Identification and characterization of oxymetazoline glucuronidation in human liver microsomes: Evidence for the involvement of UGT1A9

Article abstract of DOI:10.1002/jps.22303

The incubation of oxymetazoline, a nonprescription nasal decongestant, with human liver microsomes (HLMs) supplemented with uridine-5-diphosphoglucuronic acid (UDPGA) generated glucuronide metabolite as observed by LC/MS/MS. The uridine glucuronosyltransferases (UGTs) responsible for the O-glucuronidation of oxymetazoline remain thus far unidentified. The glucuronide formed in HLMs was identified by LC/MS/MS and characterized by one- and two-dimensional NMR to be the β-O-glucuronide of oxymetazoline. UGT screening with expressed UGTs identified UGT1A9 as the single UGT isoform catalyzing O-glucuronidation of oxymetazoline. Oxymetazoline O-glucuronidation by using HLMs was best fitted to the allosteric sigmoidal model. The derived S₅₀ and V_max values were 2.42 ± 0.40 mM and 8.69 ± 0.58 pmole/(min mg of protein), respectively, and maximum clearance (CL_max) was 3.61 L/min/mg. Oxymetazoline O-glucuronidation by using expressed UGT1A9 was best fitted to the substrate inhibition model. The derived K_m and V_max values were 2.53 ± 1.03 mM and 54.18 ± 16.92 pmole/(min mg of protein), respectively, and intrinsic clearance (CL_int) was 21.41 L/(min mg). Our studies indicate that oxymetazoline is not glucuronidated at its nanomolar intranasal dose and thus is eliminated unchanged, because UGT1A9 would only contribute to its elimination at the toxic plasma concentrations.

Full text of DOI:10.1002/jps.22303

Identification and Characterization of Oxymetazoline
Glucuronidation in Human Liver Microsomes: Evidence
for the Involvement of UGT1A9
MUKESH K. MAHAJAN,1,2 VINITA UTTAMSINGH,3 LIANG-SHANG GAN,4 BARBARA LEDUC,2 DAVID A. WILLIAMS2
1Drug Metabolism and Pharmacokinetics, Immunoinflammation Centre of Excellence in Drug Discovery, GlaxoSmithKline,
Collegeville, Pennsylvania
2Department of Pharmaceutical Sciences, Massachusetts College of Pharmacy and Health Sciences, Boston, Massachusetts
3Drug Metabolism and Pharmacokinetics, Concert Pharmaceuticals, Inc., Lexington, Massachusetts
4Drug Metabolism and Pharmacokinetics, Biogen Idec., Inc., Cambridge, Massachusetts
Received 12 April 2010; revised 4 June 2010; accepted 24 June 2010
Published online 28 July 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.22303
ABSTRACT: The incubation of oxymetazoline, a nonprescription nasal decongestant, with
human liver microsomes (HLMs) supplemented with uridine-5-diphosphoglucuronic acid
(UDPGA) generated glucuronide metabolite as observed by LC/MS/MS. The uridine glucurono-
syltransferases (UGTs) responsible for the O-glucuronidation of oxymetazoline remain thus
far unidentified. The glucuronide formed in HLMs was identified by LC/MS/MS and charac-
terized by one- and two-dimensional NMR to be the $-O-glucuronide of oxymetazoline. UGT
screening with expressed UGTs identified UGT1A9 as the single UGT isoform catalyzing O-
glucuronidation of oxymetazoline. Oxymetazoline O-glucuronidation by using HLMs was best
fitted to the allosteric sigmoidal model. The derived S50 and Vmax values were 2.42 0.40 mM
and 8.69 0.58 pmole/(min mg of protein), respectively, and maximum clearance (CLmax) was
3.61 L/min/mg. Oxymetazoline O-glucuronidation by using expressed UGT1A9 was best fit-
ted to the substrate inhibition model. The derived Km and Vmax values were 2.53 1.03 mM
and 54.18
16.92 pmole/(min mg of protein), respectively, and intrinsic clearance (CLint) was
21.41 L/(min mg). Our studies indicate that oxymetazoline is not glucuronidated at its nanomo-
lar intranasal dose and thus is eliminated unchanged, because UGT1A9 would only contribute
to its elimination at the toxic plasma concentrations. © 2010 Wiley-Liss, Inc. and the American
Pharmacists Association J Pharm Sci 100:784–793, 2011
Keywords: oxymetazoline; glucuronosyltransferases (UGTs); enzyme kinetics; phase
II metabolism; microsomes; glucuronidation; UGT1A9; liquid chromatography; mass
spectrometry
consumer because they are freely available for
self-medication without a doctor’s prescription. Ex-
cessive use of oxymetazoline nasal products has con-
tributed to rebound congestion and to cardiovascular
and central nervous system adverse events,1–3 which
are indicative of systemic absorption. Although sys-
temic effects have been minimal with recommended
intranasal doses of oxymetazoline (daily intranasal
dose of 100 :g), a potential risk of systemic absorp-
tion and adverse events increases with excessive
or overdoses. The estimated incidence of adverse
events is very low: 0.14 adverse events per 100,000
patient-years.4 Although no human metabolism,
INTRODUCTION
Oxymetazoline (Fig. 1) is a long-acting vasocon-
strictor when applied directly on nasal membranes.
It has been available as a nonprescription drug in
the United States for more than 40 years, having
been approved for the relief of nasal congestion from
common colds and allergic rhinitis. Nonprescription
drug products are often perceived as safe by the
Correspondence to: Mukesh K. Mahajan (Telephone: +610-917-
5856; Fax: +610-917-4178; E-mail: mukesh.k.mahajan@gsk.com)
Journal of Pharmaceutical Sciences, Vol. 100, 784–793 (2011)
© 2010 Wiley-Liss, Inc. and the American Pharmacists Association
784
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011

CHARACTERIZATION OF OXYMETAZOLINE GLUCURONIDATION IN HUMAN LIVER MICROSOMES
785
The objective of the current study was to iden-
tify and characterize the glucuronide metabolite of
oxymetazoline, to identify the specific UGT isoforms
involved in the in vitro glucuronidation of oxymetazo-
line by screening with expressed human UGTs, and to
determine the kinetics of oxymetazoline glucuronida-
tion in pooled HLMs and expressed human UGTs. The
oxymetazoline glucuronide metabolite was character-
ized using biosynthetic oxymetazoline glucuronide by
one-dimensional (1D) and two-dimensional (2D) NMR
as the $-O-glucuronide. UGT1A9 was identified as the
only UGT isoform catalyzing the $-O-glucuronidation
of oxymetazoline. The enzyme kinetic experiments
were performed by using pooled HLMs and expressed
UGT1A9, and their corresponding kinetic parameters
were determined.
Figure 1. Structure of oxymetazoline.
pharmacokinetic, or toxicological data have been re-
ported thus far for oxymetazoline, the phenolic
structure
of
oxymetazoline
suggests
O-
MATERIALS AND METHODS
Materials
glucuronidation to be its probable detoxification
metabolic pathway. What role glucuronide detoxifica-
tion plays in its low incidence of adverse events and
its elimination remains untested. The only intranasal
pharmacokinetic study reported for oxymetazoline is
the intranasal bioavailability of approximately 7% at
40 :g per rat.5
The human uridine glucuronosyltransferases
(UGTs) are membrane proteins of the endoplasmic
reticulum that are expressed primarily in the liver
and are also expressed in extrahepatic tissues, in-
cluding intestine, lung, kidneys, nasal olfactory ep-
ithelium, breast, placenta, testes, and prostate.6 On
the basis of the evolutionary divergence, 28 human
UGT genes have been identified and divided into
three subfamilies, 1A, 2A, and 2B7; however, only
16 of the human UGT genes have been expressed:
UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A7,
UGT1A8, UGT1A9, UGT1A10, UGT2A1, UGT2A2,
UGT2A3, UGT2B4, UGT2B7, UGT2B15, UGT2B17,
and UGT2B28.8
The identification of specific UGT isoform respon-
sible for metabolism of drugs is critical where glu-
curonidation has been established as a metabolic
pathway. The evaluation of human drug metabolism
has been explored by the use of microsomal prepa-
rations from human tissue; however, unlike P450 iso-
forms, the identification of the UGT isoform(s) respon-
sible for glucuronidation of xenobiotics is complicated
by overlapping substrate specificity and the lack of
specific UGT inhibitors.9 Emergence of human re-
combinant cDNA-expressed UGT isoforms expressed
in baculovirus-infected insect cells) has made it pos-
sible to determine the UGT isoform involved in the
glucuronidation of xenobiotics with relatively ease.
Therefore, these expressed human UGTs have led to
the identification of the UGT isoforms involved in the
glucuronidation of a number of phenolic, amino, and
carboxylic acid xenobiotics.10–13
Oxymetazoline, alamethicin, UDPGA (uridine-5-
diphosphoglucuronic acid), dimethyl sulfoxide, am-
monium formate, monobasic potassium phosphate,
and dibasic potassium phosphate were all purchased
from Sigma-Aldrich (St. Louis, Missouri). Acetoni-
trile (HPLC grade) and methanol (HPLC grade)
were obtained from Mallinckrodt Baker (Phillipsburg,
New Jersey). Magnesium chloride was purchased
from Amresco (Solon, Ohio). All other chemicals
and reagents were obtained at the highest purity
available.
Mixed-gender pooled HLMs (n = 50) were obtained
from XenoTech, LLC (Kansas City, Kansas). The hu-
man recombinant cDNA expressed SupersomesTM
(UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A8,
UGT1A9, UGT1A10, UGT2B4, UGT2B7, UGT2B15,
and UGT2B17) were purchased from BD Gentest
(Woburn, Massachusetts). HLMs and expressed UGT
enzymes were stored at −80◦C and used shortly af-
ter receiving them according to the supplier’s in-
structions. The expressed UGT2A enzymes, as well
as UGT1A5, UGT1A7, UGT2B10, UGT2B11, and
UGT2B28, were not available from BD Gentest for
testing. The protein content of all SupersomesTM was
5 mg/mL.
Oxymetazoline Glucuronidation by Pooled HLMs
A 0.1 mM of potassium phosphate buffer solution
(pH 7.4) (837.5 :L), containing 3 mM of magne-
sium chloride and 50 :L of 1 mg of protein/mL of
HLMs, was pretreated with 2.5 :L of alamethicin in
methanol (50 :g/mg of protein) on ice for 15 min
before adding 10 :L of oxymetazoline (50 :M). The
incubation tubes were preincubated for 3 min at 37◦C
in a reciprocal shak ing bath to which 100 :L of
UDPGA (5 mM) was added to initiate the reaction.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011

786
MAHAJAN ET AL.
The reactions were incubated at 37◦C in a shaking wa-
ter bath under ambient oxygenation conditions. The
final volume of each incubation was 1000 :L. The
reactions were incubated for 60 min to ensure suf-
ficient formation of the glucuronide metabolite. The
reactions were terminated at 60 min by the addi-
tion of 2 volumes of room temperature acetonitrile
to precipitate the protein, and the resulting mixture
was chilled at 4◦C for 30 min followed by centrifu-
gation at 3000 × g to pelletize the protein. The su-
pernatant was collected and dried under a stream of
SupersomesTM (0.5 mg of protein/mL) served as the
negative control.
Biosynthesis of Oxymetazoline Glucuronide
The oxymetazoline glucuronide was biosynthesized
by adding 1250 :L of rabbit liver microsomes (5 mg of
protein/mL) and 62.5 :L of alamethicin in methanol
(50 :g/mg of protein) to 3137.5 :L of 0.1 mM of potas-
sium phosphate buffer (pH 7.4) containing 3 mM of
magnesium chloride in reaction tubes. The tubes were
kept on ice for 15 min to which 50 :L of oxymetazo-
line (100 :M) was added and preincubated for 3 min
at 37◦C in a reciprocating shaking bath. The final
volume of each incubation was 5 mL. The reactions
were initiated by the addition of 500 :L of UDPGA
(10 mM), incubated at 37◦C for 3 h, and terminated
after 3 h by precipitating the protein by the addi-
tion of 10 mL of acetonitrile to each tube. The tubes
were refrigerated at 4◦C for 30 min and centrifuged
at 3000 × g for 10 min to pelletize the protein; the
supernatant was collected and dried overnight using
R
nitrogen in a TurboVapꢀ LV drier (Zymark, Hopkin-
ton, Massachusetts). The dried residues in the tubes
were reconstituted in 200 :L of mobile phase (85:15 v/
v, 10 mM of ammonium formate, pH 4.0:acetonitrile)
in preparation for LC/MS/MS analysis. Incubations of
oxymetazoline without UDPGA, drug, or protein were
run simultaneously as controls.
An ion trap LC/MS/MS system consisting of an Agi-
lent 1100 HPLC (Agilent Technologies, Palo Alto, Cal-
ifornia) coupled to a Symmetry C18 column (5 :m,
2.1 × 150 mm; Waters Corporation, Milford, Mas-
sachusetts) and Finnigan LCQ Deca XP Plus ion
trap mass spectrometer (Thermo Fisher Scientific,
Inc., Waltham, Massachusetts) was utilized to iden-
tify oxymetazoline glucuronide metabolite. The elu-
ant from HPLC column was introduced directly into
the mass spectrometer via electrospray ionization in
the positive ion mode. The gradient mobile phase con-
sisted of 10 mM of ammonium formate (pH 4.0) as
solvent A; solvent B was acetonitrile. The initial mo-
bile phase was 85:15 A/B (v/v) and by linear gradient
transitioned to 20:80 A/B (v/v) over 30 min at a flow
rate of 0.400 mL/min. Twenty microliters of the recon-
stituted samples was injected into the equilibrated
HPLC. Ionization was assisted with sheath and aux-
iliary gas (nitrogen) set at 60 and 40 psi, respectively.
The electrospray voltage was set at 5 kV, with the
heated ion transfer capillary set at 300◦C and 30 V.
Relative collision energies of 25% to 30% were used
when operating in the MS/MS mode of the ion trap.
R
the SpeedVacꢀ (Thermo Fisher Scientific Inc.). The
gummy residue remaining in each tube was dissolved
in 2 mL of mobile phase and sonicated for 2 min, af-
ter which the tubes were kept on ice for 30 min. The
solution was recentrifuged at 16,000 × g for 10 min
to pelletize any remaining protein. The supernatant
R
was collected, concentrated using the SpeedVacꢀ, and
analyzed for oxymetazoline glucuronide by using the
LC/MS/MS method as previously described.
One hundred microliters of injections were made
using a Symmetry C18 column (5 :m, 2.1 × 150 mm)
with the mobile phase and gradient conditions as de-
scribed previously. The fractions between 4 and 5 min
contained the glucuronide and were collected, pooled,
R
and dried overnight using the SpeedVacꢀ. The dried
residue was reconstituted in 2 mL of mobile phase
and further purified and collected into semipurified
fractions by repeated 100-:L LC injections using the
same column and mobile phase. Liquid–liquid extrac-
tion was then performed using an equal volume of
ethyl acetate and the pooled semipurified glucuronide
fraction. The glucuronide was back-extracted into the
10 mM of ammonium formate phase (pH 4.0) and all
of the other impurities remained in the ethyl acetate
phase. Next, solid-phase extraction was performed us-
Oxymetazoline Glucuronidation by Expressed Human
UGT Isoforms
Expressed UGT1A1, UGT1A3, UGT1A4, UGT1A6,
UGT1A8, UGT1A9, UGT1A10, UGT2B4, UGT2B7,
UGT2B15, and UGT2B17 (0.5 mg of protein/mL) in-
cubations were performed and processed in a manner
similar to that described for HLMs except that the
final volume of each incubation was 500 :L. The in-
cubations were carried out for 30 min to ensure suf-
ficient formation of the glucuronide metabolite, and
dried residues were reconstituted in 100 :L of mo-
bile phase. The incubations were performed in dupli-
cate, and the samples were analyzed by the described
LC/MS/MS method for pooled HLMs. UGT Control
R
ing a C18 Sep-Pakꢀ cartridge (Waters Corporation)
equilibrated with a 50:50 (v/v) acetonitrile:water mix-
ture. The 10 mM of ammonium formate (pH 4.0) phase
containing the glucuronide from liquid–liquid extrac-
tion was added dropwise to the extraction cartridge
and eluted using 90:10 (v/v) of acetonitrile:water. The
R
eluant was collected and dried using the SpeedVacꢀ.
Approximately 60 :g of the dried purified glucuronide
was recovered. A portion of the dried powder was
utilized for LC/MS/MS and NMR analysis, and the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011
DOI 10.1002/jps

CHARACTERIZATION OF OXYMETAZOLINE GLUCURONIDATION IN HUMAN LIVER MICROSOMES
787
remaining material was stored at −20◦C as an au-
thentic standard for quantifying the formation of
oxymetazoline glucuronide for kinetic analysis.
The mass transitions monitored were m/z 437.2/261.4
for oxymetazoline glucuronide and m/z 271.9/155.7 for
carbutamide. The mass spectrometer conditions in-
R
cluded TurboIonSprayꢀ source temperature of 500◦C;
Characterization of Oxymetazoline Glucuronide
by NMR
nebulizer and desolvation gases set to 60 and 50 psi;
and the collision cell gas to 10 (arbitrary units). For
all experiments, the spray voltage was set to 5500 V.
The declustering potential, entrance potential, col-
lision energy, and collision cell exit potential were
optimized at 61, 10, 41, and 6 V, respectively. An-
alyst software (Version 1.4.1, Applied Biosystems/
MDS Sciex, Foster City, CA) was used for data
acquisition.
1H NMR spectra were acquired on a 600-MHz Var-
ian Inova spectrometer (Varian, Inc., Palo Alto, Cal-
ifornia) fitted with a Varian 5-mm triple-resonance
cold probe. Samples were prepared in DMSO-d6 at
298.1 K and then transferred into a 5-mm, DMSO-
matched Shigemi tube. Chemical shifts are expressed
in ppm relative to tetramethylsilane as the inter-
nal standard. ROESY (rotating frame nuclear Over-
hauser effect) experiments were recorded with mix-
ing times of 500 and 750 ms, using a spin-lock field
of 4 kHz. A total of 64 complex points were recorded
in the F1 dimension of the correlation plot, with 64
scans acquired per increment. Solvent suppression
was achieved by appending the WET pulse-gradient
train to the beginning of the ROESY pulse sequence.
Enzyme Kinetic Data Analysis
All of the data points represent the mean of three
separate incubations. The kinetic parameters for
oxymetazoline glucuronidation were calculated by
fitting the untransformed experimental data to ei-
ther the single-enzyme Michaelis–Menten model,
v = {Vmax[S]}/{Km + [S]}, where v is the rate of
metabolite formation, Vmax is the maximum veloc-
ity, Km is the Michaelis constant (substrate concen-
tration at 0.5Vmax), and [S] is the substrate con-
centration; the allosteric sigmoidal (Hill) model,14
Enzyme Kinetics of Oxymetazoline Glucuronidation
The kinetics of oxymetazoline glucuronidation were
performed under the same incubation and recovery
conditions as previously described for the metabolite
identification method, except that the protein concen-
trations for pooled HLMs were 0.1 and 0.25 mg of pro-
tein/mL for expressed UGT1A9. The reactions were
incubated for 2.5 min at 37◦C for a final volume of
100 :L in 96-well polypropylene plates. The reactions
were terminated by the addition of 100 :L of ice-cold
acetonitrile containing 1 :M of the internal standard,
and the protein-free supernatants were analyzed for
the quantification of oxymetazoline glucuronide.
A triple-quadrupole LC/MS/MS system consisting
of an Agilent 1100 HPLC, a LEAP autosampler (LEAP
Technologies, Carrboro, North Carolina), and a Sciex
API 4000 mass spectrometer (Applied Biosystems/
MDS Sciex, Foster City, California) was utilized to
quantify oxymetazoline glucuronide. Peak area ra-
tios of the oxymetazoline glucuronide and the inter-
nal standard (carbutamide) were used to determine
the rate of formation of the glucuronide. Separation
of the glucuronide was performed on a Phenomenex
SynergiTM (Torrance, California) C18 Hydro-RP (4
:m; 50 × 3.0 mm) HPLC column. The mobile-phase
gradient consisted of 0.1% (v/v) formic acid in water
as solvent A; solvent B was 0.1% (v/v) formic acid in
acetonitrile. The mobile phase was programmed at a
flow rate of 0.5 mL/min, using the following gradient
expressed as changes in the mobile phase B: 0 to 1
min, hold at 5% B; 1 to 2.5 min, a linear increase from
5% B to 95% B; and 2.5 to 7 min, a linear decrease
from 95% B to 5% B. Oxymetazoline glucuronide and
the internal standard were detected by using positive
ion spray in the multiple reaction-monitoring mode.
v = Vmax × Sn/(Sn + Sn), where S50 is the substrate
50
concentration resulting in 50% Vmax (analogous to
Km in the Michaelis–Menten equation) and n is the
Hill coefficient; or the substrate inhibition model,15,16
v = Vmax × S/[Km + S(1 + S/Ksi)], where Ksi is the con-
stant describing the substrate inhibition interaction,
using Prism 5.2 (GraphPad, San Diego, California),
designed for nonlinear regression analysis. The se-
lection of the “best-fit” kinetic model was based on
comparison of the sum-of-squared residuals, standard
deviation of fit, coefficient of determination (r2), and
F test (Prism 5.2). Enzyme activity is expressed as
reaction rate [pmole/(min mg of protein)].
The intrinsic clearance for the inhibition kinetics
(CLint = Vmax/Km) is expressed as L/(min mg of pro-
tein). The maximum clearance for the Hill equation,
CLmax = Vmax (n − 1)/{S50 × n(n − 1)1/n}, provides
an estimate of the highest clearance attained, that is,
when the enzyme is fully activated before saturation
occurs.15
RESULTS
Oxymetazoline Glucuronidation by HLMs
The mass spectrum of oxymetazoline standard
showed a protonated molecular ion [M + H]+ at m/
z 261 (C16H25N2O) (Fig. 2a). The MS/MS analysis of
[M + H]+ at m/z 261 (Fig. 2b) yielded product ions at
m/z 243 (loss of H2O), m/z 205 [neutral loss of 56 amu
(C4H8, t-butyl)], m/z 191 [loss of 70 amu from cleav-
age of the imidazoline ring (C3H6N2)], and m/z 177
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011

788
MAHAJAN ET AL.
Figure 2. Full mass spectrum (a) and MS/MS spectrum of [M + H]+ at m/z 261 (b) for
oxymetazoline standard.
[loss of 84 amu from cleavage of the methylimidazo-
line moiety (C4H8N2)]. These four product ion losses
are characteristic for oxymetazoline and aid in the
identification of oxymetazoline glucuronide metabo-
lite.
retention time 4.6 min was also detected in human
liver S9 incubations supplemented with UDPGA hav-
ing a similar fragmentation pattern as the HLM prod-
uct at 4.6 min (data not shown).
Oxymetazoline Glucuronidation by Expressed
UGT Isoforms
While a UV peak was detected at retention time
4.6 min (Fig. 3a) from the incubations containing
HLMs supplemented with UDPGA, no peak was ob-
served at the similar retention time from the incu-
bations without UDPGA in HLMs. The extracted ion
chromatogram (XIC) for m/z 437 (176 amu addition to
oxymetazoline) also showed a single peak at 4.8 min
(Fig. 3b), further supporting the UV peak at 4.6 min
for the formation of oxymetazoline glucuronide. The
full mass spectrum showing the [M + H]+ at m/z 437
for oxymetazoline glucuronide is shown in Figure 4a.
The addition of 176 amu to oxymetazoline indicates
the formation of a glucuronide conjugate.17,18 The MS/
MS analysis of [M + H]+ at m/z 437 for oxymetazo-
line glucuronide from HLM incubations gave product
ions at m/z 261 (characteristic loss of 176 amu for a
glucuronide) and m/z 205 (loss of 56 amu for t-butyl
group), which are indicative of an oxymetazoline glu-
curonide (Fig. 4b). The oxymetazoline glucuronide at
The incubations of oxymetazoline with expressed hu-
man UGT isoforms revealed that UGT1A9 is the only
UGT catalyzing the glucuronidation of oxymetazo-
line (Fig. 5). The XIC of the oxymetazoline incuba-
tions supplemented with UDPGA and a panel of 11
expressed human UGTs showed a peak at 4.6 min
with [M + H]+ at m/z 437 for UGT1A9. No XIC glu-
curonide peak at m/z 437 was detected for any of the
other UGTs (UGT1A1, UGT1A3, UGT1A4, UGT1A6,
UGT1A8, UGT1A10, UGT2B4, UGT2B7, UGT2B15,
and UGT2B17). The MS/MS analysis of [M + H]+ at
m/z 437 gave product ions at m/z 261 and m/z 205
characteristic for an oxymetazoline glucuronide (data
not shown). LC/MS/MS was used to identify the rel-
evant UGT enzymes catalyzing the glucuronidation
of oxymetazoline because of its greater sensitivity
and selectivity for detecting the glucuronide relative
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011
DOI 10.1002/jps

CHARACTERIZATION OF OXYMETAZOLINE GLUCURONIDATION IN HUMAN LIVER MICROSOMES
789
Figure 4. Full mass spectrum (a) and MS/MS spectrum of
[M + H]+ at m/z 437 (b) from the oxymetazoline incubations
containing alamethicin-activated human liver microsomes
supplemented with uridine-5-diphosphoglucuronic acid.
Figure 3. Representative expanded LC–UV chro-
matogram (a) and extracted ion chromatogram (b) from
the oxymetazoline incubations containing alamethicin-
activated human liver microsomes supplemented with
uridine-5-diphosphoglucuronic acid.
for the various peaks of oxymetazoline and oxymeta-
zoline O-glucuronide are summarized in Table 1.
Table 1. Comparison of the Chemical Shift of Protons of
Oxymetazoline and Oxymetazoline $-O-Glucuronide
to UV detection. These results were used to select
UGT1A9 for more detailed kinetic analysis.
Chemical Shift in ppm (# of Integrated Hs)
Characterization of Oxymetazoline β-O-Glucuronide
by NMR
Relevant
Proton
Oxymetazoline
$-O-Glucuronide
Oxymetazoline
Because the LC/MS/MS could not identify whether
oxymetazoline had been glucuronidated at either the
phenolic group or one of the nitrogen atoms of the im-
idazoline ring, the chemical shift, coupling constants,
HA
HB
HC
HD
HE
HF
H1
H2
H3
H4
1.347 (9 Hs)
2.107 (3 Hs)
2.181 (3 Hs)
3.774 (4 Hs)
3.858 (2 Hs)
6.898 (1 H)
NA
1.372 (9 Hs)
2.170 (3 Hs)
2.267 (3 Hs)
3.496 (4 Hs)
3.585 (2 Hs)
6.977 (1 H)
4.520 (1 H)
3.088 (1 H)
2.909 (1 H)
3.159 (1 H)
1
and integration of the relevant protons in H NMR
and cross peak interactions of the adjacent protons
in 2D-ROESY spectrum of biosynthesized oxymeta-
zoline glucuronide standard were analyzed to char-
acterize the glucuronide as the O-glucuronide. The
chemical shift and the number of protons integrated
NA
NA
NA
NA, not available.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011

790
MAHAJAN ET AL.
Figure 5. Screening of 11 cDNA expressed human uridine glucuronosyltransferases (UGT)
SupersomesTM with 50 :M of oxymetazoline for the formation of oxymetazoline glucuronide.
The incubation conditions and extracted-ion LC/MS/MS analysis were as described under the
Materials and Methods section. The bar for UGT1A9 represents the mean of duplicate determi-
nations.
1
The H NMR spectrum of the purified oxymetazo-
the phenolic group. Thus, our NMR data utilizing a
combination of proton chemical shift assignments,
vicinal proton–proton coupling constants, and 2D-
ROESY cross peak interactions demonstrate that the
oxymetazoline glucuronide formed by pooled HLMs or
expressed UGT1A9 is the phenolic $-O-glucuronide.
line glucuronide standard showed that chemical shift
of the anomeric glycosidic proton (H1) occurs at 4.520
ppm, similar to reports for other O-glucuronides.18,19
Glucuronide proton signals (H2, H3, and H4) were ob-
served at 3.088, 2.909, and 3.159 ppm, respectively.19
Evidence for the $-O-glucuronide was provided by the
vicinal trans proton–proton coupling of the anomeric
proton H1 with the H4 proton of the glucuronide
ring. Our results show that the vicinal trans coupling
constant for oxymetazoline O-glucuronide is 7.751,
which is in agreement with the literature values, con-
firming the $-O-glycosidic linkage for oxymetazoline
glucuronide.20,21 Thus, 1H NMR suggests oxymetazo-
line glucuronide to be the $-O-glucuronide.
The attachment of the $-O-glucuronide to
oxymetazoline was further confirmed with the 2D-
ROESY spectrum for the purified oxymetazoline
O-glucuronide, which showed strong cross peak in-
teractions between the anomeric proton signals (H1)
at 4.520 ppm and the t-butyl protons signal (HA) at
1.372 ppm, the methyl protons (HC) at 2.267 ppm,
the glucuronide proton (H3) at 2.909 ppm, and the
glucuronide proton (H2) at 3.088 ppm (Fig. 6). Al-
though interactions between H1 and HC, H1 and H3,
and H1 and H2 would have been feasible irrespective
of the position of glucuronide attachment at a phe-
nolic hydroxyl group or one of the nitrogen atoms of
the imidazoline ring, the strong cross peak interac-
tion of t-butyl protons (HA) with the anomeric pro-
ton (H1) indicates that the glucuronide is linked to
Figure 6. Structure of oxymetazoline $-O-glucuronide
showing the key two-dimensional (2D) NMR cross peak
proton interactions of anomeric proton (H1) of biosynthe-
sized oxymetazoline $-O-glucuronide standard. Samples
were prepared in DMSO-d6 at 298.1 K, and 2D NMR spectra
were acquired on 600-MHz Varion Inova spectrometer.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011
DOI 10.1002/jps

CHARACTERIZATION OF OXYMETAZOLINE GLUCURONIDATION IN HUMAN LIVER MICROSOMES
791
Oxymetazoline O-Glucuronidation Kinetics
by Pooled HLMs
The initial rate conditions for oxymetazoline glu-
curonidation with HLMs were determined using a
range of protein concentrations (1–4 mg of protein/
mL) and incubation times (0–60 min). Oxymetazoline
O-glucuronidation was determined to be linear up to
2 mg of microsomal protein/mL in HLMs, and the rate
of glucuronidation was linear up to 2.5 min.
Oxymetazoline O-glucuronidation by pooled HLMs
exhibited nonhyperbolic (atypical) kinetics charac-
teristic of allosteric sigmoidal (Hill) kinetics when
initially fitted to the Michaelis–Menten equation
(Fig. 7a, manifested as a curvilinear Eadie–Hofstee
plot inset).15 Thus, the Hill equation was best fit-
ted to the experimental data, and the derived ki-
netic parameters by pooled HLMs for S50 were 2.42
0.40 mM, Vmax 8.69 0.58 pmole/(min mg of protein),
and Hill coefficient (n) of 1.69, with an apparent CLmax
of 3.61 L/(min mg of protein). The Hill coefficient is a
measure of sigmoidicity.
Oxymetazoline O-Glucuronidation Kinetics by
Expressed UGT1A9
The initial rate conditions for oxymetazoline O-
glucuronidation by expressed UGT1A9 were deter-
mined in same manner as those for the pooled HLMs.
The O-glucuronidation of oxymetazoline was deter-
mined to be linear up to 1 mg of protein/mL, and the
rate of glucuronidation was linear up to 2.5 min with
expressed UGT1A9.
In contrast to the sigmoidal kinetics observed by
pooled HLMs, oxymetazoline O-glucuronidation with
expressed UGT1A9 exhibited nonhyperbolic (atyp-
ical) kinetics characteristic of substrate inhibition
when initially fitted to the Michaelis–Menten equa-
tion (Fig. 7b, Eadie–Hofstee plot inset).15 Thus, the
substrate inhibition equation was best fitted to the
experimental data and the derived kinetic parame-
ters by UGT1A9 for Km were 2.53 1.03 mM, Vmax
54.18 16.92 pmole/(min mg of protein), and ki 2.44.
The apparent CLint (substrate inhibition) was 21.41 L/
(min mg of protein). Thus, UGT1A9 appears to exhibit
low affinity for oxymetazoline in HLMs.
Figure 7. The kinetics of oxymetazoline $-O-
glucuronidation activity by pooled human liver micro-
somes (HLMs) represents fitting the Hill equation to
the data. The inset shows the curvilinear Eadie–Hofstee
plot. The oxymetazoline $-O-glucuronidation activity
was determined as described under the Materials and
Methods section. Each data point represents the mean of
triplicate incubations
SD (a). The kinetics of oxymeta-
zoline O-glucuronidation activity by expressed human
UGT1A9 from baculovirus-infected insect cells represents
fitting the data to the substrate inhibition equation. The
inset shows the Eadie–Hofstee plot. The oxymetazoline
O-glucuronidation activity was determined as described
under the Materials and Methods section. Each data point
represents the mean of triplicate incubations SD (b).
DISCUSSION
The UGT enzymes responsible for catalyzing the
glucuronidation of oxymetazoline remain thus far
unidentified. Therefore, in this study, we examined
the glucuronidation of oxymetazoline by alamethicin-
activated pooled HLMs supplemented with UDPGA.
The formation of an oxymetazoline glucuronide was
coupling constant distinctive for the formation of a $-
O-glucuronide. Thus, the glucuronide was confirmed
by NMR to be the $-O-glucuronide of oxymetazoline.
The 2D-ROESY spectral analysis of cross peak in-
teractions with adjacent protons further confirmed
that the phenolic group of oxymetazoline had been
O-glucuronidated.
1
identified by LC/MS/MS. The H NMR spectrum of
this glucuronide showed an anomeric glycosidic pro-
ton with a chemical shift and a vicinal proton–proton
Screening with 11 expressed UGTs (UGT1A1,
UGT1A3, UGT1A4, UGT1A6, UGT1A8, UGT1A9,
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011

792
MAHAJAN ET AL.
UGT1A10, UGT2B4, UGT2B7, UGT2B15, and
UGT2B17) identified UGT1A9 as the only tested iso-
form catalyzing the O-glucuronidation of oxymeta-
zoline. Because no human therapeutic plasma con-
centrations for oxymetazoline have been reported
up to now following intranasal dosing, the 50 :M
concentration of substrate, which is at least 130-
fold greater than its 380-nM therapeutic dose, was
used in the screening study to predict the relative
contribution of each isoform to total oxymetazoline
glucuronidation.22 At the 50 :M concentration, no
detectable O-glucuronidation activities for the other
relevant UGTs were observed, suggesting that the
other enzymes are unlikely to participate in the in
vitroO-glucuronidation at purportedly subnanomolar
therapeutic plasma concentrations.
therapeutic dose, the drug would be eliminated
unchanged.
In conclusion, the $-O-glucuronide of oxymeta-
zoline has been identified as the only glucuronide
formed by pooled HLMs and expressed UGT1A9 and
was detected by LC/MS/MS and characterized by 1D
and 2D NMR. Screening with 11 expressed human
UGT isoforms identified UGT1A9 as the only de-
tectable UGT catalyzing the $-O-glucuronidation of
oxymetazoline by HLMs. From a clinical perspec-
tive, our results show that the low incidence of ad-
verse events for oxymetazoline at its therapeutic
dose cannot be attributed solely to its elimination
via the glucuronide detoxification pathway, because
UGT1A9 would contribute only to its elimination at
toxic plasma concentrations. This study highlights
the fact that the identification and the kinetics of the
UGT isoform are crucial to a thorough understanding
of human drug disposition and safety.
The kinetics for the O-glucuronidation of oxymeta-
zoline was assessed by alamethicin-activated pooled
HLMs and expressed UGT1A9 supplemented with
UDPGA. A limitation for this study was the lack
of human nasal epithelia–expressed UGT2A1 and
UGT2A2 to also assess the O-glucuronidation of
oxymetazoline. The O-glucuronidation of oxymetazo-
line by pooled HLMs exhibited allosteric sigmoidal
(Hill) kinetics, whereas expressed UGT1A9 showed
substrate inhibition kinetics. Because pooled HLMs
are a crude mixture, it contains many artifactual
substances that could contribute to sigmoidicity of
the observed HLM kinetics.15 The value of Km from
the results with expressed enzyme indicates UGT1A9
to be a low-affinity enzyme and thus glucuronidates
oxymetazoline with poor efficiency at the subnanomo-
lar therapeutic plasma concentrations anticipated
from a 380-nM intranasal dose. A substrate that
is glucuronidated solely by a single expressed UGT
isoform should have an apparent Km similar to the
apparent Km for pooled HLMs, which was observed
for oxymetazoline, confirming the results from the
screening study. The in vitro clearance of oxymeta-
zoline was assessed as CLint (substrate inhibition ki-
netics) for expressed UGT1A9 or CLmax (sigmoidal
Hill kinetics) for pooled HLMs. CLmax provides an
estimate of the highest clearance attained prior to
saturation of the enzyme active sites. When the ki-
netic parameters for oxymetazoline are compared
with the reported values for the sterically hindered
and structurally similar O- diisopropylphenol, propo-
fol (Km = 45 :M), oxymetazoline exhibits approxi-
mately 50-fold less affinity and 40-fold less reactivity
for UGT1A9 and Clint is 2.5-fold less than that for
propofol [Clint 54.3 L/(min mg of protein)].23 These
results would seem to indicate that the low affin-
ity for oxymetazoline is not likely due to steric hin-
drance of the phenolic group by the flanking t-butyl
and methyl groups but to other factors. Thus, because
the value of Km for glucuronidation of oxymetazoline
by UGT1A9 is more than 6000-fold than its intranasal
ACKNOWLEDGMENTS
The authors thank Drs Chuang Lu, Scott Daniels, Ian
Parsons, Ajit Chavan, and Sandeep Pusalkar for their
assistance and helpful discussions.
REFERENCES
1. Thrush DN. 1995. Cardiac arrest after oxymetazoline nasal
spray. J Clin Anesth 7:512–514.
2. Soderman P, Sahlberg D, Wiholm BE. 1984. CNS reactions to
nose drops in small children. Lancet 323:573.
3. Glazener F, Blake K, Gradman M. 1983. Bradycardia, hy-
potension, and near-syncope associated with Afrin (oxymeta-
zoline) nasal spray. N Engl J Med 309:731.
4. Savage R. 2009. Safety of bromhexine and topical nasal
decongestants presentation 18 August 2009. Accessed
February 24, 2010, at: http://www.medsafe.govt.nz/hot/alerts/
CoughandCold/Safety Bromhexine.pdf.
5. Hayes FJ, Baker TR, Dobson RLM, Tsueda MS. 1995.
Rapid liquid chromatographic-mass spectrometric assay for
oxymetazoline in whole rat blood. J Chromatogr A 692:73–81.
6. Sneitz N, Court MH, Zhang X, Laajanen K, Yee KK, Dalton P,
Ding X, Finel M. 2009. Human UDP-glucuronosyltransferase
UGT2A2: cDNA construction, expression, and functional char-
acterization in comparison with UGT2A1 and UGT2A3. Phar-
macogenet Genom 19:923–934.
7. Mackenzie PI, Owens IS, Burchell B, Bock KW, Bairoch A,
Be´langer A, Fournel-Gigleux S, Green M, Hum DW, Iyanagi
T, Lancet D, Louisot P, Magdalou J, Chowdhury JR, Ritter
JK, Schachter H, Tephly TR, Tipton KE, Nebert DW. 1997. The
UDP-glycosyltransferase gene superfamily: Recommended
nomenclature update based on evolutionary divergence. Phar-
macogenetics 7:255–269.
8. Stone AN, Mackenzie PI, Galetin A, Houston JB, Miners JO.
2003. Isoform selectivity and kinetics of morphine 3- and
6-glucuronidation by human UDP-glucuronosyltransferases:
Evidence for atypical glucuronidation kinetics by UGT2B7.
Drug Metab Dispos 31:1086–1089.
9. Miners JO, Smith PA, Sorich MJ, McKinnon RA, Mackenzie
PI. 2004. Predicting human drug glucuronidation parameters:
Application of in vitro and in silico modeling approaches. Annu
Rev Pharmacol Toxicol 44:1–25.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011
DOI 10.1002/jps

CHARACTERIZATION OF OXYMETAZOLINE GLUCURONIDATION IN HUMAN LIVER MICROSOMES
793
10. Court MH, Duan SX, Von Moltke LV, Greenblatt DJ,
Patten CJ, Miners JO, Mackenzie PI. 2001. Interindivid-
ual variability in acetaminophen glucuronidation by human
liver microsomes: Identification of relevant acetaminophen
UDP-glucuronosyltransferase isoforms. J Pharmacol Exp Ther
299:998–1006.
11. Nakajima M, Sakata N, Ohashi N, Kume T, Yokoi T.
2002. Involvement of multiple UDP-glucuronosyltransferase
1A isoforms in glucuronidation of 5-(4ꢀ-hydroxyphenyl)-5-
phenylhydantoin in human liver microsomes. Drug Metab Dis-
pos 30:1250–1256.
12. Watanabe Y, Nakazima M, Ohashi N, Kume T, Yokoi T. 2003.
Glucuronidation of etoposide in human liver microsomes is
specifically catalyzed by UDP-glucuronosyltransferase 1A1.
Drug Metab Dispos 31:589–595.
13. Zhang JY, Zhan J, Cook CS, Ings RM, Breau AP. 2003. Involve-
ment of human UGT2B7 and 2B15 in rofecoxib metabolism.
Drug Metab Dispos 31:652–658.
N-formylmethamphetamine by FAB/MS, LC/MS/MS, and
NMR. Drug Metab Dispos 20:451–460.
18. Mutlib AE, Chen H, Nemeth GA, Markwalder JA, Seitz SP,
Gan LS, Christ DD. 1999. Identification and characteriza-
tion of Efavirenz metabolites by liquid chromatography/mass
spectrometry and high field NMR: Species differences in the
metabolism of Efavirenz. Drug Metab Dispos 27:1319–1333.
19. Krishnaswamy S, Duan SX, Moltke VLL, Greenblatt DJ,
Sudmeier JL, Bachovchin WW, Court MH. 2003. Serotonin
(5-hydroxytryptamine) glucuronidation in vitro: Assay devel-
opment, human liver microsome activities and species differ-
ences. Xenobiotica 33:169–180.
20. Matsui M, Okada M, Ishidate M. 1969. Studies on the glu-
caric acid pathway in the metabolism of d-glucuronic acid in
mammals, part II. Excretion of d-glucaric acid in urine after
administration of several monosaccharides and their deriva-
tives to mammals. Chem Pharm Bull 17:1064–1071.
21. Oatis JE Jr, Baker JP, McCarthy D, Knapp DR. 1983. Syn-
thesis and chromatographic separation of the glucuronides of
(R)-and (S)-propranolol. J Med Chem 26:1687–1691.
14. Weiss JN. 1997. The Hill equation revisited: Uses and misuses.
FASEB J 11:835–841.
15. Houston JB, Kenworthy KE. 2000. In vitro–in vivo scaling of
CYP kinetic data not consistent with the classical Michaelis—
Menten model. Drug Metab Dispos 28:246–254.
16. Venkatakrishnan K, Von Moltke LL, Greenblatt DJ. 2001.
Human drug metabolism and the cytochromes P450: Appli-
cation and relevance of in vitro models. J Clin Pharmacol
41:1149–1179.
22. Ogilvie BW, Usuki E, Burton A, Yerino P, Parkinson A.
2008. In vitro approaches for studying the inhibition of drug-
metabolizing enzymes and identifying the drug-metabolizing
enzymes responsible for the metabolism of drugs (reaction phe-
notyping) with emphasis on cytochrome P450. In Drug–drug
interactions; Rodrigues AD, Ed. 2nd ed. New York: Marcel
Dekker Inc., pp 231–258.
17. Mutlib AE, Abbott FS. 1992. Isolation and characterization of
carbinolamide and phenolic glucuronide conjugates of (+−)-
N-methyl-N-(1-methyl-3,3-diphenylpropyl) formamide and
23. Soars MG, Ring BJ, Wrighton SA. 2003. The effect of
incubation conditions on the enzyme kinetics of UDP-
glucuronosyltransferases. Drug Metab Dispos 31:762–767.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 2, FEBRUARY 2011