
CH2Cl2 (N-benzoylethanolamine, N-benzoylglycinal, and N-benzo-
ylglycine). Aldehyde derivatization to PFB-oximes was done by dis-
solving the extracted and dried residue into 90
ll of CH3CN, adding
mazan with kmax at 490 nm and
e
= 20,800 MÀ1 cmÀ1 at pH 9.5
10 l of 100 mM PFBHA (O-(2,3,4,5,6-pentafluorobenzyl)hydroxyl-
l
[36]. Phenazinemethosulfate (PMS) was used as an intermediate
electron carrier.
amine), and heating the solution at 60 °C for 60 min.
N-Benzoylglycine derivatization was done by dissolving the ex-
NAEs with 10 or more carbon atoms in the acyl chain were all
assayed for ADH3 activity at the same concentration employing
the MTS-formazan assay. The assay conditions were as follows:
tracted and dried material in 100 ll of BSTFA (N,O-bis(trimethyl-
silyl)trifluoroacetamide), purging the solution with N2, and
heating the mixture for 15–20 min at 90 °C. Derivatized samples
100 mM sodium pyrophosphate pH 9.5, 55
l
M NAE, 2.5 mM
g/
(5–10 ll) were injected into the GC–MS in a splitless manner, with
NAD+, 150
l
M MTS, 8.25 M PMS, 3.0% (v/v) DMSO, and 30
l
l
injection temperature at 250 °C. The temperature program was
modified adapted from Merkler et al. [9]. Oven temperature was
raised from 55 to 150 °C, at a rate of 40 °C/min, held at 150 °C for
3.6 min, then raised to 300 °C at a rate of 10.0 °C/min, and finally
held at 300 °C for 1.0 min. Interface temperature between GC and
MS was 280 °C, and solvent cut time was 7–9 min. Peak identity
was established by comparison of retention times and mass spec-
tra with those of derivatized standards and library spectra.
ml ADH3. Decanoyl and lauroylethanolamine were soluble enough
to allow the determination of KM and VMAX values as well. All reac-
tions were done at 37 °C and the initial rates were determined by
observing the increase in absorbance at 490 nm using a JASCO V-
530 UV–Vis spectrophotometer. Sodium pyrophosphate was ob-
tained from Fisher, MTS was purchased from Promega, and PMS
was obtained from TCI America. Rates for cinnamyl alcohol ob-
tained in this manner matched the rates obtained by following
NADH production at 340 nm. Rates were normalized against the
Trapping of the N-acylglycinals with semicarbazide
rate obtained for 250 lM cinnamyl alcohol as a standard.
In order to characterize the N-acylglycinal reaction products,
which are unstable at the reaction pH, semicarbazide was em-
ployed as an aldehyde-trapping reagent. Semicarbazide was added
in large excess (9–10 Â substrate concentration) to the reaction
mix containing 100 mM sodium pyrophosphate pH 9.5, 3–5 mM
N-benzoylethanolamine, 2.5 mM NAD+, and 0.25–1.0 mg/ml
ADH3. Formation of N-benzoylglycinal semicarbazone was fol-
lowed by HPLC.
Molecular modeling
The crystal structure of human ADH3 (PDB ID 1MP0) [37] was
used for grid-based ligand docking. All co-crystallized ligands
deemed superfluous for enzyme function were removed from the
crystal structure and polar hydrogen atoms were added using
AutoDockTools. Charges were then corrected for the requisite zinc
ions and bond orders corrected for the co-substrate, NAD+. The
receptor grid was prepared with a grid point spacing of 0.2 Å using
AutoGrid. The substrates of interest were then prepared using
AutoDockTools to define torsions, rotamers, and polar hydrogen
atoms. The ligands were then docked into the active site of ADH
using AutoDock 4.0 [38,39]. All default settings were utilized with
the exception of increasing the number of energy evaluations from
2.5 Â 104 to 2.5 Â 107.
N-Acylglycinal semicarbazone characterization by 13C NMR
N-Octanoylglycinal semicarbazone was characterized by fol-
lowing the enzymatic oxidation of [1,2-13C]-N-octanoylethanol-
amine in the presence of semicarbazide by 13C NMR. The
presence of N-octanoylglycinal semicarbazone was established by
observing the reduction in intensity of the 13C labeled signals in
the substrate and the appearance of two 13C carbon signals consis-
tent with those of N-octanoylglycinal semicarbazone. Peak identity
was confirmed by comparison with 13C spectrum of synthesized
N-octanoylglycinal semicarbazone. Reaction conditions were as
follows: 50 mM sodium pyrophosphate pH 8.0, 2.5 mM [1,2-13C]-
N-octanoylethanolamine, 2.5 mM NAD+, 20 mM semicarbazide–
HCl, 10% (v/v) D2O, and 1.7 mg/ml ADH3. Reaction mix was incu-
bated at 37 °C, and 13C NMR spectra were taken after 24 and
48 h reaction time points. Control samples containing all the
Product characterization by HPLC
Separations were performed on an Agilent HP 1100, equipped
with a 4-channel solvent mixing system, a quaternary pump, and
a variable wavelength UV detector and were accomplished using
a Thermo-Scientific C18 column (4.6 Â 250 mm) with the tempera-
ture fixed at 40 °C. N-Benzoylethanolamine, N-benzoylglycinal, and
N-benzoylglycine (hippurate) were separated using either a gradi-

160
M. Ivkovic et al. / Archives of Biochemistry and Biophysics 506 (2011) 157–164
Table 1
Kinetic constants for the ADH3-mediated oxidation of NAE substrates with acyl chains containing 2–12 carbon atoms.
Substrate
KM (mM)
VMAX
(l
mol/min/mg)
VMAX/KM (sÀ1 MÀ1
)
O
OH
R
N
H
N-Acetylethanolamine
N-Propionylethanolamine
N-Butyrylethanolamine
N-Hexanoylethanolamine
N-Octanoylethanolamine
N-Decanoylethanolamine
N-Lauroylethanolamine
1-Octanol
R@CH3
R@CH3CH2
R@CH3(CH2)2
R@CH3(CH2)4
R@CH3(CH2)6
R@CH3(CH2)8
R@CH3(CH2)10
CH3–(CH2)6–CH2–OH
C6H6HC@CHCH2OH
(4.5 0.4) Â 102
79 3.8
46 5.4
9.5 0.36
5.8 0.38
0.32 0.029
0.033 0.0061
0.050 0.0130
0.035 0.0033
1.9 0.059
1.9 0.026
1.7 0.052
1.6 0.018
1.5 0.028
0.57 0.024
0.14 0.0088
2.5 0.16
2.7
15
23
110
160
1.1 Â 103
2.7 Â 103
3.2 Â 104
7.2 Â 104
Cinnamyl alcohol
4.0 0.058
reagents except for the enzyme were handled and analyzed in the
same manner. 13C NMR Spectra were collected using a Varian
iNova 400 MHz instrument.
low reaction rates. A comparison of initial rates between two med-
ium-chain (N-decanoyl- and N-lauroylethanolamine) and several
long-chain NAEs at a single substrate concentration is shown in
Table S1 (Supplementary material). The initial rates of Table S1 fol-
low the general trend observed with short- and medium-chain
NAEs as the rates decrease with the increasing length of the acyl
chain (see Table 1).
Results and discussion
ADH3 purification
The data in (Tables 1 and S1) were generated with NAEs com-
prised only of straight-chain saturated or unsaturated acyl chains.
We also found that NAEs containing a terminal benzene ring were
ADH3 substrates (Table 2). The VMAX/KM values obtained for the
NAE substrates of this class included here are low relative to cin-
namyl alcohol and are of the same magnitude determined for the
aliphatic acyl chain NAEs with acyl chain lengths of 4–8 carbons
(compare Table 2 to Table 1).
ADH3 was purified from bovine liver to ꢀ90% homogeneity and
exhibited specific activity of 0.53 U/mg. Molecular weight of ADH3
subunits, calculated based on the Rf values determined from the
SDS–PAGE was 38 kDa, which is in good agreement with previ-
ously determined values of 40–41 kDa [40].
Kinetics of ADH3-catalyzed NAE oxidation
Molecular modeling of ADH3–NAE binding
A variety of aliphatic NAEs with acyl chain lengths varying from
2 to 20 carbons, as well as several aromatic NAEs were evaluated as
substrates for ADH3. All of the tested NAEs were found to be sub-
strates for ADH3, with KM values decreasing with increase in chain
length of the acyl substituent (Table 1). This indicates that increase
in substrate size leads to tighter binding to the enzyme, or rather
that small size impairs the ability to bind to the site tightly, which
is in agreement with known substrate preferences of ADH3 [41].
The values measured for the VMAX followed the same trend,
decreasing with increase in substrate size and hydrophobicity.
However, the decrease in KM values was more pronounced, result-
ing in overall increase in VMAX/KM values with increase in length of
the acyl chain length for the NAE substrates (Table 1). While ADH3
could catalyze the oxidation of all tested NAEs, their turnover val-
ues were low compared to cinnamyl alcohol, an ADH3 substrate
with a relatively high VMAX/KM value. Also included in Table 1 are
the steady-state kinetic values for the ADH3 mediated oxidation
of 1-octanol. 1-Octanol is a substrate used in the study of many
alcohol dehydrogenases and, thus, facilitaties the comparison of
the NAE kinetic data included here to earlier work.
NAE substrates with straight-chain, saturated acyl chains be-
tween 2 and 11 carbon atoms were docked into the human
ADH3 crystal structure using AutoDock 4.0 in order to evaluate
their relative binding energies. Sequence alignment of human
and bovine ADH3 using BLAST [42] shows that they share 94%
identity, with 98% of residues conserved, suggesting that the dock-
ing results obtained with the human enzyme are applicable to bo-
vine ADH3 as well. Relative binding energies of docked NAEs were
found to follow the same general trend as KM values, decreasing
with increase in chain length (Table 3), indicating that tighter bind-
ing to the enzyme is at least partly responsible for the observed KM
trend. Substrates were docked in the enzyme active site with hy-
droxyl group coordinating with the catalytic Zn(II), in the same
manner as other ADH3 substrates have been shown to bind
[40,43].
HPLC analysis of the ADH3-mediated oxidation of the NAEs
Kinetic parameters, KM and VMAX/KM, could not be determined
for NAE substrates with acyl chains longer than 12-carbon atoms
(longer than N-lauroylethanolamine) due to low solubility and
HPLC analysis of the ADH3-catalyzed oxidation of N-benzoy-
lethanolamine revealed the presence of peaks with retention times
consistent with N-benzoylglycine and N-benzoylglycinal (Fig. 1).
Table 2
Kinetic constants for the ADH3-mediated oxidation of NAE substrates with acyl chains containing an benzyl ring.
Substrate
KM (mM)
VMAX
(lmol/min/mg)
VMAX/KM (sÀ1 MÀ1
)
O
OH
R
N
H
N-Benzoylethanolamine
R@C6H6
R@C6H5CH2
R@C6H5CH2O
R@C6H5OCH2
C6H6HC@CHCH2OH
5.2 0.52
3.9 0.24
2.3 0.14
6.3 0.36
0.36 0.0098
1.2 0.026
0.81 0.014
0.25 0.0042
4.0 0.058
44
200
220
N-Phenylacetylethanolamine
N-Benzyloxycarbonylethanolamine
N-(2-Phenoxyacetyl)ethanolamine
Cinnamyl alcohol
25
0.035 0.0033
7.2 Â 104

M. Ivkovic et al. / Archives of Biochemistry and Biophysics 506 (2011) 157–164
161
Table 3
the products generated from the NAEs by ADH3. In order to test
the validity of this approach, we first used GC–MS to characterize
the cinnamyl alcohol reaction product. Derivatization of cinnamyl
alcohol reaction extract with PFBHA gave two peaks detected by
GC–MS, whose retention times and fragmentation patterns were
consistent with those for two isomers of cinnamyl aldehyde PFB-
oxime (Fig. 3), confirming that the reaction product is cinnamyl
aldehyde. However, applying the same method to characterize N-
benzoylglycinal observed by HPLC proved unsuccessful. Although
the derivatized standard for N-benzoylglycinal-PFB-oxime was
successfully characterized using GC–MS, this compound could
not be detected by GC–MS in the reaction extracts. This can be
attributed to the very low turnover rates of N-benzoylethanol-
amine oxidation compared to cinnamyl alcohol oxidation (Table 2),
as well as to the tendency of N-benzoylglycinal to undergo
non-enzymatic oxidation to N-benzoylglycine under the reaction
conditions. Derivatization of the N-benzoylethanolamine reaction
extract with BSTFA resulted in detection of N,O-di-TMS-N-benzoyl-
glycine, confirming that the peak observed by HPLC is
N-benzoylglycine.
Comparison of the KM values to the calculated free energy of binding for selected
NAEs and cinnamyl alcohol.
Substrate
KM (mM)
Calculated free energy of
bindinga (kcal/mol)
N-Acetylethanolamine
N-Propionylethanolamine
N-Butyrlethanolamine
N-Hexanoylethanolamine
N-Octanoylethanolamine
N-Decanoylethanolamine
N-Lauroylethanolamine
N-Benzoylethanolamine
Cinnamyl alcohol
(4.5 0.4) Â 102
79 3.8
46 5.4
9.5 0.36
5.8 0.38
0.32 0.029
0.033 0.0061
5.2 0.52
À3.72
À4.14
À4.22
À4.34
À4.31
À4.34
À4.06
À4.58
À4.68
0.035 0.0033
a
Binding energies values were calculated using Autodock 4.0, and have a stan-
dard error of about 2.5 kcal/mol [39].
N-Benzoylglycine and N-benzoylglycinal are the two possible
products of the enzymatic oxidation of N-benzoylethanolamine.
However, Svensson et al. [44] found that ADH3 does not catalyze
aldehyde dismutation suggesting that N-benzoylglycinal should
be produced upon the ADH3-mediated oxidation of N-benzoyleth-
anolamine. Incubation of N-benzoylglycinal under the conditions
used for ADH3 catalyzed reactions (2.5 mM NAD+, pH 9.5–10) in
the absence of ADH3 showed low rates of non-enzymatic aldehyde
dismutation to N-benzoylglycine and N-benzoylethanolamine
(data not shown). Incubation of ADH3 with N-benzoylglycinal un-
der identical conditions does not significantly increase the rate of
aldehyde dismutation (Supplementary material, Fig. S1). HPLC
analysis of ADH3-catalyzed oxidation of cinnamyl alcohol, a sub-
strate with a VMAX/KM that is ꢀ1600 higher than that of N-benzoy-
lethanolamine, yields only a peak with a retention time consistent
with cinnamyl aldehyde (Fig. 2). These data further suggest that
the ADH3-catalyzed products of NAE oxidation are the N-acylglyc-
inals, and that the observed N-acylglycines are produced via slow
non-enzymatic aldehyde dismutation.
13C NMR characterization of the products generated by ADH3
In order to detect and characterize the unstable N-acylglycinal
product, a well-known aldehyde-trapping reagent, semicarbazide,
was added to the reaction mixture. Semicarbazide readily reacts
with aldehydes and ketones to form semicarbazones. HPLC analy-
sis of N-benzoylethanolamine reaction containing semicarbazide
revealed a peak consistent with the peak for N-benzoylglycinal
semicarbazone standard (Fig. 4). Note that the peak consistent
with N-benzoylglycine is not seen in the chromatogram. Further
characterization of the semicarbazone derivative was attempted
by LC–MS and GC–MS, but proved unsuccessful.
To better characterize the putative N-acylglycinal semicarba-
zone species detected by HPLC, the ADH3 reaction was analyzed
by 13C NMR for which the substrate was [1,2-13C]-N-octanoyletha-
nolamine. ADH3 treatment of [1,2-13C]-N-octanoylethanolamine
reaction showed a total of 4 different 13C signals – two that
matched those for the substrate, at 41 and 60 ppm, and two new
13C signals, at 40 and 143 ppm (Fig. 5). The chemical shifts of the
two new signals match those of the corresponding carbon atoms
in the N-octanoylglycinal semicarbazone standard, providing
strong confirmation that the product of the ADH3 catalyzed
GC–MS characterization of the products generated by ADH3
Our HPLC results strongly suggest that products of the ADH3-
mediated oxidation of the NAEs are the corresponding N-acylglyc-
inals. However, comparison of HPLC retention times is not defini-
tive as there is always the possibility that another compound has
a similar retention time. We used GC–MS to further characterize
14.21
2.42 – NAD+/NADH
120
O
N-Benzoylethanolamine
N
CHO
O
H
100
NAD+ Derivative
OH
N
H
3.92
N-Benzoylglycinal
80
2.53
60
5.40 – N-Benzoylglycine
O
40
20
0
N
COOH
H
12.85
0
5.0
7.5
10.0
12.5
17.5
20.0
2.5
15.0
Time (minutes)
Fig. 1. HPLC analysis of the ADH3-catalyzed oxidation of N-benzoylethanolamine. Reaction conditions were as follows: 100 mM sodium pyrophosphate pH 9.5, 2.5 mM NAD+,
6.0 mM N-benzoylethanolamine, and 2.4 mg/ml ADH3. HPLC analysis was performed after 36 h of reaction time. The peak with a retention time of 3.918 min was derived
from NAD+ that formed during prolonged incubation at high pH. A peak with the same retention time was also observed in the control sample containing only NAD+ and
buffer.

162
M. Ivkovic et al. / Archives of Biochemistry and Biophysics 506 (2011) 157–164
3000
NAD+/NAD
O
OH
H
2000
1000
0
Cinnamyl
alcohol
Cinnamyl
aldehyde
0
2
6
8
10
12
14
16
18
4
Time (minutes)
Fig. 2. HPLC analysis of the ADH3-catalyzed oxidation of cinnamyl alcohol. Reaction conditions were as follows: 100 mM sodium pyrophosphate pH 9.5, 2.5 mM NAD+,
1.75 mM cinnamyl alcohol, and 5.0 g/ml ADH3. HPLC analysis was performed after 45 min of reaction time.
l
oxidation is N-octanoylglycinal. Repeating the experiment in the
absence of semicarbazide resulted in appearance of 13C signals
consistent with those of N-octanoylglycine. This result was consis-
tent with our HPLC experiments, which showed that N-benzoyl-
glycinal is non-enzymatically oxidized to N-benzoylglycine at
basic pH. Control sample with no ADH3 present showed only the
two substrate peaks, confirming there was no detectable oxidation
of N-octanoylethanolamine through non-enzymatic chemistry.
[44,45], ADH3 does not catalyze aldehyde dismutation and, thus,
does not catalyze the conversion of N-acylglycinals and cinnamyl
aldehyde to the corresponding carboxylate. This result agrees with
earlier data from Svensson et al. [44] that human ADH3 does not
convert butanal to butyric acid. The relatively low VMAX/KM values
measured for the ADH3-catalyzed oxidation of the NAEs suggests
that it is unlikely that this enzyme plays a significant role in con-
version of NAEs to N-acylglycinals in vivo. Nevertheless, our find-
ings coupled with the work of Aneetha et al. [27] showing that
ADH4 is capable of catalyzing the conversion of anandamide to
N-arachidonoylglycinal and N-arachidonoylglycine, validate the
dehydrogenase-mediated oxidation of the NAEs is a plausible path-
way to the NAGs and PFAMs in vivo. Recent work from Bradshaw
et al. [21] does show that anandamide is a precursor to N-arachido-
noylglycine in RAW 264.7 and C6 glioma cells. The ADH4-mediated
Conclusions
The data presented here demonstrate the N-fatty acylethanola-
mines can serve as substrates for ADH3 being oxidized to the N-
fatty acylglycinals. In contrast to other alcohol dehydrogenases
Cinnamylaldehyde O-PFB-oximes
(Z-and E-isomers)
12.5
13.0
13.5
14.0
14.5
15.0
15.5
Time (minutes)
181
700,000
500,000
91
77
77
130
181
F
F
91
N
F
O
130
115
105
51
F
F
300,000
100,000
31
326
63
161
146
196
250
230
10
40
70
100 130 160 190 220 250 280 310 340
m/z
Fig. 3. GC–MS analysis of the ADH3-generated cinnamyl aldehyde. Cinnamyl aldehyde was first generated enzymatically by treating cinnamyl alcohol with ADH3 as
described in the legend to Fig. 2. Cinnamyl aldehyde was converted to the perfluorobenzyl oxime (O-PFB-oxime) with PFBHA and analyzed by GC–MS. The resulting cinnamyl
aldehyde O-PFB-oxime exists as the E- and Z-isomers, which have different retention times in the gas chromatograph (upper panel), but exhibit identical fragmentation
patterns from the mass spectrometer (lower panel). The structure and fragmentation pattern of cinnamaldehyde O-perfluorobenzyl oxime are shown in the inset to the lower
panel.

M. Ivkovic et al. / Archives of Biochemistry and Biophysics 506 (2011) 157–164
163
1000
NAD+/NADH
800
600
400
200
N-Benzoylethanolamine
O
O
N
N
N
H
NH
H
N-Benzoylglycinal
NAD+ derivative
semicarbazone
0
0
2.5
5
7.5
10
12.5
15
17.5
Time (minutes)
Fig. 4. HPLC analysis of the ADH3-catalyzed oxidation of N-benzoylethanolamine in the presence of semicarbazide. Reaction conditions were as follows: 100 mM sodium
pyrophosphate pH 9.5, 2.5 mM NAD+, 4.0 mM N-benzoylethanolamine and 0.3 mg/ml ADH3. HPLC analysis was performed after 24 h of reaction time.
authors dedicate this work to the memory of Dr. Mitchell E.
Johnson.
60.05
10000
A
B
O
C
13
NH CH
2
13
CH OH
2
60.43
CH -(CH
3
)
2
6
+
NAD
NADH
B
8000
6000
4000
ADH3
Appendix A. Supplementary data
O
C
13
13
A
CH -(CH
)
NH CH
CHO
3
2
6
2
O
C
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.abb.2010.12.002.
H
N-NH
NH
2
2
A1
B1
O
C
O
13
13
CH
CH -(CH
)
NH CH
N
NH C
NH
2
3
2
6
2
2000
0
References
B1
A2
[1] J.L.W. Thudichum,
Baillière, Tindall, and Cox, London, 1884.
[2] P.A. Levene, J. Biol. Chem. 24 (1916) 69–89.
[3] K.A. Kuehl Jr., T.A. Jacob, H.O. Ganley, R.E. Ormond, M.A.P. Meisinger, J. Am.
Chem. Soc. 79 (1957) 5577–5578.
A Treatise on the Chemical Constitution of the Brain,
150
130
110
90
70
50
13C-Chemical Shift (ppm)
Fig. 5. 13C NMR spectrum of ADH3-catalyzed oxidation of [1,2-13C]-N-octanoyleth-
anolamine in the presence of semicarbazide. Reaction conditions were as follows:
50 mM sodium pyrophosphate pH 8.0, 2.5 mM [1,2-13C]-N-octanoylethanolamine,
2.5 mM NAD+, 20 mM semicarbazide HCl, 10% (v/v) D2O, and 1.7 mg/ml ADH3. The
13C NMR spectrum was collected after reaction proceeded for 48 h.
[4] S.L. Palmer, G.A. Thakur, A. Makriyannis, Chem. Phys. Lipids 121 (2003) 3–19.
[5] H.H.O. Schmid, E.V. Berdyshev, Prostaglandins Leukot. Essent. Fatty Acids 66
(2002) 363–376.
[6] V. Di Marzo, L. De Petrocellis, T. Bisogno, Handb. Exp. Pharmacol. 168 (2005)
147–185.
[7] E.K. Farrell, D.J. Merkler, Drug Discov. Today 13 (2008) 558–568.
[8] D.J. Merkler, K.A. Merkler, W. Stern, F.F. Fleming, Arch. Biochem. Biophys. 330
(1996) 430–434.
[9] D.J. Merkler, G.H. Chew, A.J. Gee, K.A. Merkler, J.-P.O. Sorondo, M.E. Johnson,
Biochemistry 43 (2004) 12667–12674.
[10] B.J. Wilcox, K.J. Ritenour-Rodgers, A.S. Asser, L.E. Baumgart, M.A. Baumgart,
D.L. Boger, J.L. DeBlassio, M.A. deLong, U. Glufke, M.E. Henz, L. King III, K.A.
Merkler, J.E. Patterson, J.J. Robleski, J.C. Vederas, D.J. Merkler, Biochemistry 38
(1999) 3235–3245.
[11] T. Bisogno, N. Sepe, L. De Petrocellis, R. Mechoulam, V. Di Marzo, Biochem.
Biophys. Res. Commun. 239 (1997) 473–479.
[12] W.J. Driscoll, S. Chaturvedi, G.P. Mueller, J. Biol. Chem. 282 (2007) 22353–
22363.
[13] J. Caldwell, J.R. Idle, R.L. Smith, The amino acid conjugations, in: T.E. Gram
(Ed.), Extrahepatic Metabolism of Drugs and Other Compounds, SP Medical
and Scientific Books, New York, 1980, pp. 453–492.
[14] P.T. Ozand, G.G. Gascon, J. Child. Neurol. 6 (1991) 288–303.
[15] L. Bonafé, H. Troxler, T. Kuster, C.W. Heizmann, N.A. Chamoles, A.B. Burlina, N.
Blau, Mol. Genet. Metab. 69 (2000) 302–311.
[16] S.M. Bonham Carter, J.M. Midgley, D.G. Watson, R.W. Logan, J. Pharm. Biomed.
Anal. 9 (1991) 969–975.
[17] P. Rinaldo, E. Schmidt-Sommerfeld, A.P. Posaca, S.J.R. Heales, D.A. Woolf, J.V.
Leonard, J. Pediatr. 122 (1993) 580–584.
oxidation of anandamide is relatively slow hinting that another en-
zyme may catalyze NAE oxidation in vivo. In this regard, further
studies on ADH5 and ADH6 assessing the possibility of their
involvement in NAE metabolism could be revealing, as these two
alcohol dehydrogenases are poorly investigated to date. The possi-
bility that fatty aldehyde dehydrogenases could function in the
oxidation of N-fatty acylglycinals to N-fatty acylglycines also war-
rants further investigation.
Acknowledgments
This work was supported, in part, by grants from the National
Institutes of Health – General Medical Sciences (R15-GM059050
and R15-GM073659), the Florida Center for Excellence for Biomo-
lecular Identification and Targeted Therapeutics (FCoE-BITT Grant
No. GALS020), the Gustavus and Louise Pfieffer Research Founda-
tion, the Eppley Foundation for Research, the Shin Foundation for
Medical Research, and the Shirley W. & William L. Griffin Founda-
tion to D.J.M., a predoctoral fellowship to E.W.L. from the American
Heart Association (#0415259B), and a travel grant from the
University of South Florida Student Government to M.I. The
[18] S.M. Huang, T. Bisgno, T.J. Petros, S.Y. Chang, P.A. Zavitsanos, R.E. Zipkin, R.
Sivakumar, A. Coop, D.Y. Maeda, L. De Petrocellis, S. Burstein, V. Di Marzo, J.M.
Walker, J. Biol. Chem. 276 (2001) 42639–42644.
[19] N. Rimmerman, H.B. Bradshaw, H.V. Hughes, J.S.-C. Chen, S.S.-J. Hu, D. McHugh,
E. Vefring, J.A. Jahnsen, E.L. Thompson, K. Masuda, B.F. Cravatt, S. Burstein, M.R.
Vasko, A.L. Prieto, D.K. O’Dell, J.M. Walker, Mol. Pharmacol. 74 (2008) 213–224.
[20] H.B. Bradshaw, N. Rimmerman, S.S.-J. Hu, S. Burstein, J.M. Walker, Vitam.
Horm. 81 (2009) 191–205.

164
M. Ivkovic et al. / Archives of Biochemistry and Biophysics 506 (2011) 157–164
[21] H.B. Bradshaw, N. Rimmerman, S.S.-J. Hu, V.M. Benton, J.M. Stuart, K. Masuda,
B.F. Cravatt, D.K. O’Dell, J.M. Walker, BMC Biochem. 10 (2009) 14.
[22] M. Kelley, D.A. Vessey, J. Biochem. Toxicol. 9 (1994) 153–158.
[23] D.P. Waluk, N. Schultz, M.C. Hunt, FASEB J. 24 (2010) 2795–2803.
[24] J. O’Byrne, M.C. Hunt, D.K. Rai, M. Saeki, S.E. Alexson, J. Biol. Chem. 278 (2003)
34237–34244.
[25] J.M. McCue, W.J. Driscoll, G.P. Mueller, Biochem. Biophys. Res. Commun. 365
(2008) 322–327.
[26] B. Tan, D.K. O’Dell, Y.W. Yu, M.F. Monn, H.V. Hughes, S. Burstein, J.M. Walker, J.
Lipid Res. 51 (2010) 112–119.
[27] H. Aneetha, D.K. O’Dell, B. Tan, J.M. Walker, T.D. Hurley, Bioorg. Med. Chem.
Lett. 19 (2009) 237–241.
[28] S.H. Burstein, R.G. Rossetti, B. Yagen, R.B. Zurier, Prostagladins Other Lipid
Mediat. 61 (2000) 29–41.
[29] O. Danielsson, S. Atrian, T. Luque, L. Hjelmqvist, R. Gonzàlez-Duarte, H.
Jörnvall, Proc. Natl. Acad. Sci. USA 91 (1994) 4980–4984.
[30] T. Pourmotabbed, D.J. Creighton, J. Biol. Chem. 261 (1986) 14240–14244.
[31] X. Parés, B.L. Vallee, Biochem. Biophys. Res. Commun. 98 (1981) 122–130.
[32] M.M. Bradford, Anal. Biochem. 72 (1976) 248–254.
[33] K.-O. Jonsson, S. Vandevoorde, D.M. Lambert, G. Tiger, C.J. Fowler, Br. J.
Pharmacol. 133 (2001) 1263–1275.
[35] E.V. Brown, in: Penilloaldehydes and penaldic acids, in: H.T. Clarke, J.R.
Johnson, R. Johnson (Eds.), The Chemistry of Penicillin, Princeton University
Press, Princeton, NJ, 1949, pp. 473–534.
[36] J.A. Barltrop, T.C. Owen, A.H. Cory, J.G. Cory, Bioorg. Med. Chem. Lett. 1 (1991)
611–614.
[37] P.C. Sanghani, H. Robinson, R. Bennett-Lovsey, T.D. Hurley, W.F. Bosron, Chem.
Biol. Interact. 143–144 (2003) 195–200.
[38] G.M. Morris, D.S. Goodsell, R.S. Halliday, R. Huey, W.E. Hart, R.K. Belew, A.J.
Olson, J. Comput. Chem. 19 (1998) 1639–1662.
[39] R. Huey, G.M. Morris, A.J. Olson, D.S. Goodsell, J. Comput. Chem. 28 (2007)
1145–1152.
[40] P.C. Sanghani, H. Robinson, W.F. Bosron, T.D. Hurley, Biochemistry 41 (2002)
10778–10786.
[41] F.W. Wagner, X. Parés, B. Holmquist, B.L. Vallee, Biochemistry 23 (1984) 2193–
2199.
[42] S.F. Altschul, T.L. Madden, A.A. Schäffer, J. Zhang, Z. Zhang, W. Miller, D.J.
Lipman, Nucleic Acids Res. 25 (1997) 3389–3402.
[43] P.C. Sanghani, W.F. Bosron, T.D. Hurley, Biochemistry 41 (2002) 15189–
15194.
[44] S. Svensson, A. Lundsjö, T. Cronholm, J.-O. Höög, FEBS Lett. 394 (1996) 217–
220.
[34] D.M. Lambert, F.G. DiPaolo, P. Sonveaux, M. Kanyonyo, S.J. Govaerts, E.
Hermans, J.-L. Bueb, N.M. Delzenne, E.J. Tschirhart, Biochim. Biophys. Acta
1440 (1999) 266–274.
[45] G.T.M. Henehan, N.J. Oppenheimer, Biochemistry 32 (1993) 735–
738.