Structure-activity relationship of the inhibitory effects of flavonoids on nitric oxide production in RAW264.7 cells

Article abstract of DOI:10.1016/j.bmc.2016.11.055

We isolated flavonoids from herbal specimens from the Tibetan region (Sophora yunnanensis and Rhodiola sacra) that suppress nitric oxide (NO) production in macrophages stimulated by lipopolysaccharide and interferon-γ. The isolated flavonoids carry symmetric substitutions in the B ring (R_3′= R_5′). We analyzed the quantitative structure-activity relationship of the inhibitory activity by comparative molecular field analysis (CoMFA) using this series of flavonoids. Use of flavonoids with symmetrical substitutions in the B ring made it simpler to align molecules because it was not necessary to consider a huge number of combinations due to the B-ring conformation. The CoMFA model, whose cross-validated q²value was 0.705, suggested the existence of a hydroxy group at the 5-position, the choice of the A/C-ring scaffold (chromane or chromene) and electrostatic field around the B ring are important for NO inhibitory activity. Flavonoids synthesized based on the CoMFA model exhibited significant inhibitory potential against NO production, validating the predictive capability of the CoMFA model.

Full text of DOI:10.1016/j.bmc.2016.11.055

Accepted Manuscript
Structure-activity relationship of the inhibitory effects of flavonoids on nitric
oxide production in RAW264.7 cells
Wen-Jun Jiang, Akihiro Daikonya, Mitsuyoshi Ohkawara, Takashi Nemoto,
Ryusuke Noritake, Tomoko Takamiya, Susumu Kitanaka, Hiroshi Iijima
PII:
S0968-0896(16)30949-X
http://dx.doi.org/10.1016/j.bmc.2016.11.055
BMC 13422
DOI:
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
13 October 2016
26 November 2016
28 November 2016
Please cite this article as: Jiang, W-J., Daikonya, A., Ohkawara, M., Nemoto, T., Noritake, R., Takamiya, T.,
Kitanaka, S., Iijima, H., Structure-activity relationship of the inhibitory effects of flavonoids on nitric oxide
production in RAW264.7 cells, Bioorganic & Medicinal Chemistry (2016), doi: http://dx.doi.org/10.1016/j.bmc.
2016.11.055
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Structure-activity relationship of the inhibitory effects of flavonoids on nitric oxide
production in RAW264.7 cells
Wen-Jun Jiang, Akihiro Daikonya†, Mitsuyoshi Ohkawara, Takashi Nemoto, Ryusuke Noritake,
Tomoko Takamiya, Susumu Kitanaka and Hiroshi Iijima*
School of Pharmacy, Nihon University, 7-7-1 Narashino-dai, Funabashi-shi, Chiba 274-8555,
Japan
Corresponding Author
*Hiroshi Iijima. School of Pharmacy, Nihon University, 7-7-1 Narashino-dai, Funabashi-shi,
Chiba 274-8555, Japan. E-mail: iijima.hiroshi@nihon-u.ac.jp. Phone/Fax: +81 47 465 6470.
Present Addresses
†Kotaro Pharmaceutical Co., Ltd., Ro 96-1, Kashima-machi, Hakusan, Ishikawa 929-0201,
Japan 929-0201
Key Words
flavonoid; nitric oxide; macrophage; structure-activity relationship; comparative molecular field
analysis
1

Abstract
We isolated flavonoids from herbal specimens from the Tibetan region (Sophora yunnanensis
and Rhodiola sacra) that suppress nitric oxide (NO) production in macrophages stimulated by
lipopolysaccharide and interferon-. The isolated flavonoids carry symmetric substitutions in the
B ring (R3’ = R5’). We analyzed the quantitative structure-activity relationship of the inhibitory
activity by comparative molecular field analysis (CoMFA) using this series of flavonoids. Use of
flavonoids with symmetrical substitutions in the B ring made it simpler to align molecules
because it was not necessary to consider a huge number of combinations due to the B-ring
conformation. The CoMFA model, whose cross-validated q2 value was 0.705, suggested the
existence of a hydroxy group at the 5-position, the choice of the A/C-ring scaffold (chromane or
chromene) and electrostatic field around the B ring are important for NO inhibitory activity.
Flavonoids synthesized based on the CoMFA model exhibited significant inhibitory potential
against NO production, validating the predictive capability of the CoMFA model.
1. Introduction
Inflammation is one of important syndromes that relates to a wide range of modern diseases
such as diabetes and circulatory disease. Many trials have been undertaken to identify
compounds from natural resources that suppress or mitigate the progress of inflammation,
especially in herbal medicine. Flavonoids are often designated as the active constituents in many
traditional medicinal plants. The anti-inflammatory effects of flavonoids are explored by many in
vitro surrogate assays such as inhibitory tests of nitric monoxide (NO) production in
2

macrophages, inhibitory tests of the chemotaxis of neutrophils or suppression tests on the
translation of inflammation-related genes.1-4
Structure-activity relationship studies on the inhibitory effects of flavonoids on NO
production were performed by many researchers.5-8 However, due to the structural diversity of
the flavonoids, these studies remained qualitative.
During our research into anti-inflammatory flavonoids in Sophora yunnanensis and Rhodiola
sacra, we isolated the flavonoids shown in Fig. 1.9,10 Tricin 1 and apigenin 2 are flavones
(chromen-4-one scaffold). Compound 3, laurentinol 4, kaempferol 5 and robinetin 6 are
flavonols (3-hydroxychromen-4-one scaffold). Sepinol 7, sophorayunnanol 8, dihydrokaempferol
9 and dihydrorobinetin 10 are flavanonols (3-hydroxychroman-4-one scaffold). Naringenin 11,
compound 12 and robitin 13 are flavanones (chroman-4-one scaffold). The B ring of these
flavonoids happens to possess symmetrically arranged substituents (R3’ = R5’). We determined
their potential to inhibit NO production in Raw264.7 cells, a murine macrophage cell line, and
intended to build a quantitative structure-activity relationship (QSAR) model.
3

Figure 1. Structure and activity of flavonoids isolated from Sophora
yunnanensis and Rhodiola sacra. Flavonoids are categorized based on the
skeleton and substitution pattern of the B ring. IC50 values of the flavonoids for
NO production in Raw246.7 cells are also indicated. Flavonoids 1, 2, 5, 9, 11
and 12 were isolated from R. sacra, and flavonoids 3, 4, 6, 7, 8, 10 and 13 were
isolated from S. yunnanensis. Atom numbering is shown in the structure of
apigenin 2. The absolute configuration of flavanonols 7-10 and of flavanones 11-
13 are (2R, 3R) and (2S), respectively.
4

One reason making a structure-activity study of flavonoids difficult is the complexity caused
by the conformation of the B ring. For example, as shown in Fig. 2, there are four possible
combinations to overlap or align flavonoids a and b. However, if A = C and X = Z, only one
alignment is available for overlapping the two flavonoids. Thus, the use of flavonoids having
symmetrically substituted B rings reduces the number of possible combinations of B-ring
alignments and is advantageous for 3-D structure-activity relationship studies. We conducted a
comparative molecular field analysis (CoMFA) study.11 The CoMFA model was validated by (i)
predicting the activity of flavonoids of known activity reported by Matsuda et al.12 and by (ii)
synthesis of flavones that are expected to be active followed by evaluation of their biological
activity.
Figure 2. Symmetric B rings make alignment of molecules simpler. Provided that
flavones a and b have asymmetrically substituted B rings (A ≠ C and X ≠Z), we
need to consider two conformers each from which four possible combinations arise
when the flavonoid molecules are aligned (or superimposed) in space. The number
of combinations grows drastically when multiple molecules are to be aligned. If
both flavonoids carry symmetric B rings, such complexity is nonexistent.
5

2. Materials and methods
2.1 Measurement of NO Production by Macrophages in vitro.
Details of the assay method performed are as described previously.9,10 In brief, RAW264.7 cells
were stimulated with LPS (100 ng/mL) and IFN- (0.33 ng/mL). The released nitrite (NO) was
quantified by the Greiss method. Cell viability was measured using AlamarBlue® reagent (Bio-
Rad AbD Serotec Ltd.). The NO production inhibitory potency of the synthesized compounds 27
- 31 was tested using RAW264.7 cells stimulated by LPS only and also by LPS and IFN-.
2.2 Natural Flavonoids.
Flavonoids were isolated form Sophora yunnanensis and Rhodiola sacra. S. yunnanensis is used
for the treatment of asthma, bronchitis and skin diseases. Rhodiola species grow in the Changbai
Mountains area. The roots of Rhodiola have been used as a homeostasis stabilizer (adaptogen).
All compounds were crystalized and were pure specimens. Details of the isolation procedure and
structure determination were published elsewhere.9,10
2.3 CoMFA.
CoMFA available in the SYBYL-X2.0 software package13 was used to build QSAR models. The
standard procedure of CoMFA, i.e., using grid space of 2 Å and the probe atom type being sp3-
carbon, was employed.
2.3.1 Outline of CoMFA Procedure.
6

CoMFA requires molecules to be aligned properly in space.14 Our approach to finding such
alignment was as follows. First, we built several conformational models for each of the 13
flavonoids. Then, considering possible combinations of conformers, we created 32 alignments.
We submitted the 32 alignments to CoMFA and selected an alignment that gave the best partial
least square evaluation, i.e., a high cross-validated q2 value and regression coefficient r2 with a
lower number of components.
2.3.2 Generation of Conformers.
The dihedral angle along the single bond that connects the B and the C rings (defined by
atoms O1, C2, C1’ and C2’) is the most robust determinant of the topology of flavonoids (Fig.
3). Conformational search by using the grid search method available in the SYBYL-X was
conducted, and representative energetically stable conformers were selected. The values of
Figure 3. Definition of dihedral angles and generation of conformers. Dihedral angles
that determine the conformer of flavonoids are shown. Note that these dihedral angles
are not independent. For example, a conformer having (3’ = 180°, 4’ = 180°, 5’ =
180°) is allowed for robitin 13, but a conformer having (3’ = 180°, 4’ = 0°, 5’ =
180°) would not be allowed because of unfavorable van der Waals’ collision of two
hydrogen atoms of the 3’- and 4’-hydroxy groups.
7

flavanones and flavanonols (chromane scaffold) are focused around -40°. For flavones and
flavanonols (chromene scaffold),  values of stable conformers were found around -40° and 40°.
Thus, the initial dihedral angle was set to -40° for flavonoids bearing a chromane scaffold. For
those having a chromene skeleton, two conformation models (= -40° and  = 40°) were built.
The energy barrier for rotation of the C2-C1’ bond is small (~5 Kcal/mol). We surveyed the
distribution of the  torsion angle values in protein-flavonoids complex structures stored in the
Protein Data Bank (22 entries, 8 proteins, 11 flavonoids). As expected from the small energy
barrier, the  values distribute widely. However, in a same protein, the  values of the ligand
flavonoids focused to certain values that are specific to the host proteins. We carried out our
analyses using the local minima conformations, because target proteins, if exist, on which
flavonoids work to suppress NO production are unknown.
Dihedral angles of substituents on the B ring were defined as shown in Fig. 3. The dihedral
angle of a hydroxy group was set to 0° or 180°. A crystallographic substructure search for 2,6-
methoxyphenol moiety clearly demonstrated that the methoxy carbon atoms preferably exist in
the same plane of the phenyl ring, orienting two methoxy carbon atoms apart from the oxygen
atom of the central hydroxy group.15-20 Based on this fact, for the flavonoids carrying the 4’-
hydroxy-3’,5’-dimethoxy substitution on the B ring (1, 4 and 8), 3’ and 5’ were fixed at 0° and
180°, respectively. In contrast, for the flavonoids that carry the 3’,5’-dihydroxy-4’methoxy
phenyl ring (3 and 7), 4’ was inevitably set to -90° or 90° to avoid unfavorable van der Waals’
contact with the 3’- and 5’-hydroxy groups.
The dihedral angles of the 3-hydroxy group and 5-hydroxy group in the C and A rings were
fixed to form intramolecular hydrogen bonds with a carbonyl oxygen atom at the 4 position (C2-
8

C3-O3-H: 180° and C4a-C5-O5-H: 0°). The dihedral angle of the 7-hydroxy group (C6-C7-O7-
H) was fixed at 180°.
2.3.3 Alignment of Conformers.
The B rings of the flavonoids investigated in this study can be classified into three types:
3',4',5'-trihydroxyphenyl type (category 1), 3’,5’-dimethoxy-4’hydroxyphneyl type (category 2)
and 4’-methoxy-3’,5’-hydroxyphenyl type (category 3) (Fig. 1 and Fig. 4). Flavonoids 6, 10 and
13 belong to category 1. Flavanone 12 belongs to category 1 because the B ring of flavanone 12
can be recognized as a 4’-deshydroxy derivative of the category 1 framework. Similarly, the B
ring of flavonoids 2, 5, 9 and 11 can be recognized as category 1, which lacks 3',5'-dihydoroxy
groups. Trypsin 1, laurentinol 4 and sophorayunnanol 8 carry the category 2 B rings whereas
flavonol 3 and sepinol 7 have B rings of category 3.
The B ring of category 1 was chosen as the standard reference frame on which to superimpose
B rings of the compounds. For the category 1 B ring, four conformers were premised. These four
conformers were represented by an array of torsion angles of the three hydroxy groups (3’, 4’,
5’): i.e., (0°, 0°, 0°), (0°, 0°, 180°), (0°, 180°, 180°) and (180°, 180°, 180°), as shown in Fig. 4.
Let us superimpose (or align) the B ring of laurentinol 4 (category 2) over the standard B ring
having a (3’,4’, 5’) = (0°, 0°, 0°) conformation. The 4’-hydroxy group of 4 was set to the same
conformation of the 4’-hydroxy group of the reference (4’ = 0°). As mentioned above, the
allowed torsion value for the 3’-methoxy group (3’) of 4 is 0°. (3’ = 180° is not possible because
of steric collision between the methyl moiety of the 3’-methoxy group and the oxygen atom of
the 4’-hyrdroxy group.) Similarly, the only available torsion angle of the 5’-methoxy (5’) of 4 is
9

180°. Thus, the B-ring conformation of 4 in the (0°, 0°, 180°) conformation was to be
superimposed on the standard B ring in the (0°, 0°, 0°) conformation.
Figure 4. Combinations of B-ring conformers. Numbers indicated near the lines
that link B-ring conformers represent the conformer group identification
numbers.
The conformation of the B ring of category 3 that corresponds to the standard category 1 B
ring in the (0°, 0°, 0°) conformation was determined as follows. The torsion angles of the 3’- and
5’- hydroxyl groups were set as those of the standard ring, i.e., 3’ = 5’ = 0°. The methyl moiety
of the 4’-methoxy group cannot be in the plane of the phenyl ring because of steric collision with
10

the 3’- and 5’- oxygen atoms. Thus, the torsion angle for the 4’-methoxy group of the category 3
B ring was either -90° or 90°. Two possible conformers of the category 3 B ring, (0°, -90°, 0°)
and (0°, 90°, 0°), should be superimposed on the standard B ring in the (0°, 0°, 0°) conformation.
Taken together, conformers that were to be superimposed on the category 1 B ring of (0°, 0°,
0°) were (0°, 0°, 180°) for the category 2 ring and either (0°, -90°, 0°) or (0°, 90°, 0°) for the
category 3 ring, leading to two possible alignments: “category 1 (0°, 0°, 0°) – category 2 (0°, 0°,
180°) – category 3 (0°, -90°, 0°)” or “category 1 (0°, 0°, 0°) – category 2 (0°, 0°, 180°) –
category 3 (0°, 90°, 0°)”. Such combinations are summarized in Fig. 4. Eight combinations
existed.
Further, there are two possible ways to align whole structures of the 13 flavonoids. One is to
align the molecules based on the A ring, and the other is to align the molecules based on the A
and B rings. The former method uses six carbon atoms of the A ring and the carbonyl carbon and
oxygen atoms at position 4 of the flavonoids to fit molecules. The latter uses three carbon atoms
(C4’ of the B ring and C5 and C7 of the A ring) to fit molecules. The former emphasizes the
diversity of the B ring topology in space whereas the later maximizes the overall volume overlap
of the molecules.
The two combinations of the A/C ring conformations, i.e. (chroman = -40°, chromen = -40°) and
(chroman = -40°, chromen = 40°), eight combinations of the B-ring superimposition and the two fitting
methods produced a total of 32 alignments. The torsion angles to build the initial conformers are
given in supplementary data. For these 32 alignments, every conformer generated was first
optimized by using the MMFF94 force field21-25 and then by MOPAC.26 Atomic charges of each
conformer were calculated by MOPAC. Thus, the atomic charges used in this CoMFA study
were conformation dependent.
11

The torsional angles after optimization changed from the initial values, but the deviation was
small. The deviation of was under 15° for flavonols and flavones and under 30° for flavanones
and flavanonols. Deviation values of 3’, 4’and 5’ were under 10° for the hydroxy groups and
under 30° for the methoxy groups. Thus, the molecular models were in the local minima
conformations near the initial conformations.
2.4 Synthesis of 2',4',6'-Trihydroxy-2-chloroacetophenone 26.
Anhydrous ZnCl2 (1.1 g, 8.0 mmol) was added to a mixture of anhydrous phloroglucinol (5.0 g,
40 mmol) and chloroacetonitrile (3.8 mL, 60 mmol) in ether (50 mL) with stirring at 0 - 5 °C
(ice-water bath). When the solution was cooled to 0 - 5 °C, anhydrous HCl gas was bubbled
through the reaction solution until an amount of thick yellow solid precipitated. The reaction
mixture was left for 6 h. The precipitated imine was collected by filtration and washed with ether
three times. The briefly dried imine was dissolved in 25 mL of distilled water. Twenty-five
milliliters of aqueous 2M HCl was added to the solution and stirred for 8 h at room temperature.
The precipitated solid was collected by filtration, washed three times with distilled water, and
then dried under a vacuum to give 26 as a pale white powder (6.6 g, 82% yield).
13
1H NMR (600 MHz, methanol-d4) δ: 4.91 (2H, s, H-2), 5.84 (2H, s, H-3', H-5'). C NMR (150
MHz, methanol-d4) δ: 51.9 (C-2), 96.0 (C-3', C- 5'), 104.1 (C-1'), 165.8 (C-2', C-6'), 167.3 (C-4'),
196.7 (C-1). HR-EI-MS: m/z 202.0033 [M]+ (Calcd 202.0033 for C8H7O4Cl).
2.5 Synthesis of flavones
Acacetin (27).
12

KOH (250 mg, 4.47 mmol) ethanol solution (10 mL) was added to a solution of 26 (300 mg, 1.49
mmol) in EtOH (10 mL). Then, 4'-methoxy benzaldehyde (202 mg, 1.49 mmol) was added to the
reaction mixture solution and stirred at room temperature for 3 h. Five milliliters of aqueous HCl
(10 M) was added and stirred at 55 °C for 1 h. EtOH was removed under a vacuum. Distilled
water was added to the residue, and the mixture was extracted with EtOAc three times. The
organic layer was combined and dried over Na2SO4. The filtered EtOAc layer was evaporated to
give a red-yellow oil that was purified by silica gel column chromatography eluting with MeOH
and CHCl3 to give 27 as a yellow powder (75 mg, 17.8% yield).
1H NMR (600 MHz, acetone-d6) δ: 3.87 (3H, s, OCH3-4'), 6.14 (1H, s, H-6), 6.35 (1H, s, H-8),
13
6.62 (1H, s, H-3), 7.04 (1H, d, J = 8.4 Hz, H-3', H-5'), 7.90 (1H, d, J = 8.4 Hz, H-2', H-6'). C
NMR (150 MHz, acetone-d6) δ: 55.9 (OCH3-4'), 92.1 (C-8), 98.6 (C-6), 104.4 (C-10), 110.5 (C-
3), 115.4 (C-3', C-5'), 126.1 (C-1'), 133.7 (C-2', C-6'), 147.7 (C-9), 158.9 (C-5), 161.8 (C-4'),
168.4 (C-7), 168.4 (C-2), 181.7 (C-4). HR-EI-MS: m/z 284.0683 [M]+ (Calcd 284.0685 for
C16H12O5).
Diosmetin (28). KOH (250 mg, 4.47 mmol) ethanol solution (10 mL) was added to a solution of
26 (300 mg, 1.49 mmol) in EtOH (10 mL). Then, 3'-hydroxy-4'-methoxy benzaldehyde (226 mg,
1.49 mmol) was added to the reaction mixture and stirred at room temperature for 3 h. Five
milliliters of aqueous HCl (10 M) was added and stirred at 55°C for 1 h. EtOH was removed
under a vacuum. Distilled water was added to the residue, and the mixture was extracted with
EtOAc three times. The organic layer was combined and dried over Na2SO4. The filtered EtOAc
layer was evaporated to give a red-yellow oil that was purified by silica gel column
chromatography eluting with MeOH and CHCl3 to give 28 as a yellow powder (65 mg, 14.6%
yield).
13

1H NMR (600 MHz, acetone-d6) δ: 3.91 (3H, s, OCH3-4'), 6.15 (1H, d, J = 1.8 Hz, H-6), 6.29
(1H, d, J = 1.8 Hz, H-8), 6.52 (1H, s, H-3), 7.04 (1H, d, J = 8.4 Hz, H-5'), 7.34 (1H, dd, J = 8.4,
1.8 Hz, H-6'), 7.55 (1H, d, J = 1.8 Hz, H-2'). 13C NMR (150 MHz, acetone-d6) δ: 56.3 (OCH3-4'),
91.7 (C-8), 98.8 (C-6), 104.1 (C-10), 110.1 (C-3), 112.6 (C-5'), 118.0 (C-2'), 124.7 (C-6'), 126.8
(C-1'), 147.8 (C-9), 147.8 (C-3'), 150.0 (C-4'), 159.3 (C-5), 168.7 (C-7), 168.7 (C-2), 181.0 (C-
4). HR-EI-MS: m/z 300.0634 [M]+ (Calcd 300.0634 for C16H12O6).
4'-O-Methyltricetin (29). KOH (200 mg, 3.57 mmol) ethanol solution (10 mL) was added to a
solution of 26 (240 mg, 1.19 mmol) in EtOH (10 mL). Then, 3',5'-dihydroxy-4'-methoxy
benzaldehyde (200 mg, 1.19 mmol) was added to the reaction mixture and stirred at room
temperature for 3 h. Five milliliters of aqueous HCl (10 M) was added and stirred at 55 °C for 1
h. EtOH was removed under a vacuum. Distilled water was added to the residue, and the mixture
was extracted with EtOAc three times. The organic layer was combined and dried over Na2SO4.
The filtered EtOAc layer was evaporated to give a red-yellow oil that was purified by silica gel
column chromatography eluting with MeOH and CHCl3 to give 29 as a yellow powder (65 mg,
17.3% yield).
1H NMR (600 MHz, acetone-d6) δ: 3.87 (3H, s, OCH3-4'), 6.15 (1H, d, J = 1.8 Hz, H-6), 6.34
13
(1H, d, J = 1.8 Hz, H-8), 6.46 (1H, s, H-3), 7.05 (1H, s, H-2', H-6'). C NMR (150 MHz,
acetone-d6) δ: 60.8 (OCH3-4'), 92.1 (C-8), 98.7 (C-6), 104.2 (C-10), 110.7 (C-3), 111.6 (C-2', C-
6'), 129.0 (C-1'), 138.0 (C-4'), 148.2 (C-9), 151.6 (C-3', C-5'), 158.9 (C-5), 168.5 (C-7), 168.5
(C-2), 181.7 (C-4). HR-EI-MS: m/z 316.0584 [M]+ (Calcd 316.0583 for C16H12O7).
3'-O-Methyldiosmetin (30). KOH (166 mg, 2.97 mmol) ethanol solution (10 mL) was added to
a solution of 26 (200 mg, 0.99 mmol) in EtOH (10 mL). Then, 3',4'-dimethoxy benzaldehyde
14

(165 mg, 0.99 mmol) was added to the reaction mixture and stirred at room temperature for 3 h.
Five milliliters of aqueous HCl (10 M) was added and stirred at 55 °C for 1 h. EtOH was
removed under a vacuum. Distilled water was added to the residue, and the mixture was
extracted with EtOAc three times. The organic layer was combined and dried over Na2SO4. The
filtered EtOAc layer was evaporated to give a red-yellow oil that was purified by silica gel
column chromatography eluting with MeOH and CHCl3 to give 30 as a yellow powder (30 mg,
9.6% yield).
1H NMR (600 MHz, acetone-d6) δ: 3.88 (3H, s, OCH3-4'), 3.90 (3H, s, OCH3-3'), 6.15 (1H, d, J =
1.8 Hz, H-6), 6.35 (1H, d, J = 1.8 Hz, H-8), 6.61 (1H, s, H-3), 7.05 (1H, d, J = 8.4 Hz, H-5'),
7.53 (1H, dd, J = 8.4, 1.8 Hz, H-6'), 7.57 (1H, d, J = 1.8 Hz, H-2'). 13C NMR (150 MHz, acetone-
d6) δ: 56.2 (OCH3-3'), 56.2 (OCH3-4'), 92.2 (C-8), 98.6 (C-6), 104.3 (C-10), 111.0 (C-3), 112.7
(C-5'), 115.1 (C-2'), 126.0 (C-6'), 126.3 (C-1'), 147.7 (C-9), 150.4 (C-3'), 151.9 (C-4'), 158.8 (C-
5), 168.4 (C-7), 168.4 (C-2), 181.7 (C-4). HR-EI-MS: m/z 314.0791 [M]+ (Calcd 314.0790 for
C17H14O6).
Apometzgerin (31). KOH (250 mg, 4.47 mmol) ethanol solution (10 mL) was added to a
solution of 26 (300 mg, 1.49 mmol) in EtOH (10 mL). Then, 3',4'-dimethoxy-5'-hydroxy
benzaldehyde (270 mg, 1.49 mmol) was added to the reaction mixture and stirred at room
temperature for 3 h. Five milliliters of aqueous HCl (10 M) was added and stirred at 55 °C for 1
h. EtOH was removed under a vacuum. Distilled water was added to the residue, and the mixture
was extracted with EtOAc three times. The organic layer was combined and dried over Na2SO4.
The filtered EtOAc layer was evaporated to give a red-yellow oil that was purified by silica gel
column chromatography eluting with MeOH and CHCl3 to give 31 as a yellow powder (70 mg,
14.3% yield).
15

1H NMR (600 MHz, acetone-d6) δ: 3.83 (3H, s, OCH3-4'), 3.92 (3H, s, OCH3-3'), 6.15 (1H, d, J =
1.2 Hz, H-6), 6.36 (1H, d, J = 1.2 Hz, H-8), 6.54 (1H, s, H-3), 7.10 (1H, d, J = 1.8 Hz, H-6'),
13
7.22 (1H, d, J = 1.8 Hz, H-2'). C NMR (150 MHz, acetone-d6) δ: 56.4 (OCH3-3'), 60.9 (OCH3-
4'), 92.2 (C-8), 98.7 (C-6), 104.2 (C-10), 108.2 (C-6'), 110.7 (C-3), 112.5 (C-2'), 129.1 (C-1'),
138.8 (C-4'), 148.3 (C-9), 151.5 (C-5'), 154.2 (C-3'), 158.9 (C-5), 168.5 (C-7), 168.5 (C-2), 181.6
(C-4). HR-EI-MS: m/z 330.0739 [M]+ (Calcd 330.0740 for C17H14O7).
3. Results and discussion
3.1 Flavonoids and Their Inhibitory Potential on NO Production.
IC50 values for the NO production of flavonoids isolated from S. yunnanensis and R. sacra are
indicated in Fig. 1. No compound showed cell toxicity at 100 µM. Details of their isolation,
structure determination and evaluation of activity were published previously.9,10
Flavonoids carrying a chromene scaffold tended to exhibit higher potency than those having a
chromane scaffold. Flavanonols (7 - 10) showed poor inhibitory potency, which seemed to be
related to the structure of the A/C-ring system (i.e., chromane or chromene ring system). For
example, the inhibitory potency of naringenin 11, dihydrokaempferol 9, kaempferol 5 and
apigenin 2, which commonly have a 4’-hydroxyphenyl ring as the B ring, varies widely. The
existence of a hydroxy group at the 5-position tends to enhance the inhibitory potency.
3.2 CoMFA model T2c.
16

We ran 32 CoMFA calculations, applying the standard default conditions implemented in the
SYBYL-X software package. Cross-validation was carried out using the leave-one-out
procedure. The number of components was varied from 1 to 8. The inhibitory potency of the
flavonoids (IC50 value in µM) was converted into –logIC50 values, which were the dependent
variables. IC50 values higher than 100 µM were regarded as 300 µM.
skeletone conformation
noncoplanar
coplanar
= -40° for flavanones and
flavanonols (Chromane)
= 40° for flavones and
= -40°
flavonols (Chromene)
B-ring
confomer
group #
Cross-
Cross-
Atoms used
for alignment
Number of
components
(PLS)
Number of
components
(PLS)
validated Coreraltion
validated Coreraltion
q2
r2
q2
r2
1
2
3
4
5
6
7
8
4
4
4
6
8
6
8
1
0.483
0.705a)
0.429
0.648
0.313
0.519
0.396
0.293
0.954
0.977
0.967
0.992
0.995
0.983
0.997
0.638
5
4
1
5
2
1
8
1
0.363
0.555
0.355
0.594
0.442
0.334
0.495
0.378
0.991
0.974
0.592
0.988
0.831
0.584
0.998
0.609
T
C5, C7 and
C3'
(Three atoms)
1
2
3
4
5
6
7
8
4
4
6
8
4
7
8
4
0.427
0.504
0.274
0.522
0.148
0.336
0.119
0.223
0.964
0.956
0.993
1.000
0.929
0.995
1.000
0.944
1
2
1
5
2
2
6
1
0.351
0.446
0.341
0.567
0.383
0.341
0.356
0.337
0.600
0.850
0.584
0.992
0.818
0.810
0.994
0.588
E
A-ring (C4a,
C5, C6, C7,
C8, C8a) , C4
and O4
(Eight atoms)
Table 1.
Results of CoMFA runs.
a)Model T2c (Three atom alignment, group 2 type B ring combination, co-planar skeleton conforma
17

From each CoMFA run, the model with the largest cross-validated q2 (predictive coefficient)
value was selected. Table 1 summarizes the characteristics of the 32 CoMFA models. Among
them, model T2c had a q2 value of 0.705 with the number of components equal to 4. The name
-3
-2.5
-2
-1.5
-1
-0.5
-0.5
kp17
kp18
-1
kp21
kp23
2
kp20
kp16
kp15
kp22
5
kp19
-1.5
-2
kp14
kp25
Training Set
kp24
Predicted Set
-2.5
-3
-Log(observed IC50 [µM])
Figure 5. Regression plot of the predicted inhibitory potential (Y axis) versus the
experimentally determined inhibitory potential (X axis). Closed circles indicate the
experimentally determined inhibitory potential versus predicted inhibitory potential
of the 13 compounds by CoMFA model T2c. Open circles indicate the inhibitory
potential of flavonoids reported by Matsuda et al.12 (Table 2) versus the predicted
potential calculated by CoMFA model T2c (See section 3.3). Apigenin (2, kp16)
and kaempeferol (5, kp19) are common compounds between our 13 compounds
and those of Matsuda et al. Their plots are expressed using closed gray circles (our
data set) and open gray circles (their data set).
18

T2c means that the CoMFA model was derived from conformers having a B-ring conformer
group of 2 with value of -40° (coplanar) aligned by the three atom fitting (C5, C7 and C3’).
The correlation coefficient r2 of model T2c was 0.977 (Fig. 5, closed circles). The contribution
ratios of electrostatic field and steric field were 0.56 and 0.44, respectively.
We ran CoMFA that included ClogP values as an additional variable, but no improvement
was observed. Within our data set, the contribution of hydrophobicity to the inhibitory potential
was not relevant.
Fig. 6-(a) depicts the CoMFA field mapping. Steric repulsive regions exist above the 2-
position and below the 3-position of the C ring (regions 2 and 3). These regions indicate that the
chromane scaffold is less favored for higher potency because the chromane scaffold carries
hydrogen atoms at the 2- and 3-positions whereas in the chromene scaffold, the 2- and 3-
positions are unsaturated, and no atom exists to interfere with the steric regions. There is a
sterically favored region near the 5-position of the A ring (region 1). Flavonoids that carry a
substitution at the 5-postion is advantageous for the inhibitory potency. Electrostatic mappings
exist around the B ring (regions A and B). The region A is an electrostatically positive mapping
and the approach of negatively charged surface of a ligand molecular is favorable for the
inhibitory potency. Thus, within the scope of the 13 compounds, the inhibitory potency of
flavonoids on NO production is basically explained by the electrostatic property of the B ring,
the choice of the scaffold and the presence of a hydroxy group at the 5-position.
19

Figure 6. CoMFA field mapping of model T2c (stereo view; wall-eyes). The contours
of the steric map are shown in yellow (regions 2, 3, 4, and 5) and green (regions 1 and
6), and those of the electrostatic map are shown in red (regions A and C) and blue
(regions B and D). The approach of the molecular volume of a compound to the
yellow region reduces the inhibitory potency whereas the existence of the molecular
volume near the green region is favorable. The approach of a positive charge to the
blue region is attractive and favorable, but that of a negative charge to the blue region
is unfavorable. Similarly, the red region is favorable to a negative charge and
unfavorable to a positive charge. (a) All 13 compounds in CoMFA mapping, (b)
flavanonol 12 (high potency) and (c) flavanonol 13 (low potency) are shown.
20

The 13 flavanoids are shown with the CoMFA field in Fig. 6-(a). Note that the position of the
B ring differs depending upon the scaffold of the A/C ring. The 3’-position of the B ring of
chromenes is distant from the region A while the 3’-position of chromanes is closer to the region
A. The 5’-position of chromes is closer to the region B than that of chromanes. This means that
the nature of electrostatic interaction of the B rings with region A and B depends on the A/C ring
systems.
Fig. 6-(b) shows flavanone 12 (IC50 = 12 µM) in the CoMFA mapping. The electrostatic
potential was color-coded on the molecular van der Waals’ surface. The electrostatic
complementarity of the B ring of 12 with the region A and B are in good compatibility. Also
electrostatic complementarity with region C is favorable. There also exists a favorable steric
interaction between the 5-hydroxy group and the region 1. Robitin 13 is an inactive flavanone
which differ from 12 in the absence of the 5-hydroxy group and in the existence of an additional
hydroxy group at the 4’-position. Fig. 6-(c) shows the electrostatic potential on the molecular
surface of 13 and the CoMFA mapping. The presence of the 4’-hydroxy group changes
electrostatic property of the B ring into poorer complementarity with the regions A and B.
Furthermore this compound lacks the 5-hydroxy substitution and interaction with the region 1 is
smaller. CoMFA model suggests that the activity of flavonoids in the suppression of NO
production is related to the overall topology of the molecules as well as the electrostatic property
of the B ring. Indeed, flavanonol skeletons have been regarded as inactive skeletons for the
inhibition of NO production, but we succeeded in synthesizing a flavanonol that exhibited an IC50
of 60 µM.27
It is unexpected that the CoMFA model T2c suggests the 4’-hydroxy group of chromanes is
unfavorable for the inhibitory activity for NO production. The 4’-hydroxy group is the conserved
21

substitution among flavonoids because the most frequent starting unit for the biosynthesis of
flavonoids is p-coumaric acid. Substitution of a hydroxy group or methoxy group at the 3’-
position of the B ring of chromane is favorable for the inhibitory activity.
Molecular surface representations of the 13 flavonoids with the CoMFA mapping are
available as supplementary data.
3.3 Validation of CoMFA model T2c: Predictive Capability.
We applied the T2c model to the structure-activity data sets complied by Matsuda et al.12
Because CoMFA model T2c was derived from the 13 flavonoids with limited structural
variation, it would be valid within the structural variety implemented by the 13 compounds.
Thus, we selected the 12 compounds (kp14 - kp25, Table 2) that match the Markush structure
shown in Fig. 7 from the Matsuda et al. data set.12
Figure 7. Markush structure used to select compounds from the original article
of Matsuda et. al.12 The selected compounds were predicted by CoMFA model
T2c for their inhibitory potency. The correlation between predicted potency and
experimental potency is plotted in Figure 5 (open circles).
22

R3
H
R5
H
R3'
H
R4'
OH
R5'
H
Compound
4',7-dihydroxyflavone
3',4',7-trihydroxyflavone
Apigenin
IC50 [µM]
14
kp14
kp15
kp16
kp17
kp18
kp19
kp20
kp21
kp22
kp23
26
H
H
OH
H
OH
H
7.7
20
H
OH
OH
OH
OH
OH
OH
OH
OH
OH
H
Luteolin
H
OH
OH
H
OH
H
Diosmetin
8.9
29
H
OCH3
OH
H
Kaempferol
OH
OH
OH
OH
OH
H
Quercetin
36
OH
OH
OH
OH
OH
H
Tamarixetin
25
OCH3
OH
H
Myricetin
99
OH
OH
Mearncetin
84
OCH3
Name
IC50 [µM]
R5
H
R3'
H
R4'
OH
OH
Liquiritigenin
Eriodictyol
85
kp24
kp25
> 100
OH
OH
Structure-activity data from Matsuda et.al.12
Table 2.
23

It should be noted that the assay methods of Matsuda et al. were not identical to ours. They
determined the inhibitory potency of flavonoids on NO production using peritoneal exudate cells
collected from the peritoneal cavity of ddY mice, and they stimulated the cells only by LPS.12 We
tested the 13 compounds using RAW264.7 cells, a murine macrophage cell line, and we
stimulated the cells with LPS (100 ng/mL) and IFN-(0.33 ng/mL). The method used to quantify
NO production was the same (Greiss method10,12).
Two compounds were commonly included in the Matsuda et al. data set and in ours. The IC50
value of apigenin 2 was 14 µM in our evaluation and was reported to be 8 µM in theirs (kp16),
and that of kaempeferol 5 was 15 µM in our evaluation and 29 µM in theirs (kp19). Given that
these divergences are permissible, we applied the T2c model to predict the potency of the
Matsuda et al. compounds. For the compounds carrying asymmetrically substituted B rings, two
possible B ring orientations were evaluated, and the one with the lower IC50 value (more potent
activity) was chosen. In Fig. 5, the experimental IC50 values and the predicted values of the
Matsuda et al. compounds were plotted, and a reasonable predictive ability of the CoMFA model
T2c was shown. Prediction of the inhibitory activity of compounds having high lipophilicity
(kp14 and kp24) tended to be lower than the observed activity and that of the compounds with
lower lipophilicity (kp17, kp22, kp23 and kp25) was overestimated. This fact suggests that cell
permeability of the compounds or their perturbation into the cell membrane might be relevant.
Indeed, no glycosylated flavonoid is known to inhibit NO production. We have observed that the
inhibitory activity of NO production in RAW264.7 cells of flavanonols is stereospecific, and
flavanonols having (2R, 3R) stereochemistry are active but the (2S, 3S) enantiomers are not.27
Taken together, flavonoids would interact with certain target molecule(s) that stereochemically
recognizes flavonoid molecules. Many previous studies reported that the NO inhibitory function
24

of flavonoids is associated with the suppression of inducible nitric oxide synthase (iNOS)
biosynthesis.7,12
3.4 Validation of CoMFA Model T2c: Synthesis of Active Compounds.
We constructed a virtual compound library, which was composed of chromen-4-ones that
carry a phenyl having 3’-, 4’- and 5’-substitutions (Fig. 8). The inhibitory potency of the virtual
compounds was predicted by the model T2c. Of the predicted activity of the virtual compounds
having asymmetrical substitutions in the B-ring, the one indicating a lower IC50 value was
chosen. In general, compounds having a 3,5-dihydroxyflavone skeleton were predicted to
indicate higher inhibitory potential, and the compounds having a 3’-methoxy substitution in the
B ring were predicted to be preferable for their high potency. Detail of the virtual library and the
predicted IC50 values by the model T2c is available as supplementary data.
Figure 8. Structures of virtual compounds. Virtual compounds defined by this
Markush structures were modeled and their inhibitory potency was predicted by
CoMFA model T2c. Five compounds were selected and synthesized. Their
inhibitory potency is shown in Table 3.
25

Scheme 1. Synthesis of flavones. Reagents and conditions: (i) (1) ZnCl2,
Et2O (2) HCl gas, 0-5 °C; (ii) 1 M HCl aqueous, rt; (iii) (1) KOH, EtOH, rt;
(2) HCl aqueous, 55 °C.
Five flavones, 27 - 31, that were predicted to be potent by the model T2c were synthesized as
shown in Scheme 1. An intermediate, 2',4',6'-trihydroxy-2-chloroacetophenone 26, was prepared
by the hydrolysis of 2-(2-chloro-1-iminoethyl)benzene-1,3,5-triol, which was synthesized from
phloroglucinol and chloroacetonitrile aided by ZnCl2. Benzaldehyde derivatives were reacted
with 26 to yield flavones 27 - 31.
The inhibitory potency for NO production of compounds 27 - 31 is summarized in Table 3.
We carried out two types of assay. One is our conventional assay in which RAW264.7 cells are
stimulated by LPS and IFN-, and the other is similar to the Matsuda et al. method whereby the
cells are stimulated only by LPS. Compound 28 is diosmetin, and its IC50 was 16.7 µM
(LPS)/32.2 µM (LPS/IFN-). This compound was included in the Matsuda et al. data set as kp18
(Table 2). Its IC50 was reported as 8.9 µM (LPS) and was the third most potent flavonoid in their
data set. Compounds 30 (3'-O-methyldiosmetin) and 31 (apometzgerin) showed higher potency
26

than 28 in our assay: IC50 5.0 µM (LPS)/17.0 µM (LPS/IFN-) and 6.9 µM (LPS)/18.2 µM
(LPS/IFN-), respectively. It seemed that the existence of the 3’-methoxy group on the B ring
was important for the inhibitory activity of 5,7-dihydroxy-flavanones. Actually, tricine 1 has an
IC50 value of 8 µM (LSP/IFN-) as shown in Fig. 1. IC50 values of compounds 27 and 29 were
27.7 µM (LPS)/53.7 µM (LPS/IFN-) and 37.0 µM (LPS)/59.3 µM (LPS/IFN-), respectively.
Compounds 27 - 31 did not show cell toxicity at 100 µM.
Unfortunately, the diversity of the A/C ring was very small in our data set. It is interesting that
Matsuda et al. reported that 3’,4’,5,7-tetramethoxy-flavone (tetra-O-methyl-luteolin) showed the
highest inhibitory potential among the 73 compounds they investigated.12 Considering that
3’,4’,5,7-tetramethoxy-flavone is very hydrophobic, hydrophobicity might be important for its
27

activity. We are currently undertaking a synthetic study of flavonoids to further expand structural
diversity and hydrophobicity with the aim to more fully establish the structure-activity
relationship of flavonoids on their inhibitory potency against NO production.
4. Conclusion
We carried out a CoMFA study on the inhibitory activity of flavonoids against NO
production. We constructed the possible alignments of 13 flavonoids systematically. For this
process, the use of symmetrically substituted analogues was the key to reducing the possible
numbers of alignments to a feasible size. The CoMFA model T2c reasonably explained the
structure-activity relationship. The preference of the chromene scaffold over the chromane
scaffold was expressed as sterically unfavorable regions above the 2-position and below the 3-
position. The preference for the existence of a hydroxy group at the 5-position was expressed as
a sterically favorable region between the 4- and 5- positions. The CoMFA fields around the B
ring are mostly electrostatic fields, suggesting that electrostatic features of the B-ring
substitutions are important. Because the topology of the B ring alters depending on the scaffold,
the effects of the B-ring substitutions on inhibitory potency would differ scaffold by scaffold,
and this rendered CoMFA a robust methodology in this study. The CoMFA model T2c exhibited
a predictable capability, and this model rationally helped us to obtain very potent compounds 30
and 31. To our knowledge, their inhibitory potential against NO production has not been reported
so far, and this study appears to be the first to report the inhibitory potency of these compounds.
Abbreviations
28

CoMFA, comparative molecular field analysis; IFN-, interferon-; LPS, lipopolysaccharide;
NO, nitric oxide
Acknowledgement
Part of this study was supported by the Academic Frontier Project of the Ministry of Education,
Culture, Sports, Science and Technology. We thank the Otsuka Toshimi Scholarship Foundation
and Watanuki International Scholarship Foundation for supporting the study and research of
author W.-J.J. We thank Dr. Masateru Ohta (Chugai Pharmaceutical Co., Ltd.) for discussions
and insightful suggestions.
29