Three-dimensional quantitative structure-activity relationship studies on UGT1A9-mediated 3-O-glucuronidation of natural flavonols using a pharmacophore-based comparative molecular field analysis model

Article abstract of DOI:10.1124/jpet.110.175356

Glucuronidation is often recognized as one of the rate-determining factors that limit the bioavailability of flavonols. Hence, design and synthesis of more bioavailable flavonols would benefit from the establishment of predictive models of glucuronidation using kinetic parameters [e.g., K_m, V _max, intrinsic clearance (CL_int) = V_max/K _m] derived for flavonols. This article aims to construct position (3-OH)-specific comparative molecular field analysis (CoMFA) models to describe UDP-glucuronosyltransferase (UGT) 1A9-mediated glucuronidation of flavonols, which can be used to design poor UGT1A9 substrates. The kinetics of recombinant UGT1A9-mediated 3-O-glucuronidation of 30 flavonols was characterized, and kinetic parameters (K_m, V_max, CL_int) were obtained. The observed K_m, V_max, and CL_int values of 3-O-glucuronidation ranged from 0.04 to 0.68 μM, 0.04 to 12.95 nmol/mg/min, and 0.06 to 109.60 ml/mg/min, respectively. To model UGT1A9-mediated glucuronidation, 30 flavonols were split into the training (23 compounds) and test (7 compounds) sets. These flavonols were then aligned by mapping the flavonols to specific common feature pharmacophores, which were used to construct CoMFA models of V_max and CL_int, respectively. The derived CoMFA models possessed good internal and external consistency and showed statistical significance and substantive predictive abilities (V_max model: q² = 0.738, r² = 0.976, r_pred² = 0.735; CL_int model: q² = 0.561, r² = 0.938, r_pred² = 0.630). The contour maps derived from CoMFA modeling clearly indicate structural characteristics associated with rapid or slow 3-O-glucuronidation. In conclusion, the approach of coupling CoMFA analysis with a pharmacophore-based structural alignment is viable for constructing a predictive model for regiospecific glucuronidation rates of flavonols by UGT1A9. Copyright

Full text of DOI:10.1124/jpet.110.175356

Supplemental material to this article can be found at:
http://jpet.aspetjournals.org/content/suppl/2010/11/10/jpet.110.175356.DC1.html
0022-3565/11/3362-403–413$20.00
THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS
Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics
JPET 336:403–413, 2011
Vol. 336, No. 2
175356/3660793
Printed in U.S.A.
Three-Dimensional Quantitative Structure-Activity Relationship
Studies on UGT1A9-Mediated 3-O-Glucuronidation of Natural
Flavonols Using a Pharmacophore-Based Comparative
□
S
Molecular Field Analysis Model
Baojian Wu, John Kenneth Morrow, Rashim Singh, Shuxing Zhang, and Ming Hu
Department of Pharmacological and Pharmaceutical Sciences, College of Pharmacy, University of Houston, Houston, Texas
(B.W., R.S., M.H.); and Department of Experimental Therapeutics, University of Texas M. D. Anderson Cancer Center, Houston,
Texas (J.K.M., S.Z.)
Received September 23, 2010; accepted November 4, 2010
ABSTRACT
Glucuronidation is often recognized as one of the rate-deter- vonols were split into the training (23 compounds) and test (7
mining factors that limit the bioavailability of flavonols. Hence, compounds) sets. These flavonols were then aligned by map-
design and synthesis of more bioavailable flavonols would ben- ping the flavonols to specific common feature pharmacoph-
efit from the establishment of predictive models of glucuronida- ores, which were used to construct CoMFA models of Vmax and
tion using kinetic parameters [e.g., Km, Vmax, intrinsic clearance CLint, respectively. The derived CoMFA models possessed
(CLint) ϭ Vmax/Km] derived for flavonols. This article aims to good internal and external consistency and showed statistical
construct position (3-OH)-specific comparative molecular field significance and substantive predictive abilities (Vmax model:
analysis (CoMFA) models to describe UDP-glucuronosyltrans- q2 ϭ 0.738, r2 ϭ 0.976, rp2red ϭ 0.735; CLint model: q2 ϭ 0.561,
ferase (UGT) 1A9-mediated glucuronidation of flavonols, which r2 ϭ 0.938, rp2red ϭ 0.630). The contour maps derived from
can be used to design poor UGT1A9 substrates. The kinetics of CoMFA modeling clearly indicate structural characteristics as-
recombinant UGT1A9-mediated 3-O-glucuronidation of 30 fla- sociated with rapid or slow 3-O-glucuronidation. In conclusion,
vonols was characterized, and kinetic parameters (Km, Vmax
,
the approach of coupling CoMFA analysis with a pharmacoph-
CLint) were obtained. The observed Km, Vmax, and CLint values ore-based structural alignment is viable for constructing a pre-
of 3-O-glucuronidation ranged from 0.04 to 0.68 ␮M, 0.04 to dictive model for regiospecific glucuronidation rates of fla-
12.95 nmol/mg/min, and 0.06 to 109.60 ml/mg/min, respec- vonols by UGT1A9.
tively. To model UGT1A9-mediated glucuronidation, 30 fla-
antioxidant and anticancer effects (Birt et al., 2001; Ross and
Kasum, 2002). However, the undesired biopharmaceutical
Introduction
Flavonols are widely distributed in regular human diets
properties of flavonols (e.g., quercetin and kaempferol),
(D’Archivio et al., 2007). There are nearly 400 natural fla-
which undergo particularly rapid and extensive phase II
vonols reported, and the majority of them have flavonol
metabolism, produced low levels of the parent compounds but
structural scaffold with hydroxyl (OH) and/or methoxy (OMe)
high levels of the conjugated forms (e.g., glucuronides or
substitutions (Andersen and Markham, 2006). It is now well
sulfates to a lesser extent) found in the blood (Erlund et al.,
known that this class of naturally occurring compounds is
linked to a variety of health-promoting activities such as
2006; Barve et al., 2009).
UDP-glucuronosyltransferases (UGTs) are a family of en-
zymes that mediate the glucuronidation of endogenous or
exogenous compounds. The substrates need to contain one or
more nucleophilic groups (e.g., hydroxyl, alcohol, amine,
thiol, or carboxylic acid groups) to which the cofactor UDP-
glucuronic acid is covalently linked (Jancova et al., 2010).
This pattern of structural recognition explains the broad
This work was supported by the National Institutes of Health National
Institute of General Medical Sciences [Grant GM070737] (to M.H.).
Article, publication date, and citation information can be found at
http://jpet.aspetjournals.org.
doi:10.1124/jpet.110.175356.
□S The online version of this article (available at http://jpet.aspetjournals.org)
contains supplemental material.
ABBREVIATIONS: UGT, UDP-glucuronosyltransferase; CoMFA, comparative molecular field analysis; UPLC, ultra performance liquid chroma-
tography; CLint, intrinsic clearance; 2D, two-dimensional; 3D, three-dimensional; QSAR, quantitative structure-activity relationship; HF, hydroxy-
flavone; DHF, dihydroxyflavone; THF, trihydroxyflavone; QHF, tetrahydroxyflavone; RMS, root mean square.
403

404
Wu et al.
substrate specificity of UGTs and underpins its wide detoxi- electrostatic properties). The final validated CoMFA model
fication functionality by turning the substrates to hydrophilic can be used to design novel ligands and predict the biological
metabolite, which can be readily eliminated (Iyanagi, 2007). activities thereof. CoMFA methodology has been widely and
In the UGT superfamily, the enzymes of the UGT1A and successfully used to model interactions between proteins and
UGT2B subfamilies contribute significantly to phase II me- many types of biological ligands such as enzyme inhibitors
tabolism (Wong et al., 2009). To date, a total of 16 functional (Barreca et al., 1999; Holder et al., 2007), CYP2D6 substrates
isoforms have been identified (nine for UGT1A and seven for (Haji-Momenian et al., 2003), and antifungal agents (Wei et
UGT2B): UGT1A1, UGT1A3, UGT1A4, UGT1A5, UGT1A6, al., 2005).
UGT1A7, UGT1A8, UGT1A9, UGT1A10, UGT2B4, UGT2B7,
It is still not fully understood which characteristics make
UGT2B10, UGT2B11, UGT2B15, UGT2B17, and UGT2B28 flavonol a good or poor UGT1A9 substrate. UGT1A9 often
(Mackenzie et al., 2005). Liver, as the major first-pass me- generates multiple monoglucuronide isomers from a single
tabolizing organ, expresses a variety of UGT isoforms, in- flavonoid that bears more than one conjugation site. For
cluding UGT1A1, UGT1A3, UGT1A4, UGT1A5, UGT1A6, example, two glucuronides were generated from 3,7-dihy-
UGT1A9, UGT2B7, and UGT2B15. In contrast, UGT1A7, droxylfavone (Tang et al., 2010), three glucuronides were
UGT1A8, and UGT1A10 are present primarily in the gastro- generated from galangin and luteolin (Otake et al., 2002;
intestinal tract or esophagus (Fisher et al., 2001; Ohno and Chen et al., 2008), and four glucuronides were generated
Nakajin, 2009).
from quercetin (Chen et al., 2008). Currently, there is no
UGT1A9 is a unique enzyme among the nine UGT1A iso- method that can be used to predict the kinetic parameters of
forms. UGT1A9 is resistant to detergent and is stable in glucuronidation for polyphenolic compounds such as flavonol.
response to heat treatment at 57°C for 15 min (Fujiwara et It is assumed that the difficulties arose from the approach
al., 2009). Consistent with this observation, our earlier study that seeks to build a single model that will predict the rates
showed that UGT1A9 was thermostable in metabolizing pru- of glucuronidation at multiple sites. We believe that a new
netin (an isoflavone) at 37°C for more than 8 h (Joseph et al., predictive algorithm is needed to account for glucuronidation
2007). Moreover, UGT1A9 demonstrated greater proficiency at all phenolic groups (-OH) in flavonols. This algorithm
in glucuronidating flavonoids. In the study by Chen et al. must consist of multiple regiospecific (position-specific) mod-
(2008), the catalyzing efficiency (Vmax/Km) of UGT1A9 was els, which can separately predict the glucuronidation rates
higher than that of UGT1A3. It has also been observed that on a particular position (e.g., 3-O-glucuronidation in fla-
UGT1A9 ranks within the top three isoforms that most ef- vonols), and the addition of glucuronidation rates from the
fectively metabolize isoflavones, flavones, and flavonols multiple positions will represent the total glucuronidation
(Tang et al., 2009, 2010). The good correlation between gluc- rates or overall metabolic susceptibility. Therefore, the ob-
uronidation rates derived from UGT isoforms (UGT1A9 plus jective of this article is to construct a position (3-OH)-specific
UGT1A1) and those from human liver microsomes highlight CoMFA model to predict 3-O-glucuronidation. 3-O-gluc-
the fact that UGT1A9 contributes significantly to the gluc- uronidation was chosen because it is the most active position
uronidation activity of liver microsomes in metabolizing fla- in UGT1A9-mediated glucuronidation (Otake et al., 2002;
vonoids (Tang et al., 2010).
Tang et al., 2010). Kinetics parameters (Km, Vmax, CLint) of
It is therefore hypothesized that a poor flavonol substrate UGT1A9-mediated 3-O-glucuronidation with 30 flavonols
of UGT1A9 will have less efficient glucuronidation in vivo, were determined. The Vmax and CLint datasets were used to
which could lead to higher bioavailability. One approach to construct their respective position-specific (3-OH) CoMFA
finding the poor substrate (with potential for high bioavail- models, based on molecular alignments generated from a
ability) is to test every compound experimentally, which is defined pharmacophore search. Contour maps from CoMFA
labor- and cost-intensive. Alternatively, a quantitative struc- provided insightful structural characteristics associated with
ture-activity relationship (QSAR) can be established. Ap- rapidly or slowly metabolized flavonol substrates for 3-O-
proaches to predicting biological activities by building QSAR glucuronidation.
models are usually from two directions: structure-based or
ligand-based. Because of a lack of mammalian UGT crystal
structures, ligand-based approaches [e.g., 2D/3D QSAR,
pharmacophore modeling, and comparative molecular field
analysis (CoMFA)] have been applied to establish quantita-
tive or qualitative structure-glucuronidation relationships
(Sorich et al., 2008). Predictive regression and pharmacoph-
ore models have been generated with UGT1A1 and UGT1A4
Materials and Methods
Materials. Expressed human UGT1A9 isoform (Supersomes)
were purchased from BD Gentest (Woburn, MA). The recombinant
UGT1A9 displayed a Km value of 25.8 Ϯ 4.47 ␮M and a Vmax value of
148 Ϯ 7.03 pmol/mg/min when it was used to glucuronidate the probe
substrate propofol (Supplemental Fig. S1). Uridine diphosphogluc-
by using structurally diverse compounds (Sorich et al., 2002; uronic acid, alamethicin, D-saccharic-1,4-lactone monohydrate, and
magnesium chloride were purchased from Sigma-Aldrich (St. Louis,
MO). Ammonium acetate was purchased from J. T. Baker (Phillips-
burg, NT). Thirty flavonols (Fig. 1 and Table 1) were purchased from
Indofine Chemicals (Hillsborough, NJ). The flavonols structurally
differ in number and position of substituents [hydroxyl (-OH) or
methoxy (-OMe)] on the A-ring or B-ring. Particularly, compounds 16
and 27 have methyl groups at C6. C3Ј and C4Ј of compound 16 are
cosubstituted by the methylenedioxy group. All other materials (typ-
ically analytical grade or better) were used as received.
Smith et al., 2003). However, attempts to develop predictive
QSAR for substrates of UGT1A9 have been unsuccessful
(Miners et al., 2004). On the other hand, CoMFA was used to
examine a series of 18 compounds (triphenylalkyl carboxylic
acid analog) that inhibited glucuronidation of bilirubin by
UGT1A1; the resulting model allowed good prediction of in-
hibitory potency (Said et al., 1996).
CoMFA is a 3D QSAR technique (Cramer et al., 1988) that
aims to derive a correlation between the biological activity of
UGT1A9 Enzyme Kinetics. Enzyme kinetics parameters of gluc-
ligands and their structural characteristics (i.e., steric and uronidation by UGT1A9 were determined by measuring initial gluc-

Prediction of Flavonol Glucuronidation by UGT1A9 with CoMFA
405
to 90% B; 5 to 5.5 min, 90% B; and 5.5 to 6 min, 90 to 100% B.
Quantitation of the glucuronide was based on the standard curve of
the parent compound and was further calibrated using the conver-
sion factor. The conversion factors (Supplemental Table S1) were
generated for 3-O-glucuronides compared with the respective agly-
cone following the protocol described previously (Tang et al., 2009,
2010). Representative chromatograms and UV spectra are shown in
Supplemental Fig. S2.
Glucuronide Structure Confirmation. Glucuronide structures
were confirmed via a three-step process as summarized previously
(Singh et al., 2010). First, the glucuronides were hydrolyzed by
␤-D-glucuronidase to the aglycones. Second, the glucuronides were
identified as monoglucuronides that showed mass of [(aglycone’s
mass) ϩ 176] Da using UPLC/tandem mass spectroscopy, where 176
Da is the mass of single glucuronic acid. The same UPLC/tandem
mass spectroscopy instruments and methods described previously
(Singh et al., 2010) were applied in this study. Finally, the 3-O-
Fig. 1. Backbone of flavonol structures (see Table 1 for the definitions of
the substituents). Compared with flavonols, flavones lack 3-OH on the
2-phenylbenzopyran scaffold.
uronidation rates of flavonols at a series of concentrations. The
experimental procedures of UGT assays were exactly the same as
those described previously (Joseph et al., 2007; Tang et al., 2009,
2010; Singh et al., 2010). Glucuronidation rates were calculated as
the amount of glucuronides formed per protein quantity per reaction
time (or nmol/mg/min). Aglycone substrate concentrations in the
range of 0.039 to 40 ␮M were used unless method sensitivity or
substrate solubility necessitated otherwise. All experiments were
performed in triplicate.
Ultra Performance Liquid Chromatography Analysis of
Flavonols and Their Glucuronides. The Waters (Milford, MA)
ACQUITY ultra performance liquid chromatography (UPLC) system
was used to analyze the parent compounds and their corresponding
glucuronides. UPLC methods for each flavonol and the glucuronides
were essentially the same as described previously (Tang et al., 2009,
2010; Singh et al., 2010) with slight modifications. In brief, mobile
phase A was 2.5 mM ammonium acetate in purified water (pH 6.5).
Mobile phase B was 100% acetonitrile. The mobile phase was run 5
min at a flow rate of 0.45 ml/min with the predetermined gradient (0
min, 10% B; 0–2 min, 10–20% B; 2–3 min, 20–40% B; 3–3.5 min,
40–50% B; 3.5–4 min, 50–90% B; 4–4.5 min, 90% B; 4.5–5 min,
90–10% B). For compound 18 (kaempferol), different mobile phase B
and gradient were adopted: mobile phase B, 100% methanol; and
gradient, 0 min, 10% B; 0 to 2 min, 10 to 20% B; 2 to 3 min, 20 to 40%
B; 3 to 3.5 min, 40 to 50% B; 3.5 to 4 min, 50 to 70% B; 4 to 5 min, 70
glucuronides were confirmed by the UV spectrum maxima (␭
)
max
shift method (Singh et al., 2010). This method is based on the
characteristic UV shifts caused by glucuronic acid substitution on a
particular hydroxyl group. In brief, if the 3-hydroxyl group on the
flavonol nucleus was glucuronidated, disappearance of UV maxima
or hypsochromic shifts (i.e., to shorter wavelength) were observed in
band I (ϳ300–400 nm). The shift in band I associated with the
substitution was on the order of 13 to 30 nm. Detailed UV data are
shown in Supplemental Fig. S2.
Kinetics Analysis. Kinetic parameters [Vmax, Km, or Ks (sub-
strate inhibition constant)] were estimated by fitting the initial rate
data to Michaelis-Menten and substrate inhibition rate equations by
nonlinear least-squares regression. Data analysis was performed
with Prism version 5 for Windows (GraphPad Software Inc., San
Diego, CA). The goodness of fit was evaluated on the basis of R2
values, residual sum of squares, root mean square (RMS), and resid-
ual plots (Christopoulos and Lew, 2000). The Vmax values obtained
here are apparent values because the actual concentration of
UGT1A9 is unknown.
TABLE 1
Flavonols used in the work (see Fig.1 for chemical structures)
Compound
Name
Abbreviation
R1
R2
1
2
3,2Ј-Dihydroxyflavone
32ЈDHF
33Ј4ЈTHF
34ЈDHF
364ЈTHF
36DHF
37DHF
3H2ЈMF
3H3ЈMF
3H4ЈMF
3H57DMF
3H5MF
3H64ЈDMF
3H6MF
3H7MF
3HF
2Ј-OH
3,3Ј,4Ј-Trihydroxyflavone
3,4Ј-Dihydroxyflavone
3,6,4Ј-Trihydroxyflavone
3,6-Dihydroxyflavone
3,7-Dihydroxyflavone
3-Hydroxy-2Ј-methoxyflavone
3-Hydroxy-3Ј-methoxyflavone
3-Hydroxy-4Ј-methoxyflavone
3-Hydroxy-5,7-dimethoxyflavone
3-Hydroxy-5-methoxyflavone
3-Hydroxy-6,4Ј-dimethoxyflavone
3-Hydroxy-6-methoxyflavone
3-Hydroxy-7-methoxyflavone
3-Hydroxyflavone
3Ј-OH;4Ј-OH
4Ј-OH
3
4
6-OH
6-OH
7-OH
4Ј-OH
5
6
7
2Ј-OMe
3Ј-OMe
4Ј-OMe
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
5-OMe; 7-OMe
5-OMe
6-OMe
4Ј-OMe
6-OMe
7-OMe
3-Hydroxy-6-methyl-3Ј,4Ј-methylenedioxyflavone
Isorhamnetin
Dioxy
6-Me
3Ј,4Ј-methylenedioxy
4Ј-OH; 3Ј-OMe
4Ј-OH
2Ј-OH; 4Ј-OH
3Ј-OH; 4Ј-OH; 5Ј-OH
3Ј-OH; 4Ј-OH
4Ј-OH
Iso
3574ЈQHF
5-OH; 7-OH
5-OH; 7-OH
5-OH; 7-OH
5-OH; 7-OH
5-OH; 7-OH
7-OH
Kaempferol
Morin
Myricetin
Myr
Quercetin
3573Ј4ЈPHF
374ЈTHF
Rha
Resokaempferol
Rhamnetin
5-OH; 7-OMe
4Ј-OH; 3Ј-OH
3Ј-OH
3,3Ј-Dihydroxyflavone
3,5-Dihydroxyflavone
3-Hydroxy-2Ј,3Ј-dimethoxyflavone
3-Hydroxy-6-methylflavone
3-Hydroxy-7,4Ј-dimethoxyflavone
Fisetin
33ЈDHF
35DHF
3H2Ј3ЈDMF
3H6MeF
3H74ЈDMF
5-OH
2Ј-OMe;3Ј-OMe
6-Me
7-OMe
7-OH
4Ј-OMe
3Ј-OH; 4Ј-OH
Galangin
357THF
5-OH; 7-OH
-OH, hydroxyl group; -Me, methyl group; -OMe, methoxyl group.

406
Wu et al.
Pharmacophore Generation and Conformation Search. formed to determine the optimal number of components. The maxi-
Pharmacophore modeling was performed using MOE (molecular op- mal number of components was limited to eight to avoid overfitting.
erating environment) software, version 2008.10 (Chemical Comput- The definitive CoMFA model, which was used for prediction of ac-
ing Group, Montreal, Canada). All of the compounds were drawn on
tivity, was built by noncross-validated partial least-squares analysis
the builder module of MOE and subjected to energy minimization at using the optimal number of components. The q2 (cross-validated r2),
a RMS gradient of 0.00001. The conformational database was cre- cross-validated standard error of prediction, r2 (noncross-validated
ated by “conformation import” using the MMFF94x force field for r2), F values, and S.E. estimate values were computed and are shown
every compound. The force field parameters were kept at their de-
fault values of the strain limit of 4 kcal/mol and the conformations
limit of 250 conformations/molecule. Compound 15 (3-hydroxyfla-
vone) was used as a reference molecule to develop the pharmacoph-
ore query with the “pharmacophore query editor.” A pharmacophore
search using the created query was run against the conformational
database of flavonols. The best-matched conformer for each flavonol
(lowest RMS deviation) was selected, and the alignments were used
for CoMFA.
in Table 3. Pr2 denotes the probability of obtaining the observed F
ratio value by chance alone.
Results
Flavonol 3-O-Glucuronidation Kinetics by UGT1A9.
The kinetics profiles and derived kinetics parameters are
listed in Supplemental Fig. S3 and Table 2, respectively. The
Km values of 3-O-glucuronidation ranged from 0.042 to 0.68
␮M (ϳ1 log unit difference) (Table 2). The unanimous low Km
values (Ͻ 0.7 ␮M) suggested that the flavonol analogs bind
Comparative Molecular Field Analysis. The 30 flavonol mol-
ecules were arbitrarily divided into training (23 compounds) and test
(7 compounds) sets as shown in Table 2. The compounds from the
test set were assigned by considering their substitution patterns so strongly to the UGT1A9 for 3-OH catalysis. The structural
that the test set compounds could reflect the variations in gluc-
uronidation activity of the training set, and both data sets covered
similar diversity in their chemical space. The training set (com-
pounds 1-23; Table 2) was used for CoMFA modeling. The molecular
alignments generated from the pharmacophore model were used in
CoMFA studies using Sybyl8.0 software (Tripos, St. Louis, MO).
Partial charges were recalculated by using the MNDO (modified
neglect of diatomic overlap) method. The molecules were placed in a
three-dimensional grid (2.0-Å spacing). At each grid point, steric
energy (Lennard-Jones potential) and electrostatic energy were cal-
elements contributing to lower Km values (or higher binding
affinity) were identified as: substitutions of -OH at positions
of C3Ј or C4Ј; -OMe groups at C3Ј, C4Ј, C5, or C6; and -Me
group at C6 (Fig. 2A). 3H64ЈDMF (compound 12) had the
lowest Km value of 0.04 in the presence of two Km contribut-
ing elements (i.e., -OMe at both C4Ј and C6). In contrast, -OH
substitution at the 2Ј- or 6-position increased the Km values
by Ͼ1-fold (Fig. 2B). The Km values reported here were not
affected (i.e., increased) by the albumin effects as reported
culated. Cross-validated partial least-squares analysis was per- previously (Rowland et al., 2006).
TABLE 2
Kinetics parameters of UGT1A9 mediated 3-O-glucuronidation with flavonols, together with the predicted Vmax and CLint values from their
respective CoMFA models
Log (Vmax
)
Log (CLint)
Compound
Name
Km
Vmax
CLint (Vmax/Km)
Observeda
Calculatedb
Observeda
Calculatedb
␮M
nmol/mg/min
ml/mg/min
1
2
32ЈDHF
33Ј4ЈTHF
34ЈDHF
364ЈTHF
36DHF
37DHF
3H2ЈMF
3H3ЈMF
3H4ЈMF
3H57DMF
3H5MF
3H64ЈDMF
3H6MF
3H7MF
3HF
0.67
0.10
0.13
0.31
0.62
0.22
0.21
0.063
0.059
0.13
0.082
0.043
0.074
0.27
0.30
0.066
0.32
0.32
0.68
0.61
0.36
0.36
0.23
0.11
0.25
0.52
0.0616
0.19
0.63
0.68
0.052
2.2
1.8
6.8
13
4.6
3.8
2.9
1.9
3.3
1.9
2.2
5.4
10
2.1
6.0
12
1.9
0.040
0.49
3.3
2.6
10
0.078
22
1.72
3.34
3.25
3.84
4.11
3.66
3.58
3.47
3.28
3.52
3.29
3.34
3.73
4.01
3.32
3.78
4.09
3.28
1.60
2.69
3.52
3.42
4.01
3.50
3.27
3.91
3.83
4.25
3.16
3.87
1.68
3.41
3.30
3.94
4.08
3.51
3.59
3.44
3.33
3.70
3.24
3.33
3.66
4.00
3.45
3.69
4.17
3.24
1.62
2.88
3.32
3.35
3.92
3.47
3.48
3.15
3.61
3.80
3.34
3.95
1.89
4.34
4.14
4.34
4.32
4.32
4.25
4.66
4.51
4.40
4.36
4.71
4.86
4.57
3.84
4.96
4.59
3.78
1.77
2.90
3.97
3.86
4.65
4.44
3.87
3.92
5.04
4.66
3.38
4.04
2.31
4.37
4.06
4.29
4.31
4.07
4.08
4.47
4.59
4.52
4.11
4.73
4.84
4.62
4.16
5.20
4.81
3.51
1.86
3.24
3.77
3.98
4.38
4.27
3.74
3.84
4.76
5.02
4.01
3.83
3
14
4
22
5
21
6
21
7
18
8
46
9
32
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24c
25c
26c
27c
28c
29c
30c
26
24
51
73
37
7.0
91
Dioxy
Iso
38
3574ЈQHF
Morin
6.0
0.060
0.80
9.2
7.3
45
Myr
3573Ј4ЈPHF
374ЈTHF
Rha
33ЈDHF
35DHF
3H2Ј3ЈDMF
3H6MeF
3H74ЈDMF
Fisetin
3.1
1.9
4.4
6.75
8.7
1.5
7.4
29
7.4
16
110
46
2.3
11
357THF
a Experimentally determined activities.
b Calculated activities using the CoMFA models.
c These compounds were used as a test set and are not included in the derivation of equations.

Prediction of Flavonol Glucuronidation by UGT1A9 with CoMFA
407
TABLE 3
Summary of modeling parameters from CoMFA analysis
Vmax Model
CLint Model
a
q2
0.738
0.383
0.976
0.735
0.116
6
0.561
0.561
0.938
0.630
0.229
4
SEPb
c
r2
d
rpred
S.E.e
Componentsf
Fg
108.836
0.000
68.436
0.000
P
r2 ϭ 0h
Fraction
Steric
Electrostatic
0.395
0.605
0.450
0.550
a Cross-validated correlation coefficient after the leave-one-out procedure.
b Cross-validated standard error of prediction.
c Non-cross-validated correlation coefficient.
d Correlation coefficient for test set predictions.
e Standard error of estimate.
Fig. 3. Scatter plot of Vmax and Km derived from kinetics of UGT1A9-
mediated 3-O-glucuronidation of flavonols.
f Optimum number of components.
g F-test value.
h Probability of obtaining the observed F ratio value by chance alone.
Fig. 4. Two hypothetically distinct flavonol orientations that are required
to generate 3-O-glucuronide (1 or 3) and 7-O-glucuronide (2 or 4).
Arrows indicate the site of glucuronidation. In 3 and 4, G stands for
site of glucuronidation. A, B, and C indicate A-ring, B-ring, and C-ring,
respectively.
Pharmacophore Modeling. In addition to the gluc-
uronidation site (i.e., hydroxyl group), hydrophobic features
were considered in developing UGT isoform-specific pharma-
cophores, because they had been demonstrated to be critical
for interactions between UGTs and their substrates (Sorich
et al., 2002, 2008). In flavonol structure, the prominent hy-
drophobic features are the aromatic rings: B ring and A/C
bicyclic ring. Conformation analysis revealed that two dis-
tinct orientation modes of flavonol were required to generate
3-O-glucuronide or 7-O-glucuronide (Fig. 4). In this article,
the pharmacophore model was built based on 3-hydroxyfla-
vone to capture the common interaction poses for 3-O-gluc-
uronidation. The final pharmacophore query consisted of
three features: one glucuronidation site (i.e., hydroxyl group),
which is represented by “Don and Acc“ (F1) and two neigh-
Fig. 2. Structural modifications affect the Km values of UGT1A9-medi-
ated 3-O-glucuronidation. A, Km values were decreased in the presence of
1, substitutions of -OH at positions of C3Ј or C4Ј; 2, -OMe groups at C3Ј,
C4Ј, C5, or C6; or 3, -Me group at C6. B, Km values were increased by
Ͼ1-fold with the additions of 2Ј-OH or 6-OH. The compound numbers are
labeled above the abbreviated chemical names.
Vmax and CLint values of 3-O-glucuronidation ranged from boring aromatic regions (F2 and F3), as illustrated graphi-
0.04 to 12.95 nmol/mg/min and 0.06 to 109.60 ml/mg/min, cally (Fig. 5). The “Don and Acc” feature was able to distin-
respectively (Table 2). The turnover of the enzyme (reflected guish hydroxyl oxygen from other types of oxygens in the
by Vmax) varied more toward the flavonols (ϳ2.5 log folds) flavonol structures. The distances from the glucuronidation
than the Km values. Correlation between Km and Vmax is feature to the aromatic regions are 3.9 and 4.2 Å, respec-
depicted in Fig. 3. They seemed to be independent of each tively, and the angle between the glucuronidation feature
other with insignificant r2 value (correlation coefficient) of and the two aromatic regions is 84.2°. The pharmacophore
0.000. Moreover, the ratio of Vmax over Km, CLint, displayed query was used to search against conformation database of
greater divergence (ϳ3.3 log folds). Vmax best describes the 67 flavonoids including those from flavonoid subclasses of
measured susceptibility of chemicals to be glucuronidated, flavones, isoflavones, flavanone, chalcone, and flavonols. All
whereas CLint defines the catalytic efficiency of the enzyme of the flavonols were hit to match the three defined pharma-
toward its substrates at low concentrations. Both Vmax and cophoric features, whereas other flavonoids did not conform
CLint were used to construct CoMFA models.
to this pharmacophore model.

408
Wu et al.
not included in the model construction. The predicted r2
values from the Vmax and CLint CoMFA models were found to
be 0.735 and 0.630, respectively (Table 3). The predicted
activity and experimental activity are listed in Table 2, and
the correlations between them are depicted in Fig. 7.
Figs. 8 and 9 show the steric (in green and yellow) and
electrostatic (in blue and red) contours of CoMFA. The green
contour defines an area where the presence of steric bulky
groups would facilitate the UGT reaction. In contrast, the
yellow contour indicates a region where bulky groups would
diminish glucuronidation (not shown because it is unimpor-
tant). The blue contour denotes a space where UGT1A9 me-
tabolism would benefit from electropositive atoms. Con-
versely, the red contour is a space where the presence of
electronegative atoms would favor glucuronidation. Appar-
ently no sterically disfavored regions (i.e., yellow areas) sur-
rounded the molecules, indicating that the cavities of possi-
ble active site of UGT1A9 was very large. This was not
surprising, because UGT1A9 also efficiently catalyzed com-
pounds that were structurally much larger than flavonols
(Luukkanen et al., 2005).
Vmax CoMFA Model of UGT1A9. The contour map ob-
tained from the Vmax CoMFA model is illustrated together
with compound 19 (poor substrate; Fig. 8A, left) and com-
Fig. 5. A, 3-OH-specific pharmacophore model composed of one glucu-
ronidation site (red sphere with radius of 1 Å) and two aromatic
regions (yellow spheres with radius of 1.2 and 1.5 Å, respectively). B,
3-OH-specific pharmacophore model superimposed with 3HF (com-
pound 15).
CoMFA Models. CoMFA modeling was performed using
flavonol alignments based on a 3-OH-specific pharmacophore
(Fig. 6). The flavonol conformers were presumed to be in the
bioactive poses that interact with UGT1A9 to produce 3-O-
glucuronides. CoMFA resulted in high-quality models for
UGT1A9 (Vmax: q2 ϭ 0.738, r2 ϭ 0.976; CLint: q2 ϭ 0.561, r2 ϭ
0.938) (Table 3). For the Vmax model, the steric field descrip-
tors explained 39.5% of the variance, whereas the electro-
static descriptors explained 60.5%. For the CLint model, the
contributing proportions of steric and electrostatic fields to
the variance were 45 and 55%, respectively. These models
were validated by an external test set of seven compounds
Fig. 7. Correlation between the experimental glucuronidation parame-
ters and the predicted ones from the CoMFA models for the training and
test sets (Vmax, A; CLint, B).
Fig. 6. Structural alignment for constructing the 3D QSAR CoMFA
model generated from the 3-OH-specific pharmacophore.

Prediction of Flavonol Glucuronidation by UGT1A9 with CoMFA
409
Fig. 8. A, steric and electrostatic maps from the UGT1A9 CoMFA model of Vmax. Morin (compound 19; left) and isorhamnetin (compound 17; right)
are shown inside the field for reference. Green indicates areas in which bulky groups are sterically favorable for glucuronidation. Blue indicates areas
in which electropositive atoms are favorable for glucuronidation. Red indicates areas in which electronegative atoms are favorable for glucuronidation.
B, matching of the CoMFA to experimental data. I, bulky groups at C6 increased Vmax values. II, 7-OH or 7-OMe increased Vmax values. III, 2Ј-OMe
increased Vmax values. IV1, 3-OH or 3-OMe increased Vmax values. IV2, 4-OH or 4-OMe decreased Vmax values. The compound numbers are labeled
above the abbreviated chemical names.
pound 17 (good substrate; Fig. 8A, right). In the CoMFA the properties of the substrates and Vmax values, were eluci-
contour map, the green (sterically favorable) and yellow dated as follows.
(sterically unfavorable) contours represent 80 and 20% con-
In region I, the steric favorable green contour near the C6
tributions, respectively. The blue (electropositive atom favor- indicates that the bulky group in the contour is important for
able) and red (electronegative atom favorable) contours in the glucuronidation. This can explain why the Vmax values of
CoMFA electrostatic field make 80 and 20% contributions, 36DHF (6-OH), 3H6MF (6-OMe), and 3H6MeF (6-Me) are
respectively. Four regions (designated as I, II, III, and IV) generally higher than that of 3HF (6-H) (Fig. 8BI). There is
that contain the contours, indicating the relations between also a red contour positioned (near the green proportion) a bit

410
Wu et al.
Fig. 9. A, steric and electrostatic maps from the UGT1A9 CoMFA model of CLint. Morin (compound 19; left) and dioxy (compound 16; right) are shown
inside the field for reference. Green indicates areas in which bulky groups are sterically favorable for glucuronidation. Blue indicates areas in which
electropositive atoms are favorable for glucuronidation. Red indicates areas in which electronegative atoms are favorable for glucuronidation. B,
matching of the CoMFA to experimental data. I, effect of 5-OMe on CLint was not well defined. II, bulky groups at C6 increased CLint values. III, 7-OMe
increased CLint values. IV, effect of 2Ј-OMe on CLint was not well defined. V1, 3-OH increased CLint values. V2, effect of 4Ј-OMe on CLint was not well
defined. The compound numbers are labeled above the abbreviated chemical names.

Prediction of Flavonol Glucuronidation by UGT1A9 with CoMFA
411
further from the C6. This contour can be reached and occu-
In region II, a green contour presents around C6. It was
pied by the hydrogens (electropositive atom) of methoxy reasoned that steric bulks substituted at C6 would enhance
group substituted at C6. The placement of electropositive conjugation activity with higher CLint. This is supported by
atoms in this red area would disfavor the glucuronidation, the fact that 36DHF (6-OH), 3H6MF (6-OMe), and 3H6MeF
which is consistent with the fact that 3H6MF (6-OMe) had (6-Me) showed much higher catalytic efficiency than 3HF
lower Vmax than those of 36DHF (6-OH) and 3H6MeF (6-Me) (6-H), and 364ЈTHF (6-OH) had a higher CLint value than
(Fig. 8BI).
34ЈTHF (6-H) (Fig. 9BII).
In region II, a small red polyhedron (occupied by electro-
In region III, the blue polyhedron near C7 suggested that
negative oxygen of 7-OH or 7-OMe) and a blue contour (oc- the placement of electropositive atoms (e.g., hydrogens of
cupied by hydrogens of 7-OMe) aligned diagonally along the 7-OMe) would increase the metabolism. This is in agreement
C7. It suggests that an electronegative atoms in the red with the observations (ordered in term of CLint values):
polyhedron and/or electropositive atoms in the blue contour 3H7Me (7-OMe) Ͼ 3HF (7-H); 3H74ЈDMF (7-OMe) Ͼ
are favorable for UGT1A9 glucuronidation. Therefore, the 3H4ЈMF (7-H); and 3H57DMF (7-OMe) Ͼ 3H5MF (7-H) (mi-
presence of 7-OH or 7-OMe would contribute to higher Vmax nor difference) (Fig. 9BIII).
values. This is supported by the comparisons between the
In region IV, a red contour appears in the vicinity of C2Ј,
structures and Vmax values: 3HF (7-H) Ͻ 37DHF (7-OH) Ͻ which indicates electropositive atoms placed in this volume
3H7MF (7-OMe); 34ЈDHF (7-H) Ͻ 374ЈTHF (7-OH); 35DHF would be metabolism-unfavorable. This might explain why
(7-H) Ͻ 357THF(7-OH); 3H5MF(7-H) Ͻ 3H57DMF (7-OMe) 32ЈDHF (2Ј-OH) and morin (2Ј-OH) showed the poorest gluc-
and 3H4ЈMF (7-H)Ͻ 3H74ЈDMF (7-OMe) (Fig. 8BII).
uronidation with the smallest CLint values (Table 2). A
In region III, the red contour slightly further away from nearby large blue area is shown to cover the electropositive
the C2Ј suggested an electropositive atom (or hydrogen at- hydrogens of 2Ј-OMe. The presence of 2Ј-OMe therefore is
tached to -OH) substitution would compromise the gluc- predicted to enhance the glucuronidation, which is experi-
uronidation, as seen from the data that compound 1, mentally interpreted by the fact that 3H2ЈMF (2Ј-OMe) was
32ЈDHF, and compound 19, morin (with 2Ј-OH), had the more efficiently glucuronidated than 3HF (2Ј-H). Unexpect-
smallest Vmax values (Table 2). A nearby large blue area that edly, 3H2Ј3ЈMF (2Ј-OMe) had a CLint value smaller than that
is occupied by multiple electropositive hydrogen atoms of of 3H3ЈMF (2Ј-H) (Fig. 9BIV).
2Ј-OMe indicated that 2Ј-OMe contributes significantly to the
In region V, a big red contour shows around C3Ј/C4Ј. There
glucuronidation. For this reason, 3H2ЈMF (2Ј-OMe) and is also a small red contour that is occupied by oxygen of
3H2Ј3ЈDMF (2Ј-OMe) had higher Vmax values than those of 3Ј-OH. This can be indicate that glucuronidation would ben-
3HF (2Ј-H) and 3H3ЈMF (2Ј-H), respectively (Fig. 8BIII).
efit from the electronegative atoms (e.g., oxygen of -OH) in
For region IV, the big red contour is in the close vicinity of the area, as evidenced by the fact that 3HF (3Ј-H), 34ЈDHF
C3Ј, suggesting that glucuronidation would benefit from elec- (3Ј-H) and 3574ЈQHF (3Ј-H) were less metabolized (smaller
tronegative atoms such as oxygen in hydroxyl or methoxy Clint values) than 33ЈDHF (3Ј-OH), 33Ј4ЈTHF (3Ј-OH), and
groups placed in this area. This is consistent with the fact 3573Ј4ЈPHF (3Ј-OH), respectively (Fig. 9BV1). A blue contour
that Vmax values are increased in the presence of 3Ј-OH or situates in the vicinity of C4Ј, which can be occupied by
3Ј-OMe: 3HF (3Ј-H) Ͻ 33ЈDHF (3Ј-OH) or 3H3ЈMF (3Ј-OMe); electropositive hydrogens of the 4Ј-OMe group. It is sug-
3H2ЈMF (3Ј-H) Ͻ 3H2Ј3ЈDMF (3Ј-OMe); and 3574ЈQHF (3Ј- gested that CLint will increased in the presence of 4Ј-OMe, as
H) Ͻ 3573Ј4ЈPHF (3Ј-OH) (Fig. 8BIV1). There is a small red supported by the fact that 3HF (4Ј-H) and 3H7MF (4Ј-H)
contour (close to the hydrogen of 4Ј-OH) that can be occupied were less efficiently glucuronidated by 3H4ЈMF (4Ј-OMe) and
by electropositive hydrogens of 4Ј-OMe. It is suggested that 3H74ЈDMF (4Ј-OMe), respectively. However, the addition of
electropositive atoms in or near this contour will be detri- 4Ј-OMe in 3H64ЈMF did not result in a higher CLint value
mental to the glucuronidation. Therefore, 4Ј-OH or 4Ј-OMe compared with 3H6MF (4Ј-H) (Fig. 9BV2).
most likely would reduce the value of Vmax, which is evi-
denced from the observations: 3HF (4Ј-H) Ͼ 34ЈDHF (4Ј-OH)
or 3H4ЈMF(4Ј-OMe); 37DHF (4Ј-H) Ͼ 374ЈTHF (4Ј-OH);
Discussion
36DHF (4Ј-H)
3H64ЈDMF(4Ј-OMe); 3H7MF (4Ј-H) Ͼ 3H74ЈDMF (4Ј-OMe); models for glucuronidation of flavonols by human UGT1A9
and 357THF (4Ј-H) Ͼ 3574ЈQHF (4Ј-OH) (Fig. 8BIV2). that focused on predicting Vmax and CLint of 3-O-gluc-
Ͼ
364ЈTHF (4Ј-OH); 3H6MF(4Ј-H)
Ͼ
We have constructed successfully for the first time in silico
CLint CoMFA Model of UGT1A9. The contour map ob- uronidation. 2D/3D QSAR models developed for UGTs
tained from the CLint CoMFA model is shown with compound (Sorich et al., 2002, 2008; Smith et al., 2004) have predicted
19 (poor substrate; Fig. 9A, left) and compound 16 (good the substrate selectivity and/or binding affinity of inhibitors
substrate; Fig. 9A, right). It is observed that five regions [as reflected by low apparent inhibitor constant (Ki,app)].
(designated as region I, II, III, IV, and V) contain the con- However, Vmax or CLint is more relevant for defining the
tours, indicating the relationships between the properties susceptibility of chemicals to be metabolized and reflect the
of the substituents and CLint values. In region I, the blue in vivo biotransformation efficiency by individual human
contour near C5 suggested that electropositive entities UGT isoforms (at low physiological concentrations at or be-
placed in this area would result in higher CLint values. This low 1 ␮M), because higher substrate affinity toward UGTs
can be demonstrated by the fact that 3H5MF (5-OMe) is more does not always translate to faster glucuronidation rates. For
efficiently metabolized than 3HF (5-H) or 35DHF (5-OH) example, 3-hydroxyflavone (Ki,app ϭ 3.5 ␮M) binds much
(Fig. 9BI). However, 3H57DMF (5-OMe) had a smaller CLint strongly to UGT1A9 than naringenin (Ki,app ϭ 219 ␮M)
than 3H7MF (5-H), indicating some uncertainties at this (Smith et al., 2004), but 3-hydroxyflavone (1.99 Ϯ 0.09 nmol/
position.
mg/min) was glucuronidated slower than naringenin (3.26 Ϯ

412
Wu et al.
0.07 nmol/mg/min) at 10 ␮M by UGT1A9 (rates at additional This structural property was also shown to be the key de-
concentrations are in Supplemental Materials). The rele- scriptor in the successful modeling of UGT1A9-catalyzed gluc-
vance of using Vmax or CLint would be even greater if uronidation of phenols (Ethell et al., 2002). This important
UGT1A9 protein levels in recombinant microsomes and hu- information will be used to guide future descriptor selection
man tissue microsomes could be quantified.
or model refinement in 2D/3D QSAR modeling of UGT1A9
This article validates our approach of quantitatively de- using a more diverse dataset, because UGT1A9 was shown to
scribing the position (3-OH)-specific (or regiospecific) gluc- present more challenges for in silico modeling (Smith et al.,
uronidation using in silico models. It directly tackles the 2004).
challenges of predicting UGT metabolism of flavonols in
The success of CoMFA modeling in turn corroborated the
silico that potentially generate multiple conjugates. Success- underlying assumption about the alignment of bioactive
ful implementation of this approach to other hydroxyl groups poses, suggesting the aglycones interact with UGT1A9 pro-
will allow the estimation of overall glucuronidation of multi- tein using a similar mode to generate 3-O-glucuronide.
hydroxyl flavonols (by summation of predicted values from Hence, it lends strong support to the hypothesis that multi-
the separate position specific models). It is consistent with ple distinct binding modes are required to generate different
the hypothesis of Miners et al. (2004), who stated that “mul- glucuronide isomers by UGT1A9, if not all UGT1A, isoforms.
tiple binding modes within the aglycone-binding domain are It is noteworthy that plant UGTs (e.g., UGT71G1, VvGT1,
required to generate multiple metabolites from single sub- and UGT78G1) with crystal structures possess the big cata-
strate that bears more than one nucleophilic group.” Because lytic cavities in the aglycone-binding domain. The binding
available 3D QSAR algorithms are based on only one pre- pocket with sufficient space permits the distinct orientations
sumed bioactive conformation corresponding to each ligand from single flavonoids that render the respective hydroxyl
(i.e., one binding mode), it hinders one-step prediction of groups for conjugation (Osmani et al., 2009). Because of the
formation of multiple monoglucuronides from a single sub- marked similarity of catalytic mechanisms between human
strate with more than one nucleophilic site. This limitation and plant UGTs (Patana et al., 2008), UGT1A9 may also
can be readily observed in the pharmacophore models. Map- have a large aglycone-binding domain that serves as the
ping of quercetin into UGT1A9 pharmacophore generated by molecular basis for generating multiple metabolites. This
using the substrates with great structural diversity (Miners assumed large binding pocket is also supported by the
et al., 2004) indicates 3Ј-OH is the only site of glucuronida- CoMFA contours that did not show any steric disfavoring
tion, which is in conflict with the fact that UGT1A9 generates areas (Figs. 8 and 9). Therefore, only an algorithm that
multiple glucuronides from quercetin (Chen et al., 2005; considers all of the active conjugation sites will allow accept-
Singh et al., 2010).
able prediction of the overall glucuronidation rates of fla-
We chose to model 3-O-glucuronidation of flavonols using vonols with multiple hydroxyl groups. Because different hy-
CoMFA, a widely recognized 3D QSAR technique for model- droxyl groups of a flavonol or other flavones may be
ing biological properties. The main advantages of CoMFA predominantly metabolized by a particular isoform, this
include: 1) it has the ability to display the model graphically type of modeling must be extended to multiple UGT iso-
and 2) it allows inference regarding binding pocket geometry. forms if we were able to successfully extend this approach
Up to 3155 hits were generated by a PubMed search (con- to predict the metabolism of flavonols at all sites.
ducted on August 20, 2010) using “comparative molecular
An interesting discovery is that the linkage between Km
field analysis” or “CoMFA” as keywords. In contrast, linear or and the enzyme turnover is weak if they are present at all
nonlinear regression models based on 2D/3D descriptors (Fig. 3). Our data clearly showed that compared with Vmax or
showed good predictability, but the interpretation of impor- CLint, Km is less susceptible to minor structural changes. In
tant features and descriptors is difficult (Sorich et al., 2008). addition, kinetics profiling of positional (3-O- and 7-O-) gluc-
As stated by Chohan et al. (2006), the QSAR model is more uronidation showed identical Km but divergent Vmax (data
like a “black box,” with only computers able to interpret not shown). This might indicate that the UGT1A9 protein
which molecular properties modulate the metabolism.
adopts distinct conformations for aglycone binding and prod-
The predictive CoMFA models of UGT1A9 can be used to uct expelling/releasing, because the turnover is determined
guide the design of novel flavonol with desired UGT metab- in a large part by the departure of product (3-O-gluc-
olism or predict 3-O-glucuronidation of untested analogs. uronides). This hypothesis is supported by the fact that
From the perspective of reducing metabolism, introduction of UGTs undergo dramatic conformation changes during catal-
multiple electropositive atoms or groups (e.g., hydrogens of ysis (Laakkonen and Finel, 2010). Therefore, as mentioned
-OH) to the B-ring would result in poor 3-O-glucuronidation earlier, conventional use of Km or Ki as indicators of sub-
by UGT1A9. On the contrary, fast conjugation of 3-OH may strate selectivity might lead to erroneous interpretation of
result from substrates with bulky substituents at C6 or elec- interaction between substrates and UGT1As. Although inde-
tronegative atoms or groups (e.g., -Cl, -F, and -O of methoxy pendent from Vmax, Km influenced the modeling results be-
group) around the B-ring. As advancement is made in quan- tween Vmax and CLint. For example, 4Ј-OMe is unfavorable
titation technologies of specific UGT isoforms, it is envi- for Vmax, but favorable for CLint (Figs. 8IV and 9V2).
sioned that successful prediction of glucuronidation by
The predictability of the current model might be compro-
major metabolizing isoforms would enable the precise esti- mised by the unevenly distributed activity [e.g., existing gap
mation of the UGT metabolism at the tissue/organ level (e.g., of 1.8–2.8 for Log (Vmax)], especially when those compounds
liver, intestine, and kidney) (Tang et al., 2010). In addition, that are to be predicted have activities fall into the gaps.
the CoMFA model highlighted the significant role of electro- Refinement of the model seems to be essential by adding
static potential (i.e., electronegative versus electropositive those flavonols whose activity can fill in those gaps. However,
entity) in determining UGT1A9-mediated glucuronidation. considering the remarkable similarity of the chemical struc-

Prediction of Flavonol Glucuronidation by UGT1A9 with CoMFA
413
vitro approaches to elucidate the thermal stability of human UDP-glucuronosyl-
transferase (UGT) 1A9. Drug Metab Pharmacokinet 24:235–244.
Haji-Momenian S, Rieger JM, Macdonald TL, and Brown ML (2003) Comparative
molecular field analysis and QSAR on substrates binding to cytochrome p450 2D6.
Bioorg Med Chem 11:5545–5554.
Holder S, Lilly M, and Brown ML (2007) Comparative molecular field analysis of
flavonoid inhibitors of the PIM-1 kinase. Bioorg Med Chem 15:6463–6473.
Iyanagi T (2007) Molecular mechanism of phase I and phase II drug-metabolizing
enzymes: implications for detoxification. Int Rev Cytol 260:35–112.
Jancova P, Anzenbacher P, and Anzenbacherova E (2010) Phase II drug metaboliz-
ing enzymes. Biomed Pap Med Fac Univ Palacky Olomouc Czech Repub 154:103–
116.
tures between the modeled compounds, but divergent activ-
ity spanning ϳ 3 log orders, the model seemed to be able to
sufficiently capture the key chemical characteristics associ-
ated with the modeled parameters. This is evidenced by the
establishment of a predictive model with strong statistical
significance, as well as the consistency between experimental
results and the contour maps. Therefore, the current CoMFA
models are insightful with acceptable predictability.
In conclusion, 3D QSAR study of UGT1A9 flavonols was
carried out using pharmacophore-based CoMFA. The con-
structed CoMFA models possessed good internal and exter-
nal consistency and showed statistical significance and pre-
Joseph TB, Wang SW, Liu X, Kulkarni KH, Wang J, Xu H, and Hu M (2007)
Disposition of flavonoids via enteric recycling: enzyme stability affects character-
ization of prunetin glucuronidation across species, organs, and UGT isoforms. Mol
Pharm 4:883–894.
Laakkonen
L and Finel M (2010) A molecular model of the human UDP-
glucuronosyltransferase 1A1, its membrane orientation, and the interactions be-
tween different parts of the enzyme. Mol Pharmacol 77:931–939.
Luukkanen L, Taskinen J, Kurkela M, Kostiainen R, Hirvonen J, and Finel M (2005)
Kinetic characterization of the 1A subfamily of recombinant human UDP-
glucuronosyltransferases. Drug Metab Dispos 33:1017–1026.
Mackenzie PI, Bock KW, Burchell B, Guillemette C, Ikushiro S, Iyanagi T, Miners
JO, Owens IS, and Nebert DW (2005) Nomenclature update for the mammalian
UDP glycosyltransferase (UGT) gene superfamily. Pharmacogenet Genomics 15:
677–685.
Miners JO, Smith PA, Sorich MJ, McKinnon RA, and Mackenzie PI (2004) Predicting
human drug glucuronidation parameters: application of in vitro and in silico
modeling approaches. Annu Rev Pharmacol Toxicol 44:1–25.
Ohno S and Nakajin S (2009) Determination of mRNA expression of human UDP-
glucuronosyltransferases and application for localization in various human tissues
by real-time reverse transcriptase-polymerase chain reaction. Drug Metab Dispos
37:32–40.
Osmani SA, Bak S, and Møller BL (2009) Substrate specificity of plant UDP-
dependent glycosyltransferases predicted from crystal structures and homology
modeling. Phytochemistry 70:325–347.
Otake Y, Hsieh F, and Walle T (2002) Glucuronidation versus oxidation of the
flavonoid galangin by human liver microsomes and hepatocytes. Drug Metab
Dispos 30:576–581.
Patana AS, Kurkela M, Finel M, and Goldman A (2008) Mutation analysis in
UGT1A9 suggests a relationship between substrate and catalytic residues in
UDP-glucuronosyltransferases. Protein Eng Des Sel 21:537–543.
Ross JA and Kasum CM (2002) Dietary flavonoids: bioavailability, metabolic effects,
and safety. Annu Rev Nutr 22:19–34.
Rowland A, Elliot DJ, Williams JA, Mackenzie PI, Dickinson RG, and Miners JO
(2006) In vitro characterization of lamotrigine N2-glucuronidation and the lam-
otrigine-valproic acid interaction.Drug Metab Dispos 34:1055–1062.
Said M, Ziegler JC, Magdalou J, Elass A, and Vergoten G (1996) Inhibition of
bilirubin Udp-glucuronosyltransferase: a comparative molecular field analysis
(COMFA). Quant Struct Act Relat 15:382–388.
dictive abilities (Vmax model: q2 ϭ 0.738, r2 ϭ 0.976, rp2red
ϭ
0.735; CLint model: q2 ϭ 0.561, r2 ϭ 0.938, r2pred ϭ 0.630). The
contour maps from CoMFA clearly indicated key structural
characteristics (e.g., electropositive entities at C2Ј or C3Ј)
that were associated with poor 3-O-glucuronidation. The
results suggested that the approach of coupling CoMFA
analysis with pharmacophoric alignments is viable for con-
structing predictive models regarding regiospecific or 3-O-
glucuronidation of flavonols by UGT1A9.
Acknowledgments
We thank Dr. Xiaoqiang Wang of the Nobel Foundation for sug-
gestions and comments about this article.
Authorship Contributions
Participated in research design: Wu, Zhang, and Hu.
Conducted experiments: Wu, Morrow, and Singh.
Contributed new reagents or analytic tools: Morrow and Zhang.
Performed data analysis: Wu and Hu.
Wrote or contributed to the writing of the manuscript: Wu and Hu.
Other: Hu acquired funding for the research.
References
Singh R, Wu B, Tang L, Liu Z, and Hu M (2010) Identification of the position of
mono-O-glucuronide of flavones and flavonols by analyzing shift in online UV
spectrum (␭max) generated from an online diode-arrayed detector. J Agric Food
Chem 58:9384–9395.
Smith PA, Sorich MJ, Low LS, McKinnon RA, and Miners JO (2004) Towards
integrated ADME prediction: past, present and future directions for modelling
metabolism by UDP-glucuronosyltransferases. J Mol Graph Model 22:507–517.
Smith PA, Sorich MJ, McKinnon RA, and Miners JO (2003) In silico insights:
chemical and structural characteristics associated with uridine diphosphate glu-
curonosyltransferase substrate selectivity. Clin Exp Pharmacol Physiol 30:836–
840.
Andersen OM and Markham KR (2006) Flavonoids – Chemistry, Biochemistry and
Applications. CRC Taylor & Francis, Boca Raton, FL.
Barreca ML, Carotti A, Carrieri A, Chimirri A, Monforte AM, Calace MP, and Rao A
(1999) Comparative molecular field analysis (CoMFA) and docking studies of
non-nucleoside HIV-1 RT inhibitors (NNIs). Bioorg Med Chem 7:2283–2292.
Barve A, Chen C, Hebbar V, Desiderio J, Saw CL, and Kong AN (2009) Metabolism,
oral bioavailability and pharmacokinetics of chemopreventive kaempferol in rats.
Biopharm Drug Dispos 30:356–365.
Birt DF, Hendrich S, and Wang W (2001) Dietary agents in cancer prevention:
flavonoids and isoflavonoids. Pharmacol Ther 90:157–177.
Sorich MJ, Smith PA, McKinnon RA, and Miners JO (2002) Pharmacophore and
quantitative structure activity relationship modelling of UDP-glucuronosyltrans-
ferase 1A1 (UGT1A1) substrates. Pharmacogenetics 12:635–645.
Sorich MJ, Smith PA, Miners JO, Mackenzie PI, and McKinnon RA (2008) Recent
advances in the in silico modelling of UDP glucuronosyltransferase substrates.
Curr Drug Metab 9:60–69.
Tang L, Singh R, Liu Z, and Hu M (2009) Structure and concentration changes affect
characterization of UGT isoform-specific metabolism of isoflavones. Mol Pharm
6:1466–1482.
Tang L, Ye L, Singh R, Wu B, Lv C, Zhao J, Liu Z, and Hu M (2010) Use of
glucuronidation fingerprinting to describe and predict mono- and dihydroxyfla-
vone metabolism by recombinant UGT isoforms and human intestinal and liver
microsomes. Mol Pharm 7:664–679.
Wei DG, Yang GF, Wan J, and Zhan CG (2005) Binding model construction of
antifungal 2-aryl-4-chromanones using CoMFA, CoMSIA, and QSAR analyses. J
Agric Food Chem 53:1604–1611.
Chen Y, Xie S, Chen S, and Zeng S (2008) Glucuronidation of flavonoids by recom-
binant UGT1A3 and UGT1A9. Biochem Pharmacol 76:416–425.
Chen YK, Chen SQ, Li X, and Zeng S (2005) Quantitative regioselectivity of glucu-
ronidation of quercetin by recombinant UDP-glucuronosyltransferases 1A9 and
1A3 using enzymatic kinetic parameters. Xenobiotica 35:943–954.
Chohan KK, Paine SW, and Waters NJ (2006) Quantitative structure activity rela-
tionships in drug metabolism. Curr Top Med Chem 6:1569–1578.
Christopoulos A and Lew MJ (2000) Beyond eyeballing: fitting models to experimen-
tal data. Crit Rev Biochem Mol Biol 35:359–391.
Cramer RD III, Patterson DE, and Bunce JD (1988) Comparative molecular field
analysis (CoMFA). 1. Effect of shape on binding of steroids to carrier proteins.
J Am Chem Soc 110:5959–5967.
D’Archivio M, Filesi C, Di Benedetto R, Gargiulo R, Giovannini C, and Masella R
(2007) Polyphenols, dietary sources and bioavailability. Ann Ist Super Sanita
43:348–361.
Erlund I, Freese R, Marniemi J, Hakala P, and Alfthan G (2006) Bioavailability of
quercetin from berries and the diet. Nutr Cancer 54:13–17.
Ethell BT, Ekins S, Wang J, and Burchell B (2002) Quantitative structure activity
relationships for the glucuronidation of simple phenols by expressed human
UGT1A6 and UGT1A9. Drug Metab Dispos 30:734–738.
Wong YC, Zhang L, Lin G, and Zuo Z (2009) Structure-activity relationships of the
glucuronidation of flavonoids by human glucuronosyltransferases. Expert Opin
Drug Metab Toxicol 5:1399–1419.
Fisher MB, Paine MF, Strelevitz TJ, and Wrighton SA (2001) The role of hepatic and
extrahepatic UDP-glucuronosyltransferases in human drug metabolism. Drug
Metab Rev 33:273–297.
Address correspondence to: Dr. Ming Hu, Department of Pharmacological
and Pharmaceutical Sciences, College of Pharmacy, University of Houston,
1441 Moursund Street, Houston, TX 77030. E-mail: mhu@uh.edu
Fujiwara R, Nakajima M, Yamamoto T, Nagao H, and Yokoi T (2009) In silico and in